Isilon OneFS 7.1 Web Administration Guide - Dell EMC US · PDF fileAuditing settings ... Modify the password of an NDMP user account ... OneFS 7.1 Web Administration Guide 9

IsilonOneFSVersion 7.1

Web Administration Guide

Copyright © 2001-2014 EMC Corporation. All rights reserved. Published in USA.

Published March, 2014

EMC believes the information in this publication is accurate as of its publication date. The information is subject to changewithout notice.

The information in this publication is provided as is. EMC Corporation makes no representations or warranties of any kind withrespect to the information in this publication, and specifically disclaims implied warranties of merchantability or fitness for aparticular purpose. Use, copying, and distribution of any EMC software described in this publication requires an applicablesoftware license.

EMC², EMC, and the EMC logo are registered trademarks or trademarks of EMC Corporation in the United States and othercountries. All other trademarks used herein are the property of their respective owners.

For the most up-to-date regulatory document for your product line, go to EMC Online Support (https://support.emc.com). Fordocumentation on EMC Data Domain products, go to the EMC Data Domain Support Portal (https://my.datadomain.com).

EMC CorporationHopkinton, Massachusetts 01748-91031-508-435-1000 In North America 1-866-464-7381www.EMC.com

2 OneFS 7.1 Web Administration Guide

Introduction to this guide 17

About this guide............................................................................................18Isilon scale-out NAS overview........................................................................18Where to go for support.................................................................................18

Isilon scale-out NAS 19

OneFS storage architecture........................................................................... 20Isilon node components................................................................................20Internal and external networks...................................................................... 21Isilon cluster................................................................................................. 21

Cluster administration......................................................................21Quorum............................................................................................21Splitting and merging.......................................................................22Storage pools...................................................................................23IP address pools.............................................................................. 23

The OneFS operating system......................................................................... 23Data-access protocols......................................................................24Identity management and access control......................................... 24

Structure of the file system............................................................................25Data layout...................................................................................... 25Writing files......................................................................................26Reading files.................................................................................... 26Metadata layout...............................................................................26Locks and concurrency.....................................................................27Striping............................................................................................27

Data protection overview...............................................................................27N+M data protection........................................................................ 28Data mirroring..................................................................................29The file system journal..................................................................... 29Virtual hot spare.............................................................................. 29Balancing protection with storage space.......................................... 29

VMware integration....................................................................................... 29The iSCSI option............................................................................................30Software modules......................................................................................... 30

General cluster administration 33

General cluster administration overview........................................................34User interfaces.............................................................................................. 34Connecting to the cluster...............................................................................35

Log in to the web administration interface........................................35Open an SSH connection to a cluster................................................35

Licensing.......................................................................................................36License status..................................................................................36License configuration.......................................................................39Activate a license............................................................................. 39View license information.................................................................. 40Unconfigure a license.......................................................................40

Certificates....................................................................................................40

Chapter 1

Chapter 2

Chapter 3

CONTENTS

OneFS 7.1 Web Administration Guide 3

Replace or renew the SSL certificate.................................................41Verify an SSL certificate update........................................................42Self-signed SSL certificate data example..........................................42

General cluster settings.................................................................................42Set the cluster name........................................................................ 43Specify contact information..............................................................43Configuring the cluster date and time...............................................43Set the cluster date and time........................................................... 44Specify an NTP time server............................................................... 44Configure SMTP email settings......................................................... 45Specify the cluster join mode........................................................... 45Cluster join modes........................................................................... 45Enable or disable access time tracking.............................................46Specify the cluster character encoding............................................. 46

Cluster statistics........................................................................................... 47Performance monitoring................................................................................ 47Cluster monitoring.........................................................................................47

Monitor the cluster...........................................................................48View node status............................................................................. 49Events and notifications...................................................................49

Monitoring cluster hardware..........................................................................57View node hardware status.............................................................. 57Chassis and drive states.................................................................. 57SNMP monitoring............................................................................. 59

Cluster maintenance..................................................................................... 62Replacing node components............................................................ 62Upgrading node components........................................................... 63Managing cluster nodes................................................................... 63

Remote support using SupportIQ...................................................................64Configuring SupportIQ......................................................................65Enable and configure SupportIQ.......................................................65Disable SupportIQ............................................................................66SupportIQ scripts............................................................................. 66

Upgrading OneFS.......................................................................................... 68

Access zones 69

Access zones overview..................................................................................70Access zone features.....................................................................................70Managing access zones................................................................................ 71

Create an access zone......................................................................71Access zone settings........................................................................72Associate an IP address pool with an access zone............................73Modify an access zone..................................................................... 73Delete an access zone......................................................................74

Authentication and access control 75

Authentication and access control overview.................................................. 76Authentication and access control features......................................76

Data access control.......................................................................................77ACLs................................................................................................ 77UNIX permissions.............................................................................78Mixed-permission environments...................................................... 78

Roles.............................................................................................................79Built-in roles.................................................................................... 79

Chapter 4

Chapter 5

CONTENTS


OneFS privileges.............................................................................. 82Command-line interface privileges................................................... 84

Authentication.............................................................................................. 88Supported authentication providers.................................................88Authentication provider features...................................................... 89LDAP................................................................................................ 89Active Directory................................................................................ 90NIS...................................................................................................90File provider..................................................................................... 90Local provider.................................................................................. 91

Managing access permissions.......................................................................91Configure access management settings........................................... 91Modify ACL policy settings................................................................92ACL policy settings options.............................................................. 92Update cluster permissions..............................................................97

Managing roles............................................................................................. 99View roles........................................................................................ 99View privileges.................................................................................99Create a custom role...................................................................... 100Modify a role..................................................................................100Delete a custom role...................................................................... 100

Managing authentication providers............................................................. 100Configure an LDAP provider............................................................101Managing LDAP providers.............................................................. 102Configure an Active Directory provider............................................105Managing Active Directory providers.............................................. 106Managing NIS providers................................................................. 108Configuring file providers............................................................... 110Managing file providers..................................................................111Create a local user......................................................................... 114Create a local group....................................................................... 115Managing local users and groups...................................................115

Identity management 121

Identity management.................................................................................. 122Identity types.............................................................................................. 122Access tokens............................................................................................. 123Access token generation............................................................................. 124

ID mapping.................................................................................... 124User mapping across systems and identities..................................125On-disk identity............................................................................. 130

Auditing 133

Auditing overview........................................................................................134Protocol audit events.................................................................................. 134Supported event types................................................................................ 134Supported audit tools................................................................................. 135Enable system configuration auditing..........................................................136Enable protocol access auditing..................................................................136Auditing settings.........................................................................................137Integrating with the EMC Common Event Enabler.........................................137

Install CEE for Windows..................................................................138Configure CEE for Windows............................................................ 139

Chapter 6

Chapter 7

CONTENTS


File sharing 141

File sharing overview...................................................................................142SMB............................................................................................................ 142

Overlapping display names for SMB shares....................................143NFS............................................................................................................. 143HTTP and HTTPS.......................................................................................... 143FTP..............................................................................................................144Mixed protocol environments......................................................................144Write caching with SmartCache................................................................... 144

Write caching for asynchronous writes........................................... 145Write caching for synchronous writes............................................. 145

Managing the SMB service.......................................................................... 146Configure SMB file sharing............................................................. 146Limit access to /ifs share for the Everyone account........................ 146Snapshots directory settings..........................................................147File and directory permission settings............................................ 147SMB performance settings............................................................. 148SMB security settings.....................................................................148

Managing SMB shares.................................................................................148Create an SMB share......................................................................149Modify SMB share permissions, performance, or security...............150Add a user or group to an SMB share..............................................151Configure overlapping share display names................................... 151Configure multi-protocol home directory access............................. 152Delete an SMB share......................................................................152

Managing the NFS service........................................................................... 152Configure NFS file sharing.............................................................. 152Create a root-squashing rule for the default NFS export.................. 153NFS service settings....................................................................... 154NFS performance settings.............................................................. 154NFS client compatibility settings.................................................... 155NFS export behavior settings..........................................................155

Managing NFS exports.................................................................................156View and configure default NFS export settings.............................. 156Create an NFS export......................................................................156Modify an NFS export..................................................................... 158Delete an NFS export......................................................................158Check NFS exports for errors.......................................................... 158

Enable and configure FTP file sharing.......................................................... 159Enable and configure HTTP.......................................................................... 159Home directories.........................................................................................160

Home directory permissions...........................................................161Home directory creation through SMB............................................ 161Home directory creation through SSH and FTP................................ 163Home directory creation in a mixed environment............................166Interactions between ACLs and mode bits...................................... 166Interactions with dot-file provisioning............................................ 166Default home directory settings in authentication providers........... 167Supported expansion variables......................................................167Domain variables in home directory provisioning........................... 168

Snapshots 171

Snapshots overview.................................................................................... 172Data protection with SnapshotIQ.................................................................172

Chapter 8

Chapter 9

CONTENTS


Snapshot disk-space usage........................................................................ 172Snapshot schedules....................................................................................173Snapshot aliases........................................................................................ 173File and directory restoration.......................................................................173Snapshot best practices..............................................................................174Best practices for creating snapshot schedules...........................................174File clones...................................................................................................175

Shadow store considerations......................................................... 176iSCSI LUN clones............................................................................176

Snapshot locks........................................................................................... 176Snapshot reserve........................................................................................ 177SnapshotIQ license functionality.................................................................177Creating snapshots with SnapshotIQ...........................................................177

Create a SnapRevert domain.......................................................... 178Create a snapshot schedule........................................................... 178Create a snapshot.......................................................................... 179Snapshot naming patterns............................................................. 180

Managing snapshots ..................................................................................182Reducing snapshot disk-space usage............................................ 182Delete snapshots........................................................................... 183Modify snapshot attributes............................................................ 183Modify a snapshot alias................................................................. 184View snapshots..............................................................................184Snapshot information.................................................................... 184

Restoring snapshot data............................................................................. 184Revert a snapshot.......................................................................... 185Restore a file or directory using Windows Explorer..........................185Restore a file or directory through a UNIX command line.................186Clone a file from a snapshot...........................................................186

Managing snapshot schedules....................................................................186Modify a snapshot schedule.......................................................... 186Delete a snapshot schedule........................................................... 187View snapshot schedules...............................................................187

Managing with snapshot locks.................................................................... 187Create a snapshot lock...................................................................187Modify a snapshot lock expiration date..........................................188Delete a snapshot lock...................................................................188Snapshot lock information............................................................. 189

Configure SnapshotIQ settings.................................................................... 189SnapshotIQ settings.......................................................................189

Set the snapshot reserve.............................................................................190

Deduplication with SmartDedupe 193

Deduplication overview...............................................................................194Deduplication jobs......................................................................................194Data replication and backup with deduplication..........................................195Snapshots with deduplication.....................................................................195Deduplication considerations......................................................................195Shadow store considerations...................................................................... 196SmartDedupe license functionality..............................................................196Managing deduplication............................................................................. 197

Assess deduplication space savings.............................................. 197Specify deduplication settings....................................................... 197View deduplication space savings..................................................198View a deduplication report........................................................... 198

Chapter 10

CONTENTS


Deduplication job report information............................................. 198Deduplication information............................................................. 199

Data replication with SyncIQ 201

SyncIQ backup and recovery overview.........................................................202Replication policies and jobs...................................................................... 202

Automated replication policies.......................................................203Source and target cluster association.............................................203Full and differential replication.......................................................204Controlling replication job resource consumption.......................... 204Replication reports.........................................................................205

Replication snapshots.................................................................................205Source cluster snapshots...............................................................205Target cluster snapshots................................................................ 206

Data failover and failback with SyncIQ.........................................................206Data failover.................................................................................. 207Data failback..................................................................................207

Recovery times and objectives for SyncIQ....................................................207SyncIQ license functionality........................................................................ 208Creating replication policies........................................................................208

Excluding directories in replication.................................................208Excluding files in replication.......................................................... 209File criteria options........................................................................ 210Configure default replication policy settings...................................212Create a replication policy..............................................................212Create a SyncIQ domain................................................................. 217Assess a replication policy............................................................. 218

Managing replication to remote clusters......................................................218Start a replication job.....................................................................218Pause a replication job...................................................................219Resume a replication job................................................................219Cancel a replication job..................................................................219View active replication jobs............................................................219Replication job information............................................................ 219

Initiating data failover and failback with SyncIQ.......................................... 220Fail over data to a secondary cluster...............................................220Revert a failover operation............................................................. 220Fail back data to a primary cluster.................................................. 221

Performing disaster recovery for SmartLock directories................................222Recover SmartLock directories on a target cluster...........................222Migrate SmartLock directories........................................................223

Managing replication policies..................................................................... 224Modify a replication policy............................................................. 224Delete a replication policy..............................................................224Enable or disable a replication policy.............................................224View replication policies................................................................ 225Replication policy information........................................................225Replication policy settings............................................................. 225

Managing replication to the local cluster..................................................... 228Cancel replication to the local cluster.............................................228Break local target association........................................................ 229View replication policies targeting the local cluster........................ 229Remote replication policy information............................................ 229

Managing replication performance rules..................................................... 230Create a network traffic rule........................................................... 230

Chapter 11

CONTENTS


Create a file operations rule........................................................... 230Modify a performance rule............................................................. 230Delete a performance rule.............................................................. 231Enable or disable a performance rule............................................. 231View performance rules..................................................................231

Managing replication reports.......................................................................231Configure default replication report settings...................................231Delete replication reports...............................................................232View replication reports................................................................. 232Replication report information........................................................232

Managing failed replication jobs................................................................. 233Resolve a replication policy............................................................234Reset a replication policy............................................................... 234Perform a full or differential replication.......................................... 234

Data layout with FlexProtect 237

FlexProtect overview....................................................................................238File striping................................................................................................. 238Requested data protection.......................................................................... 238FlexProtect data recovery.............................................................................239

Smartfail........................................................................................ 239Node failures................................................................................. 239

Requesting data protection......................................................................... 240Requested protection settings.....................................................................240Requested protection disk space usage...................................................... 241

NDMP backup 243

NDMP backup and recovery overview.......................................................... 244NDMP two way backup................................................................................ 244Snapshot-based incremental backups........................................................ 245NDMP protocol support............................................................................... 246Supported DMAs......................................................................................... 246NDMP hardware support............................................................................. 247NDMP backup limitations............................................................................247NDMP performance recommendations........................................................ 247Excluding files and directories from NDMP backups.................................... 249Configuring basic NDMP backup settings.................................................... 250

Configure and enable NDMP backup.............................................. 250Disable NDMP backup....................................................................250View NDMP backup settings...........................................................250NDMP backup settings................................................................... 251

Managing NDMP user accounts................................................................... 251Create an NDMP user account........................................................ 251Modify the password of an NDMP user account.............................. 251Delete an NDMP user account........................................................ 251View NDMP user accounts..............................................................252

Managing NDMP backup devices.................................................................252Detect NDMP backup devices.........................................................252Modify the name of an NDMP backup device..................................252Delete an entry for an NDMP backup device................................... 253View NDMP backup devices........................................................... 253NDMP backup device settings........................................................ 253

Managing NDMP backup ports.................................................................... 254Modify NDMP backup port settings................................................ 254

Chapter 12

Chapter 13

CONTENTS


Enable or disable an NDMP backup port.........................................254View NDMP backup ports............................................................... 254NDMP backup port settings............................................................254

Managing NDMP backup sessions...............................................................255End an NDMP session.................................................................... 255View NDMP sessions......................................................................255NDMP session information.............................................................256

Managing restartable backups.................................................................... 257Configure restartable backups....................................................... 257Delete a restartable backup context............................................... 258View restartable backup contexts...................................................258Configure restartable backup settings............................................ 259View restartable backup settings................................................... 259

Sharing tape drives between clusters.......................................................... 259Managing default NDMP settings.................................................................259

Set default NDMP settings for a directory....................................... 260Modify default NDMP settings for a directory.................................. 260View default NDMP settings for directories..................................... 260NDMP environment variables......................................................... 261

Managing snapshot based incremental backups.........................................263Enable snapshot-based incremental backups for a directory.......... 263Delete snapshots for snapshot-based incremental backups...........263View snapshots for snapshot-based incremental backups............. 264

View NDMP backup logs..............................................................................264

File retention with SmartLock 265

SmartLock overview.................................................................................... 266Compliance mode....................................................................................... 266SmartLock directories................................................................................. 266Replication and backup with SmartLock...................................................... 267

SmartLock replication and backup limitations................................267SmartLock license functionality...................................................................268SmartLock considerations........................................................................... 268Set the compliance clock............................................................................ 268View the compliance clock.......................................................................... 269Creating a SmartLock directory.................................................................... 269

Retention periods...........................................................................269Autocommit time periods...............................................................269Create a SmartLock directory..........................................................270

Managing SmartLock directories................................................................. 271Modify a SmartLock directory......................................................... 271View SmartLock directory settings..................................................271SmartLock directory configuration settings.....................................272

Managing files in SmartLock directories...................................................... 274Set a retention period through a UNIX command line..................... 274Set a retention period through Windows Powershell.......................275Commit a file to a WORM state through a UNIX command line........ 275Commit a file to a WORM state through Windows Explorer..............275Override the retention period for all files in a SmartLock directory.. 276Delete a file committed to a WORM state .......................................276View WORM status of a file.............................................................277

Protection domains 279

Protection domains overview...................................................................... 280

Chapter 14

Chapter 15

CONTENTS


Protection domain considerations...............................................................280Create a protection domain......................................................................... 281Delete a protection domain......................................................................... 281View protection domains.............................................................................281Protection domain types............................................................................. 281

Data-at-rest-encryption 283

Data-at-rest encryption overview................................................................. 284Self-encrypting drives..................................................................................284Data security on self-encrypted drives......................................................... 284Data migration to a self-encrypted-drives cluster.........................................285Chassis and drive states............................................................................. 285Smartfailed drive ERASE and REPLACE state examples................................ 287

SmartQuotas 291

SmartQuotas overview................................................................................ 292Quota types................................................................................................ 292Default quota type.......................................................................................293Usage accounting and limits....................................................................... 294Disk-usage calculations.............................................................................. 296Quota notifications..................................................................................... 297Quota notification rules...............................................................................297Quota reports..............................................................................................298Creating quotas...........................................................................................298

Create an accounting quota........................................................... 299Create an enforcement quota......................................................... 299

Managing quotas........................................................................................ 300Search for quotas...........................................................................300Manage quotas.............................................................................. 301Export a quota configuration file.................................................... 302Import a quota configuration file.................................................... 302

Managing quota notifications......................................................................303Configure default quota notification settings..................................303Configure custom quota notification rules...................................... 304Map an email notification rule for a quota...................................... 305Configure a custom email quota notification template....................305

Managing quota reports.............................................................................. 306Create a quota report schedule...................................................... 306Generate a quota report................................................................. 307Locate a quota report..................................................................... 307

Basic quota settings....................................................................................307Advisory limit quota notification rules settings............................................ 308Soft limit quota notification rules settings................................................... 309Hard limit quota notification rules settings..................................................309Limit notification settings............................................................................310Quota report settings.................................................................................. 310Custom email notification template variable descriptions........................... 311

Storage Pools 313

Storage pools overview............................................................................... 314About storage pools.................................................................................... 314Autoprovisioning.........................................................................................315Virtual hot spare..........................................................................................316

Chapter 16

Chapter 17

Chapter 18

CONTENTS


Spillover..................................................................................................... 316Node pools................................................................................................. 317

Manual node pool management.....................................................317SSD pools................................................................................................... 318Tiers............................................................................................................319File pools.................................................................................................... 319File pool policies......................................................................................... 319Managing node pools..................................................................................320

Add or move node pools in a tier....................................................320Change the name or requested protection of a node pool...............320

Managing tiers............................................................................................ 321Create a tier................................................................................... 321Rename a tier.................................................................................321Delete a tier................................................................................... 321

Creating file pool policies............................................................................322Add a file pool policy......................................................................323File pool file-matching options....................................................... 324Valid wildcard characters............................................................... 325Default file pool requested protection settings............................... 326Default file pool I/O optimization settings...................................... 327

Managing file pool policies......................................................................... 327Configure default file pool policy settings.......................................328Configure default file pool protection settings................................ 328Configure default I/O optimization settings....................................328Modify a file pool policy................................................................. 329Copy a file pool policy.................................................................... 329Prioritize a file pool policy.............................................................. 329Use a file pool template policy....................................................... 330Delete a file pool policy..................................................................330SmartPools settings....................................................................... 331

Monitoring storage pools............................................................................ 332Monitor storage pools.................................................................... 333View unhealthy subpools............................................................... 333View file pool job results................................................................ 333

System jobs 335

System jobs overview..................................................................................336System jobs library......................................................................................336Job operation.............................................................................................. 339Job performance impact.............................................................................. 340Job priorities............................................................................................... 340Managing system jobs................................................................................ 341

View active jobs............................................................................. 341View job history............................................................................. 341Start a job...................................................................................... 341Pause a job.................................................................................... 342Resume a job................................................................................. 342Cancel a job................................................................................... 342Update a job.................................................................................. 342Modify job type settings.................................................................343

Managing impact policies........................................................................... 343Create an impact policy..................................................................343Copy an impact policy.................................................................... 344Modify an impact policy................................................................. 344Delete an impact policy..................................................................345

Chapter 19

CONTENTS


View impact policy settings............................................................345Viewing job reports and statistics................................................................345

View statistics for a job in progress................................................ 346View a report for a completed job...................................................346

Networking 347

Networking overview................................................................................... 348About the internal network.......................................................................... 348

Internal IP address ranges..............................................................348Internal network failover................................................................ 349

External client network overview................................................................. 349External network settings............................................................... 349IP address pools............................................................................ 350Connection balancing with SmartConnect...................................... 350External IP failover......................................................................... 351NIC aggregation............................................................................. 352VLANs............................................................................................ 353DNS name resolution..................................................................... 353IPv6 support.................................................................................. 353

Configuring the internal network................................................................. 353Modify the internal IP address range.............................................. 354Modify the internal network netmask............................................. 354Configure and enable an internal failover network..........................355Disable internal network failover....................................................355

Configuring an external network.................................................................. 356Adding a subnet.............................................................................356Managing external network subnets...............................................361Managing IP address pools............................................................ 364Managing network interface members............................................366

Managing external client connections with SmartConnect........................... 370Configure client connection balancing............................................370

Managing network interface provisioning rules............................................372Create a node provisioning rule......................................................372Modify a node provisioning rule..................................................... 373Delete a node provisioning rule......................................................374

Hadoop 375

Hadoop overview........................................................................................ 376OneFS Hadoop support............................................................................... 376Hadoop cluster integration..........................................................................377Managing HDFS...........................................................................................377

Configure the HDFS protocol.......................................................... 377Create a local user......................................................................... 378Enable or disable the HDFS service................................................ 379

Securing HDFS connections through Kerberos............................................. 379Configuring HDFS authentication with MIT Kerberos....................... 379Configuring HDFS authentication with Active Directory Kerberos.....381Modifying Hadoop configuration files for Kerberos authentication..383Sample commands for configuring MIT Kerberos authentication overHDFS..............................................................................................387Troubleshooting Kerberos authentication.......................................388

Antivirus 389

Chapter 20

Chapter 21

Chapter 22

CONTENTS


Antivirus overview....................................................................................... 390On-access scanning.................................................................................... 390Antivirus policy scanning............................................................................ 391Individual file scanning............................................................................... 391Antivirus scan reports................................................................................. 391ICAP servers................................................................................................ 392Supported ICAP servers...............................................................................392Anitvirus threat responses...........................................................................392Configuring global antivirus settings........................................................... 393

Exclude files from antivirus scans.................................................. 393Configure on-access scanning settings...........................................394Configure antivirus threat response settings.................................. 395Configure antivirus report retention settings...................................395Enable or disable antivirus scanning..............................................395

Managing ICAP servers................................................................................395Add and connect to an ICAP server................................................. 395Test an ICAP server connection.......................................................396Modify ICAP connection settings.................................................... 396Temporarily disconnect from an ICAP server................................... 396Reconnect to an ICAP server........................................................... 396Remove an ICAP server...................................................................396

Create an antivirus policy............................................................................ 397Managing antivirus policies.........................................................................397

Modify an antivirus policy.............................................................. 397Delete an antivirus policy............................................................... 398Enable or disable an antivirus policy.............................................. 398View antivirus policies................................................................... 398

Managing antivirus scans............................................................................398Scan a file...................................................................................... 398Manually run an antivirus policy.....................................................398Stop a running antivirus scan.........................................................399

Managing antivirus threats..........................................................................399Manually quarantine a file..............................................................399Rescan a file.................................................................................. 399Remove a file from quarantine........................................................399Manually truncate a file..................................................................399View threats...................................................................................400Antivirus threat information........................................................... 400

Managing antivirus reports..........................................................................401Export an antivirus report............................................................... 401View antivirus reports.................................................................... 401View antivirus events..................................................................... 401

iSCSI 403

iSCSI overview............................................................................................ 404iSCSI targets and LUNs................................................................................404

SmartConnect and iSCSI targets.....................................................405iSNS client service.......................................................................................405Access control for iSCSI targets................................................................... 405

CHAP authentication...................................................................... 405Initiator access control...................................................................406

iSCSI considerations and limitations........................................................... 406Supported SCSI mode pages....................................................................... 406Supported iSCSI initiators........................................................................... 407Configuring the iSCSI and iSNS services...................................................... 407

Chapter 23

CONTENTS


Configure the iSCSI service.............................................................407Configure the iSNS client service....................................................408View iSCSI sessions and throughput.............................................. 408

Managing iSCSI targets............................................................................... 409Create an iSCSI target.................................................................... 409Modify iSCSI target settings........................................................... 410Delete an iSCSI target.....................................................................411View iSCSI target settings...............................................................411

Configuring iSCSI initiator access control.................................................... 411Configure iSCSI initiator access control.......................................... 412Control initiator access to a target..................................................412Modify initiator name.....................................................................413Remove an initiator from the access list......................................... 413Create a CHAP secret......................................................................413Modify a CHAP secret..................................................................... 414Delete a CHAP secret......................................................................414Enable or disable CHAP authentication.......................................... 414

Creating iSCSI LUNs.................................................................................... 415Create an iSCSI LUN....................................................................... 415Clone an iSCSI LUN........................................................................ 417iSCSI LUN cloning operations......................................................... 418

Managing iSCSI LUNs.................................................................................. 418Modify an iSCSI LUN.......................................................................419Delete an iSCSI LUN....................................................................... 419Migrate an iSCSI LUN to another target...........................................419Import an iSCSI LUN....................................................................... 420View iSCSI LUN settings................................................................. 420

VMware integration 423

VMware integration overview.......................................................................424VAAI............................................................................................................424

VAAI support for block storage....................................................... 424VAAI support for NAS......................................................................425

VASA...........................................................................................................425Isilon VASA alarms.........................................................................425VASA storage capabilities.............................................................. 425

Configuring VASA support........................................................................... 426Enable VASA.................................................................................. 426Download the Isilon vendor provider certificate..............................426Add the Isilon vendor provider....................................................... 426

Disable or re-enable VASA...........................................................................427

File System Explorer 429

File System Explorer overview......................................................................430Browse the file system................................................................................ 430Create a directory........................................................................................ 430Modify file and directory properties............................................................. 431View file and directory properties................................................................ 431File and directory properties........................................................................431

Chapter 24

Chapter 25

CONTENTS


CONTENTS


CHAPTER 1

Introduction to this guide

This section contains the following topics:

u About this guide....................................................................................................18u Isilon scale-out NAS overview................................................................................18u Where to go for support.........................................................................................18

Introduction to this guide 17

About this guideThis guide describes how the Isilon OneFS web administration interface provides accessto cluster configuration, management, and monitoring functionality.

Isilon scale-out NAS overviewThe EMC Isilon scale-out NAS storage platform combines modular hardware with unifiedsoftware to harness unstructured data. Powered by the distributed OneFS operatingsystem, an EMC Isilon cluster delivers a scalable pool of storage with a globalnamespace.

The platform's unified software provides centralized web-based and command-lineadministration to manage the following features:

u A symmetrical cluster that runs a distributed file system

u Scale-out nodes that add capacity and performance

u Storage options that manage files, block data, and tiering

u Flexible data protection and high availability

u Software modules that control costs and optimize resources

Where to go for supportYou can contact EMC Isilon Technical Support for any questions about EMC Isilonproducts.

Online Support Live Chat

Create a Service Request

Telephone Support United States: 800-782-4362 (1-800-SVC-4EMC)

Canada: 800-543-4782

Worldwide: +1-508-497-7901

For local phone numbers in your country, see EMC CustomerSupport Centers.

Help with onlinesupport

For questions specific to EMC Online Support registration oraccess, email [email protected].

Introduction to this guide


https://support.emc.com/servicecenter/liveChat/

https://support.emc.com/servicecenter/createSR/

http://www.emc.com/collateral/contact-us/h4165-csc-phonelist-ho.pdf

http://www.emc.com/collateral/contact-us/h4165-csc-phonelist-ho.pdf

mailto:[email protected]

CHAPTER 2

Isilon scale-out NAS


u OneFS storage architecture................................................................................... 20u Isilon node components........................................................................................20u Internal and external networks.............................................................................. 21u Isilon cluster......................................................................................................... 21u The OneFS operating system................................................................................. 23u Structure of the file system....................................................................................25u Data protection overview.......................................................................................27u VMware integration............................................................................................... 29u The iSCSI option....................................................................................................30u Software modules................................................................................................. 30

Isilon scale-out NAS 19

OneFS storage architectureEMC Isilon takes a scale-out approach to storage by creating a cluster of nodes that runsa distributed file system. OneFS combines the three layers of storage architecture—filesystem, volume manager, and data protection—into a scale-out NAS cluster.

Each node adds resources to the cluster. Because each node contains globally coherentRAM, as a cluster becomes larger, it becomes faster. Meanwhile, the file system expandsdynamically and redistributes content, which eliminates the work of partitioning disksand creating volumes.

Nodes work as peers to spread data across the cluster. Segmenting and distributing data—a process known as striping—not only protects data, but also enables a userconnecting to any node to take advantage of the entire cluster's performance.

OneFS uses distributed software to scale data across commodity hardware. Each nodehelps control data requests, boosts performance, and expands the cluster's capacity. Nomaster device controls the cluster; no slaves invoke dependencies. Instead, each nodehelps control data requests, boosts performance, and expands the cluster's capacity.

Isilon node componentsAs a rack-mountable appliance, a node includes the following components in a 2U or 4Urack-mountable chassis with an LCD front panel: memory, CPUs, RAM, NVRAM, networkinterfaces, InfiniBand adapters, disk controllers, and storage media. An Isilon clustercomprises three or more nodes, up to 144.

When you add a node to a cluster, you increase the cluster's aggregate disk, cache, CPU,RAM, and network capacity. OneFS groups RAM into a single coherent cache so that adata request on a node benefits from data that is cached anywhere. NVRAM is grouped towrite data with high throughput and to protect write operations from power failures. Asthe cluster expands, spindles and CPU combine to increase throughput, capacity, andinput-output operations per second (IOPS).

EMC Isilon makes several types of nodes, all of which can be added to a cluster tobalance capacity and performance with throughput or IOPS:

Node Use Case- -S-Series IOPS-intensive applications

X-Series High-concurrency and throughput-driven workflows

NL-Series Near-primary accessibility, with near-tape value

The following EMC Isilon nodes improve performance:

Node Function- -Performance Accelerator Independent scaling for high performance

Backup Accelerator High-speed and scalable backup-and-restore solution



Internal and external networksA cluster includes two networks: an internal network to exchange data between nodesand an external network to handle client connections.

Nodes exchange data through the internal network with a proprietary, unicast protocolover InfiniBand. Each node includes redundant InfiniBand ports so you can add a secondinternal network in case the first one fails.

Clients reach the cluster with 1 GigE or 10 GigE Ethernet. Since every node includesEthernet ports, the cluster's bandwidth scales with performance and capacity as you addnodes.

Isilon clusterAn Isilon cluster consists of three or more hardware nodes, up to 144. Each node runs theIsilon OneFS operating system, the distributed file-system software that unites the nodesinto a cluster. A cluster’s storage capacity ranges from a minimum of 18 TB to a maximumof 15.5 PB.

Cluster administrationOneFS centralizes cluster management through a web administration interface and acommand-line interface. Both interfaces provide methods to activate licenses, check thestatus of nodes, configure the cluster, upgrade the system, generate alerts, view clientconnections, track performance, and change various settings.

In addition, OneFS simplifies administration by automating maintenance with a jobengine. You can schedule jobs that scan for viruses, inspect disks for errors, reclaim diskspace, and check the integrity of the file system. The engine manages the jobs tominimize impact on the cluster's performance.

With SNMP versions 1, 2c, and 3, you can remotely monitor hardware components, CPUusage, switches, and network interfaces. EMC Isilon supplies management informationbases (MIBs) and traps for the OneFS operating system.

OneFS also includes a RESTful application programming interface—known as the PlatformAPI—to automate access, configuration, and monitoring. For example, you can retrieveperformance statistics, provision users, and tap the file system. The Platform APIintegrates with OneFS role-based access control to increase security. See the IsilonPlatform API Reference.

QuorumAn Isilon cluster must have a quorum to work properly. A quorum prevents data conflicts—for example, conflicting versions of the same file—in case two groups of nodes becomeunsynchronized. If a cluster loses its quorum for read and write requests, you cannotaccess the OneFS file system.

For a quorum, more than half the nodes must be available over the internal network. Aseven-node cluster, for example, requires a four-node quorum. A 10-node cluster requiresa six-node quorum. If a node is unreachable over the internal network, OneFS separatesthe node from the cluster, an action referred to as splitting. After a cluster is split, clusteroperations continue as long as enough nodes remain connected to have a quorum.

In a split cluster, the nodes that remain in the cluster are referred to as the majoritygroup. Nodes that are split from the cluster are referred to as the minority group.


Internal and external networks 21

When split nodes can reconnect with the cluster and resynchronize with the other nodes,the nodes rejoin the cluster's majority group, an action referred to as merging.

A OneFS cluster contains two quorum properties:

u read quorum (efs.gmp.has_quorum)

u write quorum (efs.gmp.has_super_block_quorum)

By connecting to a node with SSH and running the sysctl command-line tool as root,you can view the status of both types of quorum. Here is an example for a cluster that hasa quorum for both read and write operations, as the command's output indicates with a1, for true:

sysctl efs.gmp.has_quorum efs.gmp.has_quorum: 1 sysctl efs.gmp.has_super_block_quorum efs.gmp.has_super_block_quorum: 1

The degraded states of nodes—such as smartfail, read-only, offline, and so on—affectquorum in different ways. A node in a smartfail or read-only state affects only writequorum. A node in an offline state, however, affects both read and write quorum. In acluster, the combination of nodes in different degraded states determines whether readrequests, write requests, or both work.

A cluster can lose write quorum but keep read quorum. Consider a four-node cluster inwhich nodes 1 and 2 are working normally. Node 3 is in a read-only state, and node 4 isin a smartfail state. In such a case, read requests to the cluster succeed. Write requests,however, receive an input-output error because the states of nodes 3 and 4 break thewrite quorum.

A cluster can also lose both its read and write quorum. If nodes 3 and 4 in a four-nodecluster are in an offline state, both write requests and read requests receive an input-output error, and you cannot access the file system. When OneFS can reconnect with thenodes, OneFS merges them back into the cluster. Unlike a RAID system, an Isilon nodecan rejoin the cluster without being rebuilt and reconfigured.

Splitting and mergingSplitting and merging optimize the use of nodes without your intervention.

OneFS monitors every node in a cluster. If a node is unreachable over the internalnetwork, OneFS separates the node from the cluster, an action referred to as splitting.When the cluster can reconnect to the node, OneFS adds the node back into the cluster,an action referred to as merging.

When a node is split from a cluster, it will continue to capture event information locally.You can connect to a split node with SSH and run the isi events list command toview the local event log for the node. The local event log can help you troubleshoot theconnection issue that resulted in the split. When the split node rejoins the cluster, localevents gathered during the split are deleted. You can still view events generated by asplit node in the node's event log file located at /var/log/isi_celog_events.log.

If a cluster splits during a write operation, OneFS might need to re-allocate blocks for thefile on the side with the quorum, which leads allocated blocks on the side without aquorum to become orphans. When the split nodes reconnect with the cluster, the OneFSCollect system job reclaims the orphaned blocks.

Meanwhile, as nodes split and merge with the cluster, the OneFS AutoBalance jobredistributes data evenly among the nodes in the cluster, optimizing protection andconserving space.



Storage poolsStorage pools segment nodes and files into logical divisions to simplify the managementand storage of data.

A storage pool comprises node pools and tiers. Node pools group equivalent nodes toprotect data and ensure reliability. Tiers combine node pools to optimize storage byneed, such as a frequently used high-speed tier or a rarely accessed archive.

The SmartPools module groups nodes and files into pools. If you do not activate aSmartPools license, the module provisions node pools and creates one file pool. If youactivate the SmartPools license, you receive more features. You can, for example, createmultiple file pools and govern them with policies. The policies move files, directories, andfile pools among node pools or tiers. You can also define how OneFS handles writeoperations when a node pool or tier is full. SmartPools reserves a virtual hot spare toreprotect data if a drive fails regardless of whether the SmartPools license is activated.

IP address poolsWithin a subnet, you can partition a cluster's external network interfaces into pools of IPaddress ranges. The pools empower you to customize your storage network to servedifferent groups of users. Although you must initially configure the default external IPsubnet in IPv4 format, you can configure additional subnets in IPv4 or IPv6.

You can associate IP address pools with a node, a group of nodes, or NIC ports. Forexample, you can set up one subnet for storage nodes and another subnet for acceleratornodes. Similarly, you can allocate ranges of IP addresses on a subnet to different teams,such as engineering and sales. Such options help you create a storage topology thatmatches the demands of your network.

In addition, network provisioning rules streamline the setup of external connections.After you configure the rules with network settings, you can apply the settings to newnodes.

As a standard feature, the OneFS SmartConnect module balances connections amongnodes by using a round-robin policy with static IP addresses and one IP address pool foreach subnet. Activating a SmartConnect Advanced license adds features, such asdefining IP address pools to support multiple DNS zones.

The OneFS operating systemA distributed operating system based on FreeBSD, OneFS presents an Isilon cluster's filesystem as a single share or export with a central point of administration.

The OneFS operating system does the following:

u Supports common data-access protocols, such as SMB and NFS.

u Connects to multiple identity management systems, such as Active Directory andLDAP.

u Authenticates users and groups.

u Controls access to directories and files.


Storage pools 23

Data-access protocolsWith the OneFS operating system, you can access data with multiple file-sharing andtransfer protocols. As a result, Microsoft Windows, UNIX, Linux, and Mac OS X clients canshare the same directories and files.

OneFS supports the following protocols.

Protocol Description- -SMB Server Message Block gives Windows users access to the cluster. OneFS works with

SMB 1, SMB 2, and SMB 2.1. With SMB 2.1, OneFS supports client opportunity locks(oplocks) and large (1 MB) MTU sizes. The default file share is /ifs.

NFS The Network File System enables UNIX, Linux, and Mac OS X systems to remotelymount any subdirectory, including subdirectories created by Windows users. OneFSworks with versions 2 through 4 of the Network File System protocol (NFSv2, NFSv3,NFSv4). The default export is /ifs.

FTP File Transfer Protocol lets systems with an FTP client connect to the cluster to exchangefiles.

iSCSI The Internet Small Computer System Interface protocol provides access to blockstorage. iSCSI integration requires you to activate a separate license.

HDFS The Hadoop Distributed File System protocol makes it possible for a cluster to workwith Apache Hadoop, a framework for data-intensive distributed applications. HDFSintegration requires you to activate a separate license.

HTTP Hyper Text Transfer protocol gives systems browser-based access to resources. OneFSincludes limited support for WebDAV.

Identity management and access controlOneFS works with multiple identity management systems to authenticate users andcontrol access to files. In addition, OneFS features access zones that allow users fromdifferent directory services to access different resources based on their IP address. Role-based access control, meanwhile, segments administrative access by role.

OneFS authenticates users with the following identity management systems:

u Microsoft Active Directory (AD)

u Lightweight Directory Access Protocol (LDAP)

u Network Information Service (NIS)

u Local users and local groups

u A file provider for accounts in /etc/spwd.db and /etc/group files. With the fileprovider, you can add an authoritative third-party source of user and groupinformation.

You can manage users with different identity management systems; OneFS maps theaccounts so that Windows and UNIX identities can coexist. A Windows user accountmanaged in Active Directory, for example, is mapped to a corresponding UNIX account inNIS or LDAP.

To control access, an Isilon cluster works with both the access control lists (ACLs) ofWindows systems and the POSIX mode bits of UNIX systems. When OneFS must



transform a file's permissions from ACLs to mode bits or from mode bits to ACLs, OneFSmerges the permissions to maintain consistent security settings.

OneFS presents protocol-specific views of permissions so that NFS exports display modebits and SMB shares show ACLs. You can, however, manage not only mode bits but alsoACLs with standard UNIX tools, such as the chmod and chown commands. In addition,ACL policies enable you to configure how OneFS manages permissions for networks thatmix Windows and UNIX systems.

Access zonesOneFS includes an access zones feature. Access zones allow users from differentauthentication providers, such as two untrusted Active Directory domains, to accessdifferent OneFS resources based on an incoming IP address. An access zone cancontain multiple authentication providers and SMB namespaces.

RBAC for administrationOneFS includes role-based access control (RBAC) for administration. In place of aroot or administrator account, RBAC lets you manage administrative access by role.A role limits privileges to an area of administration. For example, you can createseparate administrator roles for security, auditing, storage, and backup.

Structure of the file systemOneFS presents all the nodes in a cluster as a global namespace—that is, as the defaultfile share, /ifs.

In the file system, directories are inode number links. An inode contains file metadataand an inode number, which identifies a file's location. OneFS dynamically allocatesinodes, and there is no limit on the number of inodes.

To distribute data among nodes, OneFS sends messages with a globally routable blockaddress through the cluster's internal network. The block address identifies the node andthe drive storing the block of data.

Note

It is recommended that you do not save data to the root /ifs file path but in directories

below /ifs. The design of your data storage structure should be planned carefully. A

well-designed directory optimizes cluster performance and cluster administration.

Data layoutOneFS evenly distributes data among a cluster's nodes with layout algorithms thatmaximize storage efficiency and performance. The system continuously reallocates datato conserve space.

OneFS breaks data down into smaller sections called blocks, and then the system placesthe blocks in a stripe unit. By referencing either file data or erasure codes, a stripe unithelps safeguard a file from a hardware failure. The size of a stripe unit depends on thefile size, the number of nodes, and the protection setting. After OneFS divides the datainto stripe units, OneFS allocates, or stripes, the stripe units across nodes in the cluster.

When a client connects to a node, the client's read and write operations take place onmultiple nodes. For example, when a client connects to a node and requests a file, thenode retrieves the data from multiple nodes and rebuilds the file. You can optimize howOneFS lays out data to match your dominant access pattern—concurrent, streaming, orrandom.


Structure of the file system 25

Writing filesOn a node, the input-output operations of the OneFS software stack split into twofunctional layers: A top layer, or initiator, and a bottom layer, or participant. In read andwrite operations, the initiator and the participant play different roles.

When a client writes a file to a node, the initiator on the node manages the layout of thefile on the cluster. First, the initiator divides the file into blocks of 8 KB each. Second, theinitiator places the blocks in one or more stripe units. At 128 KB, a stripe unit consists of16 blocks. Third, the initiator spreads the stripe units across the cluster until they span awidth of the cluster, creating a stripe. The width of the stripe depends on the number ofnodes and the protection setting.

After dividing a file into stripe units, the initiator writes the data first to non-volatilerandom-access memory (NVRAM) and then to disk. NVRAM retains the information whenthe power is off.

During the write transaction, NVRAM guards against failed nodes with journaling. If anode fails mid-transaction, the transaction restarts without the failed node. When thenode returns, it replays the journal from NVRAM to finish the transaction. The node alsoruns the AutoBalance job to check the file's on-disk striping. Meanwhile, uncommittedwrites waiting in the cache are protected with mirroring. As a result, OneFS eliminatesmultiple points of failure.

Reading filesIn a read operation, a node acts as a manager to gather data from the other nodes andpresent it to the requesting client.

Because an Isilon cluster's coherent cache spans all the nodes, OneFS can store differentdata in each node's RAM. By using the internal InfiniBand network, a node can retrievefile data from another node's cache faster than from its own local disk. If a read operationrequests data that is cached on any node, OneFS pulls the cached data to serve itquickly.

In addition, for files with an access pattern of concurrent or streaming, OneFS pre-fetchesin-demand data into a managing node's local cache to further improve sequential-readperformance.

Metadata layoutOneFS protects metadata by spreading it across nodes and drives.

Metadata—which includes information about where a file is stored, how it is protected,and who can access it—is stored in inodes and protected with locks in a B+ tree, astandard structure for organizing data blocks in a file system to provide instant lookups.OneFS replicates file metadata across the cluster so that there is no single point offailure.

Working together as peers, all the nodes help manage metadata access and locking. If anode detects an error in metadata, the node looks up the metadata in an alternatelocation and then corrects the error.



Locks and concurrencyOneFS includes a distributed lock manager that orchestrates locks on data across all thenodes in a cluster.

The lock manager grants locks for the file system, byte ranges, and protocols, includingSMB share-mode locks and NFS advisory locks. OneFS also supports SMB opportunisticlocks and NFSv4 delegations.

Because OneFS distributes the lock manager across all the nodes, any node can act as alock coordinator. When a thread from a node requests a lock, the lock manager's hashingalgorithm typically assigns the coordinator role to a different node. The coordinatorallocates a shared lock or an exclusive lock, depending on the type of request. A sharedlock allows users to share a file simultaneously, typically for read operations. Anexclusive lock allows only one user to access a file, typically for write operations.

StripingIn a process known as striping, OneFS segments files into units of data and thendistributes the units across nodes in a cluster. Striping protects your data and improvescluster performance.

To distribute a file, OneFS reduces it to blocks of data, arranges the blocks into stripeunits, and then allocates the stripe units to nodes over the internal network.

At the same time, OneFS distributes erasure codes that protect the file. The erasure codesencode the file's data in a distributed set of symbols, adding space-efficient redundancy.With only a part of the symbol set, OneFS can recover the original file data.

Taken together, the data and its redundancy form a protection group for a region of filedata. OneFS places the protection groups on different drives on different nodes—creatingdata stripes.

Because OneFS stripes data across nodes that work together as peers, a user connectingto any node can take advantage of the entire cluster's performance.

By default, OneFS optimizes striping for concurrent access. If your dominant accesspattern is streaming--that is, lower concurrency, higher single-stream workloads, such aswith video--you can change how OneFS lays out data to increase sequential-readperformance. To better handle streaming access, OneFS stripes data across more drives.Streaming is most effective on clusters or subpools serving large files.

Data protection overviewAn Isilon cluster is designed to serve data even when components fail. By default, OneFSprotects data with erasure codes, enabling you to retrieve files when a node or disk fails.As an alternative to erasure codes, you can protect data with two to eight mirrors.

When you create a cluster with five or more nodes, erasure codes deliver as much as 80percent efficiency. On larger clusters, erasure codes provide as much as four levels ofredundancy.

In addition to erasure codes and mirroring, OneFS includes the following features to helpprotect the integrity, availability, and confidentiality of data:

Feature Description- -Antivirus OneFS can send files to servers running the Internet Content Adaptation

Protocol (ICAP) to scan for viruses and other threats.


Locks and concurrency 27

Feature Description- -Clones OneFS enables you to create clones that share blocks with other files to save

space.

NDMP backup andrestore

OneFS can back up data to tape and other devices through the Network DataManagement Protocol. Although OneFS supports both NDMP 3-way and 2-way backup, 2-way backup requires an Isilon Backup Accelerator node.

Protectiondomains

You can apply protection domains to files and directories to preventchanges.

The following software modules also help protect data, but they require you to activate aseparate license:

LicensedFeature

Description

- -SyncIQ SyncIQ replicates data on another Isilon cluster and automates failover and

failback operations between clusters. If a cluster becomes unusable, you canfail over to another Isilon cluster.

SnapshotIQ You can protect data with a snapshot—a logical copy of data stored on acluster.

SmartLock The SmartLock tool prevents users from modifying and deleting files. You cancommit files to a write-once, read-many state: The file can never be modifiedand cannot be deleted until after a set retention period. SmartLock can helpyou comply with Securities and Exchange Commission Rule 17a-4.

N+M data protectionOneFS supports N+M erasure code levels of N+1, N+2, N+3, and N+4.

In the N+M data model, N represents the number of nodes, and M represents the numberof simultaneous failures of nodes or drives that the cluster can handle without losingdata. For example, with N+2 the cluster can lose two drives on different nodes or lose twonodes.

To protect drives and nodes separately, OneFS also supports N+M:B. In the N+M:Bnotation, M is the number of disk failures, and B is the number of node failures. With N+3:1 protection, for example, the cluster can lose three drives or one node without losingdata.

The default protection level for clusters larger than 18 TB is N+2:1. The default forclusters smaller than 18 TB is N+1.

The quorum rule dictates the number of nodes required to support a protection level. Forexample, N+3 requires at least seven nodes so you can maintain a quorum if three nodesfail.

You can, however, set a protection level that is higher than the cluster can support. In afour-node cluster, for example, you can set the protection level at 5x. OneFS protects thedata at 4x until a fifth node is added, after which OneFS automatically reprotects the dataat 5x.



Data mirroringYou can protect on-disk data with mirroring, which copies data to multiple locations.OneFS supports two to eight mirrors. You can use mirroring instead of erasure codes, oryou can combine erasure codes with mirroring.

Mirroring, however, consumes more space than erasure codes. Mirroring data threetimes, for example, duplicates the data three times, which requires more space thanerasure codes. As a result, mirroring suits transactions that require high performance,such as with iSCSI LUNs.

You can also mix erasure codes with mirroring. During a write operation, OneFS dividesdata into redundant protection groups. For files protected by erasure codes, a protectiongroup consists of data blocks and their erasure codes. For mirrored files, a protectiongroup contains all the mirrors of a set of blocks. OneFS can switch the type of protectiongroup as it writes a file to disk. By changing the protection group dynamically, OneFS cancontinue writing data despite a node failure that prevents the cluster from applyingerasure codes. After the node is restored, OneFS automatically converts the mirroredprotection groups to erasure codes.

The file system journalA journal, which records file-system changes in a battery-backed NVRAM card, recoversthe file system after failures, such as a power loss. When a node restarts, the journalreplays file transactions to restore the file system.

Virtual hot spareWhen a drive fails, OneFS uses space reserved in a subpool instead of a hot spare drive.The reserved space is known as a virtual hot spare.

In contrast to a spare drive, a virtual hot spare automatically resolves drive failures andcontinues writing data. If a drive fails, OneFS migrates data to the virtual hot spare toreprotect it. You can reserve as many as four disk drives as a virtual hot spare.

Balancing protection with storage spaceYou can set protection levels to balance protection requirements with storage space.

Higher protection levels typically consume more space than lower levels because youlose an amount of disk space to storing erasure codes. The overhead for the erasurecodes depends on the protection level, the file size, and the number of nodes in thecluster. Since OneFS stripes both data and erasure codes across nodes, the overheaddeclines as you add nodes.

VMware integrationOneFS integrates with several VMware products, including vSphere, vCenter, and ESXi.

For example, OneFS works with the VMware vSphere API for Storage Awareness (VASA) sothat you can view information about an Isilon cluster in vSphere. OneFS also works withthe VMware vSphere API for Array Integration (VAAI) to support the following features forblock storage: hardware-assisted locking, full copy, and block zeroing. VAAI for NFSrequires an ESXi plug-in.

With the Isilon for vCenter plug-in, you can backup and restore virtual machines on anIsilon cluster. With the Isilon Storage Replication Adapter, OneFS integrates with the


Data mirroring 29

VMware vCenter Site Recovery Manager to recover virtual machines that are replicatedbetween Isilon clusters.

The iSCSI optionBlock-based storage offers flexible storage and access. OneFS enables clients to storeblock data on an Isilon cluster by using the Internet Small Computer System Interface(iSCSI) protocol. With the iSCSI module, you can configure block storage for Windows,Linux, and VMware systems.

On the network side, the logical network interface (LNI) framework dynamically managesinterfaces for network resilience. You can combine multiple network interfaces with LACPand LAGG to aggregate bandwidth and to fail over client sessions. The iSCSI modulerequires you to activate a separate license.

Software modulesYou can access advanced features by activating licenses for EMC Isilon softwaremodules.

SmartLockSmartLock protects critical data from malicious, accidental, or premature alterationor deletion to help you comply with SEC 17a-4 regulations. You can automaticallycommit data to a tamper-proof state and then retain it with a compliance clock.

SyncIQ automated failover and failbackSyncIQ replicates data on another Isilon cluster and automates failover and failbackbetween clusters. If a cluster becomes unusable, you can fail over to another Isiloncluster. Failback restores the original source data after the primary cluster becomesavailable again.

File clonesOneFS provides provisioning of full read/write copies of files, LUNs, and otherclones. OneFS also provides virtual machine linked cloning through VMware APIintegration.

SnapshotIQSnapshotIQ protects data with a snapshot—a logical copy of data stored on acluster. A snapshot can be restored to its top-level directory.

SmartPoolsSmartPools enable you to create multiple file pools governed by file-pool policies.The policies move files and directories among node pools or tiers. You can alsodefine how OneFS handles write operations when a node pool or tier is full.

SmartConnectIf you activate a SmartConnect Advanced license, you can balance policies to evenlydistribute CPU usage, client connections, or throughput. You can also define IPaddress pools to support multiple DNS zones in a subnet. In addition, SmartConnectsupports IP failover, also known as NFS failover.

InsightIQThe InsightIQ virtual appliance monitors and analyzes the performance of your Isiloncluster to help you optimize storage resources and forecast capacity.



Aspera for IsilonAspera moves large files over long distances fast. Aspera for Isilon is a cluster-awareversion of Aspera technology for non-disruptive, wide-area content delivery.

iSCSIOneFS supports the Internet Small Computer System Interface (iSCSI) protocol toprovide block storage for Windows, Linux, and VMware clients. The iSCSI moduleincludes parallel LUN allocation and zero-copy support.

HDFSOneFS works with the Hadoop Distributed File System protocol to help clientsrunning Apache Hadoop, a framework for data-intensive distributed applications,analyze big data.

SmartQuotasThe SmartQuotas module tracks disk usage with reports and enforces storage limitswith alerts.


Software modules 31

CHAPTER 3

General cluster administration


u General cluster administration overview................................................................34u User interfaces...................................................................................................... 34u Connecting to the cluster.......................................................................................35u Licensing...............................................................................................................36u Certificates............................................................................................................40u General cluster settings.........................................................................................42u Cluster statistics................................................................................................... 47u Performance monitoring........................................................................................ 47u Cluster monitoring.................................................................................................47u Monitoring cluster hardware..................................................................................57u Cluster maintenance............................................................................................. 62u Remote support using SupportIQ...........................................................................64u Upgrading OneFS.................................................................................................. 68

General cluster administration 33

General cluster administration overviewYou can manage general OneFS settings and module licenses for the EMC Isilon cluster.

General cluster administration covers several areas. You can manage general settingssuch as cluster name, date and time, and email. You can monitor the cluster status andperformance, including hardware components. You can configure how events andnotifications are handled, and you can perform cluster maintenance such as adding,removing, and restarting nodes.

Most management tasks are accomplished through both the web administration orcommand-line interface; however, you will occasionally encounter a task that can only bemanaged by one or the other.

User interfacesDepending on your preference, location, or task, OneFS provides several interfaces formanaging the EMC Isilon cluster.

Interface Description Comment- - -OneFS webadministrationinterface

The browser-based OneFS webadministration interface providessecure access with OneFS-supportedbrowsers. You can use this interface toview robust graphical monitoringdisplays and to perform cluster-management tasks.

The OneFS web administrationinterface uses port 8080 as itsdefault port.

OneFS command-line interface

You can run OneFS isi commands in

the command-line interface toconfigure, monitor, and manage thecluster. Access to the command-lineinterface is through a secure shell(SSH) connection to any node in thecluster.

The OneFS command-line interfaceprovides an extended standardUNIX command set for managingthe cluster.

OneFS Platform API The OneFS Platform API providesaccess to cluster configuration,management, and monitoringfunctionality through an HTTP-basedinterface.

You should have a solidunderstanding of HTTP/1.1 andexperience writing HTTP-basedclient software before youimplement client-based softwarethrough the Platform API.

OneFS RESTfulAccess to theNamespace API

You can create, delete, and modify dataon the OneFS file system through theRESTful Access to the Namespace(RAN) application programing interface(API).

You should have a solidunderstanding of HTTP/1.1 andexperience writing HTTP-basedclient software before youimplement client-based softwarethrough the RAN API.

Node front panel The front panel of each node containsan LCD screen with five buttons, whichyou can use to monitor node andcluster details.

Node status, events, clusterdetails, capacity, IP and MACaddresses, throughput, and drivestatus are available through thenode front panel.



Interface Description Comment- - -

Note

Accelerator nodes do not havefront panels.

Connecting to the clusterEMC Isilon cluster access is provided through the web administration interface or throughSSH. A serial connection can be used to perform cluster-administration tasks through thecommand-line interface.

You can also access the cluster through the node front panel to accomplish a subset ofcluster-management tasks. For information about connecting to the node front panel, seethe installation documentation for your node.

Log in to the web administration interfaceYou can monitor and manage your EMC Isilon cluster from the browser-based webadministration interface.

Procedure

1. Open a browser window and type the URL for your cluster in the address field,replacing <yourNodeIPaddress> in the following example with the first IP address youprovided when you configured ext-1:

http://<yourNodeIPaddress>

If your security certificates have not been configured, a message displays. Resolve anycertificate configurations and continue to the web site.

2. Log in to OneFS by typing your OneFS credentials in the Username and Passwordfields.

After you log into the web administration interface, there is a 4-hour login timeout anda 24-hour session inactivity timeout.

Open an SSH connection to a clusterYou can use any SSH client such as OpenSSH or PuTTY to connect to an EMC Isiloncluster.

Before you begin

You must have valid OneFS credentials to log in to a cluster after the connection is open.

Procedure

1. Open a secure shell (SSH) connection to any node in the cluster, using the IP addressand port number for the node.

2. Log in with your OneFS credentials.

At the OneFS command line prompt, you can use isi commands to monitor andmanage your cluster.


Connecting to the cluster 35

LicensingAdvanced cluster features are available when you activate licenses for OneFS softwaremodules. Each optional OneFS software module requires you to activate a separatelicense.

For more information about the following optional software modules, contact your EMCIsilon sales representative.

u HDFS

u InsightIQ

u Isilon for vCenter

u SmartConnect Advanced

u SmartDedupe

u SmartLock

u SmartPools

u SmartQuotas

u SnapshotIQ

u SyncIQ

u iSCSI

License statusThe status of a OneFS module license indicates whether the functionality provided by themodule are available on the cluster.

Licenses exist in one of the following states:

Status Description- -Inactive The license has not been activated on the cluster. You cannot access the features

provided by the corresponding module.

Evaluation The license has been temporarily activated on the cluster. You can access thefeatures provided by the corresponding module for a limited period of time. After thelicense expires, the features will become unavailable, unless the license isreactivated.

Activated The license has been activated on the cluster. You can access the features providedby the corresponding module.

Expired The evaluation license has expired on the cluster. You can no longer access thefeatures provided by the corresponding module. The features will remainunavailable, unless you reactivate the license.

The following table describes what functionality is available for each license dependingon the license's status:

License Inactive Evaluation/Activated

Expired

- - - -HDFS Clients cannot

access the clusterthrough HDFS.

You can configureHDFS settings andclients can access

You cannot configure HDFSsettings. After the HDFSservice restarts, clients can




Expired

- - - -the cluster throughHDFS.

no longer access the clusterthrough HDFS.

InsightIQ You cannot monitorthe cluster withInsightIQ.

You can monitor thecluster withInsightIQ.

InsightIQ stops monitoringthe cluster. Data previouslycollected by InsightIQ is stillavailable on the InsightIQinstance.

Isilon for vCenter You cannot back upvirtual machines thatare stored on anIsilon cluster withIsilon for vCenter.

You can back upvirtual machines thatare stored on anIsilon cluster withIsilon for vCenter.

You cannot create newbackups of virtual machinesthat are stored on an Isiloncluster.

SmartPools All files belong to thedefault file pool andare governed by thedefault file poolpolicy. Virtual hotspare allocation,which reservesspace for data repairif a drive fails, is alsoavailable.

You can createmultiple file poolsand file poolpolicies. You canalso managespillover, whichdefines how writeoperations arehandled when astorage pool is notwritable.

You can no longer managefile pool policies, and theSmartPools job will nolonger run. Newly addedfiles will be governed by thedefault file pool policy, andthe SetProtectPlus job willeventually apply the defaultfile pool policy to all files inthe cluster.If the SmartPools job isrunning when the licenseexpires, the job completesbefore becoming disabled.

SmartConnectAdvanced

Client connectionsare balanced byusing a round robinpolicy. IP addressallocation is static.Each externalnetwork subnet canbe assigned onlyone IP address pool.

You can accessfeatures such as CPUutilization,connection counting,and clientconnection policiesin addition to theround robin policy.You can alsoconfigure addresspools to supportmultiple DNS zoneswithin a singlesubnet, and supportIP failover.

You can no longer specifySmartConnect Advancedsettings.

SmartDedupe You cannotdeduplicate datawith SmartDedupe.

You can deduplicatedata withSmartDedupe.

You can no longerdeduplicate data. Previouslydeduplicated data remainsdeduplicated.

SmartLock You cannot enforcefile retention withSmartLock.

You can enforce fileretention withSmartLock.

You cannot create newSmartLock directories ormodify SmartLock directory


License status 37


Expired

- - - -configuration settings forexisting directories.You can still commit files toa write once read many(WORM) state, even after theSmartLock license isunconfigured, but youcannot delete WORM-committed files fromenterprise directories.

SnapshotIQ You can view andmanage snapshotsgenerated by OneFSapplications.However, you cannotcreate snapshots orconfigureSnapshotIQ settings.

You can create, view,and managesnapshots. You canalso configuresnapshot settings.

You will no longer be able togenerate snapshots.Existing snapshot schedulesare not deleted; however,the schedules will notgenerate snapshots.You can still deletesnapshots and accesssnapshot data.

SmartQuotas You cannot createquotas withSmartQuotas.

You can createquotas withSmartQuotas.

OneFS disables all quotas.Exceeding advisory and softthresholds does not triggerevents. Hard and softthresholds are not enforced.

SyncIQ You cannot replicatedata with SyncIQ.

You can replicatedata with SyncIQ

You will no longer be able toreplicate data to remoteclusters, and remoteclusters will not be able toreplicate data to the localcluster. Replication policieswill still display a status ofenabled; however, futurereplication jobs created bythe policy will fail.If a replication job is inprogress when the licenseexpires, the job completes.

iSCSI Clients cannotaccess the clusterthrough iSCSI.

Clients can accessthe cluster throughiSCSI.

You can no longer configureiSCSI settings on the cluster.HDFS clients can read datafrom cluster; however, theyare unable to write data tothe cluster.



License configurationYou can configure or unconfigure some OneFS module licenses.

You can configure a license by performing specific operations through the correspondingmodule. Not all actions that require you to activate a license will configure the license.Also, not all licenses can be configured. Configuring a license does not add or removeaccess to any features provided by a module.

You can unconfigure a license only through the isi license unconfigurecommand. You may want to unconfigure a license for a OneFS software module if, forexample, you enabled an evaluation version of a module but later decided not topurchase a permanent license. Unconfiguring a module license does not deactivate thelicense. Unconfiguring a license does not add or remove access to any features providedby a module.

The following table describes both the actions that cause each license to be configuredand the results of unconfiguring each license:

License Cause of configuring Result of unconfiguring- - -HDFS Cannot configure this license. No system impact.

InsightIQ Cannot configure this license. No system impact.

Isilon for vCenter Cannot configure this license. No system impact.

SmartPools Create a file pool policy (other than thedefault file pool policy).

OneFS deletes all file pool policies(except the default file pool policy).

SmartConnect Configure SmartConnect Advancedsettings for at least one IP addresspool.

OneFS converts dynamic IP addresspools to static IP address pools.

SmartDedupe Cannot configure this license. No system impact.

SmartLock Cannot configure this license. No system impact.

SnapshotIQ Create a snapshot schedule. Deletes all snapshot schedules.

SmartQuotas Create a quota. No system impact.

SyncIQ Create a replication policy. No system impact.

iSCSI Cannot configure this license. No system impact.

Activate a licenseTo access a OneFS module, you must activate a license.

Before you begin

Before you can activate a license, you must obtain a valid license key, and you must haveroot user privileges on your cluster. To obtain a license key, contact your EMC Isilon salesrepresentative.Procedure

1. Click Help > About This Cluster.

2. In the Licensed Modules section, click Activate license.

3. In the License key field, type the license key for the module that you want to enable.


License configuration 39

4. Read the end user license agreement, click I have read and agree, and then clickSubmit.

View license informationYou can view information about the current status of any optional Isilon softwaremodules.

Procedure

1. Click Help > About This Cluster.

2. In the Licensed Modules area, review information about licenses, including status andexpiration date.

Unconfigure a licenseYou can unconfigure a licensed module through the command-line interface.

You must have root user privileges on your Isilon cluster to unconfigure a module license.This procedure is available only through the command-line interface (CLI).

Note

Unconfiguring a license does not deactivate the license.

Procedure

1. Open a secure shell (SSH) connection to any node in the cluster.

You must log in as root.

2. Run the isi license unconfigure command.

The following command unconfigures the license for SmartConnect:isi license unconfigure -m smartconnect

If you do not know the module name, run the isi license command for a list ofOneFS modules and their status.

OnesFS returns a confirmation message similar to the following text: TheSmartConnect module has been unconfigured. The license isunconfigured, and any processes enabled for the module are disabled.

CertificatesYou can renew the Secure Sockets Layer (SSL) certificate for the Isilon web administrationinterface or replace it with a third-party SSL certificate.

All Platform API communication, which includes communication through the webadministration interface, is over SSL. You can replace or renew the self-signed certificatewith a certificate that you generate. To replace or renew an SSL certificate, you must belogged on as root.



Replace or renew the SSL certificateYou can replace or renew the Secure Sockets Layer (SSL) certificate, which is used toaccess the EMC Isilon cluster through a browser.

Before you begin

When you renew or replace a self-signed SSL certificate, you must provide information foryour organization in the format that is described in the Self-signed SSL certificate dataexample.

The following folders are the default locations for the server.crt and server.keyfiles in OneFS 6.0 and higher.

u SSL certificate: /usr/local/apache2/conf/ssl.crt/server.crtu SSL certificate key: /usr/local/apache2/conf/ssl.key/server.keyProcedure

1. Establish an SSH connection to any node in the cluster.

2. At the command prompt, run the following command to create the appropriatedirectory.

mkdir /ifs/local/3. At the command prompt, run the following command to change to the directory.

cd /ifs/local/4. Choose the type of certificate you want to install.

Option Description

Third-party(public orprivate) CA-issuedcertificate

a. At the command prompt, run the following command togenerate a new Certificate Signing Request (CSR) in addition toa new key, where <common_name> is the host name, such asisilon.example.com:openssl req -new -nodes -newkey rsa:1024 -keyout <common name>.key \ -out <common-name>.csr

b. Send the contents of the <common_name>.csr file from thecluster to your Certificate Authority (CA) for signing. When youreceive the signed certificate (now a .crt file) from the CA,copy the certificate to /ifs/local/<common-name>.crt.

Self-signedcertificatebased on theexisting (stock)ssl.key

a. At the command prompt, run the following command to createa two-year certificate. Increase or decrease the value for -daysto generate a certificate with a different expiration date.cp /usr/local/apache2/conf/ssl.key/server.key ./openssl req -new \/ -days 730 -nodes -x509 -key server.key -out server.crt

A renewal certificate is created, based on the existing (stock) ssl.key file.

5. Optional: At the command prompt, run the following command to verify the attributesin an SSL certificate.

openssl x509 -text -noout -in <common-name>.crt6. Run the following commands to install the certificate and key:

isi services -a isi_webui disablechmod 640 <common name>.key


Replace or renew the SSL certificate 41

isi_for_array -s 'cp /ifs/local/<common-name>.key /usr/local/apache2/conf/ssl.key/<common-name>.key'isi_for_array -s 'cp /ifs/local/<common-name>.crt /usr/local/apache2/conf/ssl.crt/<common-name>.crt'isi services -a isi_webui enable

7. Run the following command to remove the files in /ifs/local.

rm /ifs/local/*

Verify an SSL certificate updateYou can verify the details stored in a Secure Sockets Layer (SSL) certificate.

Procedure

1. Open a web browser window.

2. Browse to https://<common name>:8080, where <common name> is the hostname for the EMC Isilon web administration interface, such as isilon.example.com.

3. In the security details for the web page, verify that the subject line and other detailsthat you provided are correct.

Note

The steps to view security details vary by browser. For example, in some browsers, youcan click the padlock icon in the address bar to view the security details for the webpage. Follow the steps that are specific to your browser.

Self-signed SSL certificate data exampleSelf-signed SSL certificate renewal or replacement requires you to provide data such asyour fully qualified domain name and a contact email address.

When you renew or replace a self-signed SSL certificate, you are asked to provide data inthe format shown in the following example. Some fields in the certificate file contain adefault value. If you type '.', the field is left blank when the certificate is generated.

u Country Name (2 letter code) [XX]:USu State or Province Name (full name) [Some-State]:Washingtonu Locality Name (for example, city) [default city]:Seattleu Organization Name (for example, company) [Internet Widgits Pty Ltd]:Isilonu Organizational Unit Name (for example, section) []:Supportu Common Name (for example, server FQDN or server name) []:isilon.example.com

u Email Address []:[email protected] addition, you should add the following attributes to be sent with your certificaterequest:

u Challenge password []:Isilon1u Optional company name []:

General cluster settingsGeneral settings that are applied across the entire cluster can be modified.

You can modify the following general settings to customize the Isilon cluster for yourneeds:



u Cluster name

u Cluster date and time, NTP settings

u Character encoding

u Email settings

u SNMP monitoring

u SupportIQ settings

Set the cluster nameYou can assign a name and add a login message to your EMC Isilon cluster to make thecluster and its nodes more easily recognizable on your network.

Cluster names must begin with a letter and can contain only numbers, letters, andhyphens. The cluster name is added to the node number to identify each node in thecluster. For example, the first node in a cluster named Images may be named Images-1.

Procedure

1. Click Cluster Management > General Settings > Cluster Identity.

2. Optional: In the Cluster Name and Description area, type a name for the cluster in theCluster Name field and type a description in the Cluster Description field.

3. Optional: In the Login Message area, type a title in the Message Title field and amessage in the Message Body field.

4. Click Submit.

After you finish

You must add the cluster name to your DNS servers.

Specify contact informationYou can specify contact information so that Isilon Technical Support personnel and eventnotification recipients can contact you.

Procedure

1. Click Dashboard > Events > Notification Settings.

2. In the Contact Information area, click Modify contact information settings.

3. In the Contact Information area, type the name and contact information in the fieldsfor those details.

4. Click Submit.

Configuring the cluster date and timeThe NTP service is configurable manually, so you can ensure that all nodes in a cluster aresynchronized to the same time source.

The Network Time Protocol (NTP) method automatically synchronizes cluster date andtime settings through an NTP server. Alternatively, you can set the date and time reportedby the cluster by manually configuring the service.

Windows domains provide a mechanism to synchronize members of the domain to amaster clock running on the domain controllers, so OneFS adjusts the cluster time to thatof Active Directory with a service. Whenever a cluster is joined to an Active Directorydomain and an external NTP server is not configured, the cluster is set automatically toActive Directory time, which is synchronized by a job that runs every 6 hours. When the


Set the cluster name 43

cluster and domain time become out of sync by more than 4 minutes, OneFS generatesan event notification.

Note

If the cluster and Active Directory become out of sync by more than 5 minutes,authentication will not work.

To summarize:

u If no NTP server is configured but the cluster is joined to an Active Directory domain,the cluster synchronizes with Active Directory every 6 hours.

u If an NTP server is configured, the cluster synchronizes the time with the NTP server.

Set the cluster date and timeYou can set the date, time, and time zone that is used by the EMC Isilon cluster.

Procedure

1. Click Cluster Management > General Settings > Date & Time.

The Date and Time page displays a list of each node's IP address and the date andtime settings for each node.

2. From the Date and time lists, select the month, date, year, hour, and minute settings.

3. From the Time zone list, select a value.

If the time zone that you want is not in the list, select Advanced from the Time zonelist, and then select the time zone from the Advanced time zone list.

4. Click Submit.

Specify an NTP time serverYou can specify one or more Network Time Protocol (NTP) servers to synchronize thesystem time on the EMC cluster. The cluster periodically contacts the NTP servers andsets the date and time based on the information it receives.

Procedure

1. Click Cluster Management > General Settings > NTP.

2. Optional: Add a server.

a. In the Server IP or hostname field, type the host name or IP address of the NTPserver, click Add, and then click Submit.

b. Optional: To enable NTP authentication with a keyfile, type the path and file namein the Keyfile field, and then click Submit.

3. Optional: Delete a server.

a. Select the check box next to the server name in the Server list for each server thatyou want to delete.

b. Click Delete.

c. Click Submit.



Configure SMTP email settingsYou can send event notifications through the SMTP mail server. You can also enableSMTP authentication if your SMTP server is configured to use it.

You can configure SMTP email settings if your network environment requires the use of anSMTP server or if you want to route EMC Isilon cluster event notifications with SMTPthrough a port.

Procedure

1. Click Cluster Management > General Settings > Email Settings.

2. In the Email Settings area, type the SMTP information for your environment in eachfield.

3. Optional: For the Use SMTP AUTH option, select Yes, type the user credentials, andthen select a connection security option.

4. Click Submit.

You can test your configuration by sending a test event notification.

Specify the cluster join modeYou can specify the method to use when nodes are added to the EMC Isilon cluster.

Procedure

1. Click Cluster Management > General Settings > Join Mode.

2. Optional: In the Settings area, select the mode that you want to be used when nodesare added to the cluster.

l Manual— joins can be initiated by either the node or the cluster.

l Secure—joins can be initiated only by the cluster.

3. Click Submit.

Cluster join modesYou can specify the method that you want to use to join nodes to a cluster. The join modedetermines how the system responds when new nodes are added to the subnet occupiedby the Isilon cluster. You can set join modes through the web administration interface orthe command-line interface.

Mode Description Notes- - -Manual Configures OneFS to join

new nodes to the clusterin a separate manualprocess, allowing theaddition of a nodewithout requiringauthorization

N/A

Secure Requires authorizationof every node added tothe cluster

If you use the secure join mode, you cannot use the serial

console wizard option [2] Join an existing cluster to join anode to the cluster. You must add the node from cluster byusing the web administration interface or the isi devices


Configure SMTP email settings 45

Mode Description Notes- - -

-a add -d <unconfigured_node_serial_no>command in the command-line interface.

Enable or disable access time trackingYou can enable access time tracking to support features that require it.

By default, the EMC Isilon cluster does not track the timestamp when files are accessed.You can enable this feature to support OneFS features that use it. For example, access-time tracking must be enabled to configure SyncIQ policy criteria that match files basedon when they were last accessed.

Note

Enabling access-time tracking may affect cluster performance.

Procedure

1. Click File System Management > File System Settings > Access Time Tracking.

2. In the Access Time Tracking area, select a configuration option.

l To enable access time tracking, click Enabled, and then specify in the Precisionfields how often to update the last-accessed time by typing a numeric value and byselecting a unit of measure, such as Seconds, Minutes, Hours, Days, Weeks,Months, or Years.

For example, if you configure a Precision setting of 1 day, the cluster updates thelast-accessed time once each day, even if some files were accessed more oftenthan once during the day.

l To disable access-time tracking, click Disabled.

3. Click Submit.

Specify the cluster character encodingYou can modify the character encoding set for the EMC Isilon cluster after installation.

Only OneFS-supported character sets are available for selection. UTF-8 is the defaultcharacter set for OneFS nodes. You must restart the cluster to apply character encodingchanges.

CAUTION

Character encoding is typically established during installation of the cluster. Modifyingthe character encoding setting after installation may render files unreadable if doneincorrectly. Modify settings only if necessary after consultation with Isilon TechnicalSupport.

Procedure

1. Click File System Management > File System Settings > Character Encoding.

2. Optional: From the Character encoding list, select the character-encoding set that youwant to use.

3. Click Submit, and then click Yes to acknowledge that the encoding change becomeseffective after the cluster is restarted.



4. Restart the cluster.

Results

After the cluster restarts, the web administration interface reflects your change.

Cluster statisticsYou can view performance, historical, and in-depth usage statistics for your EMC Isiloncluster, and control the output for each mode of statistics reporting.

The isi statistics and isi status command-line tools include options forquerying and filtering the display of EMC Isilon cluster performance and usage statistics.

Performance monitoringYou can view cluster throughput either graphically and numerically for average andmaximum usage.

Performance information is monitored through the web administration interface, orthrough the command-line interface by using the isi statistics command options.You can view details about the input and output traffic to and from the cluster's filesystem. You can also monitor throughput distribution across the cluster.

Advanced performance monitoring and analytics are available through the InsightIQmodule, which requires you to activate a separate license. For more information aboutoptional software modules, contact your EMC Isilon Storage Division salesrepresentative.

Cluster monitoringYou can monitor the health, performance, and status of your EMC Isilon cluster.

Information is available for individual nodes, including node-specific network traffic,internal and external network interfaces, and details about node pools, tiers, and overallcluster health. You can monitor the following areas of your EMC Isilon cluster health andperformance:

Node statusHealth and performance statistics for each node in the cluster, including hard diskdrive (HDD) and solid-state drive (SSD) usage.

Client connectionsNumber of clients connected per node.

New eventsList of event notifications generated by system events, including the severity, uniqueinstance ID, start time, alert message, and scope of the event.

Cluster sizeCurrent view: Used and available HDD and SSD space and space reserved for thevirtual hot spare (VHS). Historical view: Total used space and cluster size for a one-year period.


Cluster statistics 47

Cluster throughput (file system)Current view: Average inbound and outbound traffic volume passing through thenodes in the cluster for the past hour. Historical view: Average inbound andoutbound traffic volume passing through the nodes in the cluster for the past twoweeks.

CPU usageCurrent view: Average system, user, and total percentages of CPU usage for the pasthour. Historical view: CPU usage for the past two weeks.

Using the OneFS dashboard, you can monitor the status and health of the OneFS systemhardware. In addition, you can use SNMP to remotely monitor hardware components,such as fans, hardware sensors, power supplies, and disks.

Monitor the clusterYou can monitor the health and performance of an EMC Isilon cluster with charts andtables that show the status and performance of nodes, client connections, events, clustersize, cluster throughput, and CPU usage.

Procedure

1. Click Dashboard > Cluster Overview > Cluster Status.

2. Optional: View cluster details.

l Status: To view details about a node, click the ID number of the node.

l Client connection summary: To view a list of current connections, click Dashboard >Cluster Overview > Client Connections Status.

l New events: To view more information about an event, click View details in theActions column.

l Cluster size: To switch between current and historical views, click Historical orCurrent near the Monitoring section heading. In historical view, click Used orCluster size to change the display.

l Cluster throughput (file system): To switch between current and historical views,click Historical or Current next to the Monitoring section heading. To viewthroughput statistics for a specific period within the past two weeks, clickDashboard > Cluster Overview > Throughput Distribution.

Note

You can hide or show inbound or outbound throughput by clicking Inbound orOutbound in the chart legend. To view maximum throughput, next to Show, selectMaximum.

l CPU usage: To switch between current and historical views, click Historical orCurrent near the Monitoring section heading.

Note

You can hide or show a plot by clicking System, User, or Total in the chart legend.To view maximum usage, next to Show, select Maximum.



View node statusYou can view the current and historical status of a node.

Procedure


2. Optional: In the Status area, click the ID number for the node that you want to viewstatus for.

3. View node details.

l Status: To view networks settings for a node interface or subnet or pool, click thelink in the Status area.

l Client connections: To view current clients connected to this node, review the listin this area.

l Chassis and drive status: To view the state of drives in this node, review this area.To view details about a drive, click the name link of the drive; for example, Bay1.

l Node size: To switch between current and historical views, click Historical orCurrent next to the Monitoring area heading. In historical view, click Used orCluster size to change the display accordingly.

l Node throughput (file system): To switch between current and historical views,click Historical or Current next to the Monitoring area heading. To view throughputstatistics for a period within the past two weeks, click Dashboard > ClusterOverview > Throughput Distribution.

Note

You can hide or show inbound or outbound throughput by clicking Inbound orOutbound in the chart legend. To view maximum throughput, next to Show, selectMaximum.

l CPU usage: To switch between current and historical views, click Historical orCurrent next to the Monitoring area heading.

Note

You can hide or show a plot by clicking System, User, or Total in the chart legend.To view maximum usage, next to Show, select Maximum.

Events and notificationsYou can monitor the health and performance of your EMC Isilon cluster through OneFSevent notifications.

You can select the OneFS hardware, software, network, and system events that you wantto monitor, and you can cancel, quiet, or unquiet events. In addition, you can configureevent notification rules to send an email notification or SNMP trap when a threshold isexceeded.

Event notification methodsYou can configure event notification rules to generate and deliver event notificationswhen an event occurs.

You can notify users by email, SupportIQ, or SNMP trap.


View node status 49

EmailYou can designate recipients and specify SMTP, authorization, and security settings.You can specify batch email settings and the email notification template.

SupportIQYou can specify a protocol that you prefer to use for notifications: HTTPS, SMTP, orboth.

SNMP trapYou can send SNMP traps to one or more network monitoring stations or trapreceivers. Each event can generate one or more SNMP traps. You can downloadmanagement information base files (MIBs) from the cluster at /usr/local/share/snmp/mibs/. The ISILON-TRAP-MIB.txt file describes the traps thatthe cluster can generate, and the ISILON-MIB.txt file describes the associatedvarbinds that accompany the traps.

Note

You must configure an event notification rule to generate SNMP traps.

Event notification settingsYou can specify whether you want to receive event notifications as aggregated batches oras individual notifications for each event. Batch notifications are sent every 10 seconds.

The batch options that are described in this table affect both the content and the subjectline of notification emails that are sent in response to system events. You can specifyevent notification batch options when you configure SMTP email settings.

Setting Option Description- - -Notification batchmode

Batch all Generates a single email for each eventnotification.

Batch by severity Generates an email that containsaggregated notifications for each event ofthe same severity, regardless of eventcategory.

Batch by category Generates an email an email that containsaggregated notifications for event of thesame category, regardless of severity.

No batching Generates one email per event.

Custom notificationtemplate

No custom notificationtemplate is set

Sends the email notification in the defaultOneFS notification template format.

Set custom notificationtemplate

Sends the email notifications in the formatthat you defined in your custom templatefile.



Coalesced eventsRelated or repeated events are grouped, or coalesced, into one event by the OneFSsystem.

There are two types of coalesced events.

Group eventsGroup events are different types of events that are all related to a single problem. Forexample, a single connection problem might generate the following events:

Event Description- -100010005 A SAS PHY topology problem or change was detected.

100010006 A drive's error log counter indicates there may be a problem.

100010007 A SAS link has exceeded the maximum Bit Error Rate (BER) .

100010008 A SAS link has been disabled for exceeding the maximum Bit Error Rate (BER).

Because the events are all related to a single problem, OneFS creates a group eventand adds the related errors to that event. Instead of seeing four events, you will seea single group event alerting you to storage transport problems. You can still view allthe grouped events individually if you choose.This message is representative of group coalesced event output.

# isi events show 24.924 ID: 24.924 Type: 199990001 Severity: critical Value: 0.0 Message: Disk Errors detected (Bay 1) Node: 21 Lifetime: Sun Jun 17 23:29:29 2012 - Now Quieted: Not quieted Specifiers: disk: 35 val: 0.0 devid: 24 drive_serial: 'XXXXXXXXXXXXX' lba: 1953520064L lnn: 21 drive_type: 'HDD' device: 'da1' bay: 1 unit: 805306368Coalesced by: --Coalescer Type: GroupCoalesced events:ID STARTED ENDED SEV LNN MESSAGE24.911 06/17 23:29 -- I 21 Disk stall: Bay 1, Type HDD, LNUM 35. Disk ...24.912 06/17 23:29 -- I 21 Sector error: da1 block 195352006424.913 06/17 23:29 -- I 21 Sector error: da1 block 220223224.914 06/17 23:29 -- I 21 Sector error: da1 block 220212024.915 06/17 23:29 -- I 21 Sector error: da1 block 220210424.916 06/17 23:29 -- I 21 Sector error: da1 block 220261624.917 06/17 23:29 -- I 21 Sector error: da1 block 220216824.918 06/17 23:29 -- I 21 Sector error: da1 block 220210624.919 06/17 23:29 -- I 21 Sector error: da1 block 220210524.920 06/17 23:29 -- I 21 Sector error: da1 block 104867024.921 06/17 23:29 -- I 21 Sector error: da1 block 22324.922 06/17 23:29 -- C 21 Disk Repair Initiated: Bay 1, Type HDD, LNUM...


Events and notifications 51

Repeat or duplicate eventsRepeat or duplicate events are the same type of event repeated in response to anongoing problem. For example, if a CPU fan crosses the speed threshold more thanten times in an hour, the system coalesces this sequence of identical but discreteoccurrences into one event.This message is representative of coalesced repeat event output.

# isi events show 1.3035 ID: 1.3035 Type: 500010001Severity: info Value: 0.0 Message: SmartQuotas threshold violation on quota violated, domain direc... Node: All Lifetime: Thu Jun 14 01:00:00 2012 - Now Quieted: Not quieted Specifiers: enforcement: 'advisory' domain: 'directory /ifs/quotas' name: 'violated' val: 0.0 devid: 0 lnn: 0Coalesced by: --Coalescer Type: DuplicateCoalesced events:ID STARTED ENDED SEV LNN MESSAGE18.621 06/14 01:00 -- I All SmartQuotas threshold violation on quota vio...18.630 06/15 01:00 -- I All SmartQuotas threshold violation on quota vio...18.638 06/16 01:00 -- I All SmartQuotas threshold violation on quota vio...18.647 06/17 01:00 -- I All SmartQuotas threshold violation on quota vio...18.655 06/18 01:00 -- I All SmartQuotas threshold violation on quota vio...

You can view coalesced events and details through the web administration interface orthe command-line interface.

Responding to eventsYou can view event details and respond to cluster events.

You can view and manage new events, open events, and recently ended events. You canalso view coalesced events and additional, more-detailed information about a specificevent. You also can quiet or cancel events.

View event details

You can view the details of an event and you can add a new notification rule or addsettings to another notification rule.

Procedure

1. Click Dashboard > Events > Summary.

2. In the Actions column of an event whose details you want to view, click View details.

3. Optional: To acknowledge the event, click Quiet Event.

4. Optional: To create a new event notification rule or to add the event settings of thisevent to an existing event notification rule, click Create Notification Rule.l To add a new notification rule for this event, in the Create Rule area, select Create

a new notification rule for event, click Submit, and then specify the settings for therule.



l To add the settings of this event to an existing event notification rule, in the CreateRule area, select Add to an existing notification rule, select the existing eventnotification rule from the list, and then click Submit.

View the event history

You can view all events in a chronological list and then select an event to view additionalinformation.

Procedure

1. Click Dashboard > Events > Events History.

The Events History page displays a list of all events in chronological order—newest tooldest—that have occurred on your Isilon cluster.

View the event log

You can log in to a node through the command-line interface and view the contents of thelocal event log.

Event logs are typically used for support purposes. You can only view the event log usingthe command-line interface.

Procedure

1. Establish an SSH connection to any node in the EMC Isilon cluster.

2. View the /var/log/isi_celog_events.log file.

The log file lists all event activity. Each event row contains one of the following eventlabels:

Event label Description- -COALESCED: FIRST EVENT An event was tagged as a possible first event in a series of events

that can be coalesced. The first event label is a only a placeholderfor a potential parent coalescer event.

COALESCER EVENT: ADDED A parent coalescer event was created.

COALESCED An event was added as a child beneath a coalescer event.

CREATOR EV COALID UPDATED A group was created and the placeholder first event label wasupdated to include actual group information.

DROPPED An event did not include any new information and was not storedin the master event database.

FORWARDED_TO_MASTER An event was forwarded to the master node to be stored in themaster event database.

DB: STORED An event was stored in the master event database.

DB: PURGED An event was removed from the master event database. Thedatabase has a limit of 50,000 entries, and old events are purgedwhen that limit is reached.

INVALID EVENT: DROPPED An event contained invalid information and was not stored in themaster event database.

UPDATE EVENT: DROPPED A request to update the group information in a parent coalescerevent was discontinued.



Quieting, unquieting, and canceling events

You can change an event's state by quieting, unquieting, or canceling an event.

You can select the following actions to change the state of an event:

QuietAcknowledges and removes the event from the list of new events and adds the eventto a list of quieted events.

Note

If a new event of the same event type is triggered, it is a separate new event andmust be quieted.

UnquietReturns a quieted event to an unacknowledged state in the list of new events andremoves the event from the list of quieted events.

CancelPermanently ends an occurrence of an event. The system cancels an event whenconditions are met that end its duration, which is bounded by a start time and anend time, or when you cancel the event manually.

Most events are canceled automatically by the system when the event reaches the end ofits duration. The event remains in the system until you manually acknowledge or quietthe event. You can acknowledge events through either the web administration interfaceor the command-line interface.

Manage an event

You can change the status of an event by quieting, unquieting, or canceling it.

Procedure

1. Click Dashboard > Events > Summary.

2. Perform the following actions as needed.

l To view additional information about an event, in the Actions column for thatevent, click View details.

l To acknowledge an event, click Quiet.

l To restore an event to an unacknowledged state, click Unquiet.

l To permanently remove an occurrence of an event, click Cancel.

Managing event notification rulesYou can modify or delete notification rules, configure event notification settings, andconfigure batch notification settings through the web administration interface or thecommand-line interface.

You can specify event notification settings, and you can create, modify, or delete eventnotification rules. You can configure the setting for how notifications are received,individually or in a batch.

Create an event notification rule

You can configure event notification rules based on specified events and event types.

You can configure email notification and SNMP trap generation for a specific event.



Procedure

1. Click Dashboard > Events > Event Notification Rules.

2. In the Notification Rules area on the Cluster Events page, click Add Rule.

3. In the Rule name field on the Add Notification Rule page, type a name for the rule.

4. In the Recipients area, specify a notification method.

a. To notify a recipient through email, select Email, type the email address to whichnotifications will be sent, and then click Add.

b. To notify a community through SNMP traps, select SNMP, select the communityname and the SNMP host, which is the network monitoring station, from therespective lists, and then click Add.

c. To add additional notification recipients or communities, repeat these steps.

5. In the Events area, expand the event types and select the check boxes for the eventsand event types that you want to trigger this notification.

6. Click Submit.

Send a test event notification

You can generate a test event notification to confirm that event notifications are workingas you intend.

Procedure


2. In the Send Test Event area on the Cluster Events page, click Send test event.

3. On the Cluster Events page, click Summary to verify whether the test event wassuccessful.

A corresponding test event notification appears in the New Events list, which appearsin the Message column as a message similar to Test event sent from WebUI.

View event notification rules

You can view a list of event notification rules and details about specific rules.

Procedure


2. In the Actions column of the rule whose settings you want to view, click Edit.

3. When you have finished viewing the rule details, click Cancel.

Modify an event notification rule

You can modify event notification rules that you created. System event notification rulescannot be modified.

Procedure


2. In the Actions column for the rule that you want to modify, click Edit.

3. Modify the event notification rule settings as needed.

4. Click Submit.



Delete an event notification rule

You can delete event notification rules that you created, but system event notificationrules cannot be deleted.

Procedure


2. In the Notification Rules area, in the Actions column for the rule that you want todelete, click Delete.

3. Click Yes to confirm the deletion.

View event notification settings

You can view email, SupportIQ, and contact information for event notifications.

Procedure


Modify event notification settings

You can modify email, SupportIQ, and contact settings for event notifications.

Procedure


2. Click the Modify link for the setting that you want to change.

3. Click Submit.

Specify event-notification batch mode or template settings

You can choose an event-notification batch option to specify whether you want to receivenotifications individually or as an aggregate. You also can specify a custom notificationtemplate for email notifications.

Before you begin

You must first create a custom notification template and then upload it to a directory atthe same level or below /ifs; for example, /ifs/templates.

Procedure

1. Click Cluster Management > General Settings > Email Settings.

2. In the Event Notification Settings area on the General Settings page, select aNotification batch mode option.

3. Leave the Set custom notification template field blank to use the default notificationtemplate.

4. In the Custom notification template field, select the custom event notificationtemplate.

l Click Browse, navigate to and select the template file that you want to use, andthen click OK.

l In the Set custom notification template field, type the path and file name of thetemplate file that you want to use.

5. Click Submit.



Monitoring cluster hardwareThe default Linux SNMP tools or a GUI-based SNMP tool of your choice can be used tomonitor cluster hardware.

You can enable SNMP on all OneFS nodes to remotely monitor the hardware componentsacross the cluster, including fans, hardware sensors, power supplies, and disks.

You can run the isi batterystatuscommand to monitor the status of NVRAMbatteries and charging systems. This functionality is available only from the command-line on node hardware that supports the command.

To maintain optimal cluster health, you can enable and configure SupportIQ to forward allcluster events to Isilon Technical Support for analysis and resolution.

View node hardware statusYou can view the hardware status of a node.

Procedure


2. Optional: In the Status area, click the ID number for a node.

3. In the Chassis and drive status area, click Platform.

Chassis and drive statesYou can view chassis and drive state details.

In a cluster, the combination of nodes in different degraded states determines whetherread requests, write requests, or both work. A cluster can lose write quorum but keepread quorum. OneFS provides details about the status of chassis and drives in yourcluster. The following table describes all the possible states that you may encounter inyour cluster.

State Description Interface Errorstate

- - - -HEALTHY All drives in the node are functioning correctly. CLI, web

administrationinterface

SMARTFAIL or

Smartfail orrestripe inprogress

The drive is in the process of being removedsafely from the file system, either because ofan I/O error or by user request. Nodes or drivesin a smartfail or read-only state affect onlywrite quorum.

CLI, webadministrationinterface

NOT AVAILABLE A drive can be unavailable for a variety ofreasons. You can click the bay to view detailedinformation about this condition.

Note

In the web administration interface, this state

includes the ERASE and SED_ERRORcommand-line interface states.


X


Monitoring cluster hardware 57


- - - -SUSPENDED This state indicates that drive activity is

temporarily suspended and the drive is not inuse. The state is manually initiated and doesnot occur during normal cluster activity.


NOT IN USE A node in an offline state affects both read andwrite quorum.


REPLACE The drive was smartfailed successfully and isready to be replaced.

CLI only

STALLED The drive is stalled and undergoing stallevaluation. Stall evaluation is the process ofchecking drives that are slow or having otherissues. Depending on the outcome of theevaluation, the drive may return to service orbe smartfailed. This is a transient state.

CLI only

NEW The drive is new and blank. This is the statethat a drive is in when you run the isi dev -a add command.

CLI only

USED The drive was added and contained an IsilonGUID but the drive is not from this node. Thisdrive likely will be formatted into the cluster.

CLI only

PREPARING The drive is undergoing a format operation.The drive state changes to HEALTHY when theformat is successful.

CLI only

EMPTY No drive is in this bay. CLI only

WRONG_TYPE The drive type is wrong for this node. Forexample, a non-SED drive in a SED node, SASinstead of the expected SATA drive type.

CLI only

BOOT_DRIVE Unique to the A100 drive, which has bootdrives in its bays.

CLI only

SED_ERROR The drive cannot be acknowledged by theOneFS system.

Note


is included in Not available.

CLI , webadministrationinterface

X

ERASE The drive is ready for removal but needs yourattention because the data has not beenerased. You can erase the drive manually toguarantee that data is removed.

Note



CLI only




- - - -INSECURE Data on the self-encrypted drive is accessible

by unauthorized personnel. Self-encryptingdrives should never be used for non-encrypteddata purposes.

Note


is labeled Unencrypted SED.

CLI only X

UNENCRYPTEDSED

Data on the self-encrypted drive is accessibleby unauthorized personnel. Self-encryptingdrives should never be used for non-encrypteddata purposes.

Note

In the command-line interface, this state is

labeled INSECURE.

Webadministrationinterface only

X

SNMP monitoringYou can use SNMP to remotely monitor the EMC Isilon cluster hardware components,such as fans, hardware sensors, power supplies, and disks. The default Linux SNMP toolsor a GUI-based SNMP tool of your choice can be used for this purpose.

You can enable SNMP monitoring on individual nodes on your cluster, and you can alsomonitor cluster information from any node. Generated SNMP traps are sent to your SNMPnetwork. You can configure an event notification rule that specifies the network stationwhere you want to send SNMP traps for specific events, so that when an event occurs, thecluster sends the trap to that server. OneFS supports SNMP in read-only mode. SNMP v1and v2c is the default value, but you can configure settings for SNMP v3 alone or SNMPv1, v2c, and v3.

Note

When SNMP v3 is used, OneFS requires the SNMP-specific security level of AuthNoPriv asthe default value when querying the cluster. The security level AuthPriv is not supported.

Elements in an SNMP hierarchy are arranged in a tree structure, similar to a directory tree.As with directories, identifiers move from general to specific as the string progresses fromleft to right. Unlike a file hierarchy, however, each element is not only named, but alsonumbered.

For example, the SNMPentity .iso.org.dod.internet.private.enterprises.isilon.oneFSss.ssLocalNodeId.0 maps to .1.3.6.1.4.1.12124.3.2.0. The part of the name thatrefers to the OneFS SNMP namespace is the 12124 element. Anything further to the rightof that number is related to OneFS-specific monitoring.

Management Information Base (MIB) documents define human-readable names formanaged objects and specify their data types and other properties. You can downloadMIBs that are created for SNMP-monitoring of an Isilon cluster from the web-administration interface or manage them using the command-line interface. MIBs are


SNMP monitoring 59

stored in /usr/local/share/snmp/mibs/ on a OneFS node. The OneFS ISILON-MIBs serve two purposes:

u Augment the information available in standard MIBs

u Provide OneFS-specific information that is unavailable in standard MIBs

ISILON-MIB is a registered enterprise MIB. Isilon clusters have two separate MIBs:

ISILON-MIBDefines a group of SNMP agents that respond to queries from a network monitoringsystem (NMS) called OneFS Statistics Snapshot agents. As the name implies, theseagents snapshot the state of the OneFS file system at the time that it receives arequest and reports this information back to the NMS.

ISILON-TRAP-MIBGenerates SNMP traps to send to an SNMP monitoring station when thecircumstances occur that are defined in the trap protocol data units (PDUs).

The OneFS MIB files map the OneFS-specific object IDs with descriptions. Download orcopy MIB files to a directory where your SNMP tool can find them, such as /usr/share/snmp/mibs/ or /usr/local/share/snmp/mibs, depending on the tool that youuse.

To have Net-SNMP tools read the MIBs to provide automatic name-to-OID mapping, add -m All to the command, as in the following example.

snmpwalk -v2c -c public -m All <node IP> isilon

If the MIB files are not in the default Net-SNMP MIB directory, you may need to specify thefull path, as in the following example. Note that all three lines are one command.

snmpwalk -m /usr/local/share/snmp/mibs/ISILON-MIB.txt:/usr/local/share/snmp/mibs/ISILON-TRAP-MIB.txt:/usr/local/share/snmp/mibs/ONEFS-TRAP-MIB.txt \ -v2c -C c -c public <node IP> enterprises.onefs

Note

The examples are from running the snmpwalk command on a cluster. Your SNMP versionmay require different arguments.

Managing SNMP settingsSNMP can be used to monitor cluster hardware and system information. Settings can beconfigured through either the web administration interface or the command-lineinterface.

You can enable SNMP monitoring on individual nodes in the cluster, and you can monitorinformation cluster-wide from any node when you enable SNMP on each node. Whenusing SNMP on an Isilon cluster, you should use a fixed general username. A passwordfor the general user can be configured in the web administration interface.

You should configure a network monitoring system (NMS) to query each node directlythrough a static IP address. This approach allows you to confirm that all nodes haveexternal IP addresses and therefore respond to SNMP queries. Because the SNMP proxyis enabled by default, the SNMP implementation on each node is configuredautomatically to proxy for all other nodes in the cluster except itself. This proxyconfiguration allows the Isilon Management Information Base (MIB) and standard MIBs tobe exposed seamlessly through the use of context strings for supported SNMP versions.



After you download and save the appropriate MIBs, you can configure SNMP monitoringthrough either the web administration interface or though the command-line interface.

Configure the cluster for SNMP monitoringYou can configure your EMC Isilon cluster to remotely monitor hardware componentsusing SNMP.

Before you begin

When SNMP v3 is used, OneFS requires the SNMP-specific security level of AuthNoPriv asthe default value when querying the cluster. The security level AuthPriv is not supported.

You can enable or disable SNMP monitoring, allow SNMP access by version, andconfigure other settings, some of which are optional. All SNMP access is read-only.

Note

The Isilon cluster does not generate SNMP traps unless you configure an eventnotification rule to send events.

Procedure

1. Click Cluster Management > General Settings > SNMP Monitoring.

2. In the Service area of the SNMP Monitoring page, enable or disable SNMP monitoring.

a. To disable SNMP monitoring, click Disable, and then click Submit.

b. To enable SNMP monitoring, click Enable, and then continue with the followingsteps to configure your settings.

3. In the Downloads area, click Download for the MIB file that you want to download.

Follow the download process that is specific to your browser.

4. Optional: If you are using Internet Explorer as your browser, right-click the Downloadlink, select Save As from the menu, and save the file to your local drive.

You can save the text in the file format that is specific to your Net-SNMP tool.

5. Copy MIB files to a directory where your SNMP tool can find them, such as /usr/share/snmp/mibs/ or /usr/local/share/snmp/mibs, depending on theSNMP tool that you use.

To have Net-SNMP tools read the MIBs to provide automatic name-to-OID mapping,add -m All to the command, as in the following example: snmpwalk -v2c -cpublic -m All <node IP> isilon

6. Navigate back to the SNMP Monitoring page and configure General Settings.

a. In the Settings area, configure protocol access by selecting the version that youwant.

OneFS does not support writable OIDs; therefore, no write-only community stringsetting is available.

b. In the System location field, type the system name.

This setting is the value that the node reports when responding to queries. Type aname that helps to identify the location of the node.

c. Type the contact email address in the System contact field.

7. Optional: If you selected SNMP v1/v2 as your protocol, locate the SNMP v1/v2cSettings section and type the community name in the Read-only community field.


SNMP monitoring 61

8. Configure SNMP v3 Settings.

a. In the Read-only user field, type the SNMP v3 security name to change the name ofthe user with read-only privileges.

The default read-only user is general.

The password must contain at least eight characters and no spaces.

b. in the SNMP v3 password field, type the new password for the read-only user toset a new SNMP v3 authentication password.

The default password is password.

c. Type the new password in the Confirm password field to confirm the newpassword.

9. Click Submit.

View SNMP settingsYou can review SNMP monitoring settings.

Procedure

1. Click Cluster Management > General Settings > SNMP Monitoring.

Cluster maintenanceTrained service personnel can replace or upgrade components in Isilon nodes.

Isilon Technical Support can assist you with replacing node components or upgradingcomponents to increase performance.

Replacing node componentsIf a node component fails, Isilon Technical Support will work with you to quickly replacethe component and return the node to a healthy status.

Trained service personnel can replace the following field replaceable units (FRUs):

u battery

u boot flash drive

u SATA/SAS Drive

u memory (DIMM)

u fan

u front panel

u intrusion switch

u network interface card (NIC)

u IB/NVRAM card

u SAS controller

u NVRAM battery

u power supply

If you configure your cluster to send alerts to Isilon, Isilon Technical Support will contactyou if a component needs to be replaced. If you do not configure your cluster to sendalerts to Isilon, you must initiate a service request.



Upgrading node componentsYou can upgrade node components to gain additional capacity or performance.

Trained service personnel can upgrade the following components in the field:

u drive

u memory (DIMM)

u network interface card (NIC)

If you want to upgrade components in your nodes, contact Isilon Technical Support.

Managing cluster nodesYou can add and remove nodes from a cluster. You can also shut down or restart theentire cluster.

Add a node to a clusterYou can add a new node to an existing EMC Isilon cluster.

Before you begin

Before you add a node to a cluster, verify that an internal IP address is available. Add IPaddresses as necessary before you add a new node.

If a new node is running a different version of OneFS than a cluster, the system changesthe node version of OneFS to match the cluster.

Procedure

1. Click Cluster Management > Hardware Configuration > Add Nodes.

2. In the Available Nodes table, click Add for the node that you want to add to thecluster.

Remove a node from the clusterYou can remove a node from an EMC Isilon cluster. When you remove a node, the systemsmartfails the node to ensure that data on the node is transferred to other nodes in thecluster.

Removing a storage node from a cluster deletes the data from that node. Before thesystem deletes the data, the FlexProtect job safely redistributes data across the nodesremaining in the cluster.

Procedure

1. Navigate to Cluster Management > Hardware Configuration > Remove Nodes.

2. In the Remove Node area, specify the node you want to remove.

3. Click Submit.

If you remove a storage node, the Cluster Status area displays smartfail progress. Ifyou remove a non-storage accelerator node, it is immediately removed from thecluster.


Upgrading node components 63

Shut down or restart a clusterYou can shut down or restart an entire EMC Isilon cluster.

Procedure

1. Navigate to Cluster Management > Hardware Configuration > Shutdown & RebootControls.

2. In the Shut Down or Reboot This Cluster area, specify an action:

Options Description

Shut down Shuts down the cluster.

Reboot Stops then restarts the cluster.

3. Click Submit.

Remote support using SupportIQIsilon Technical Support personnel can remotely manage your Isilon cluster totroubleshoot an open support case with your permission. The Isilon SupportIQ moduleallows Isilon Technical Support personnel to gather diagnostic data about the cluster.

Isilon Technical Support representatives run scripts that gather data about clustersettings and operations. The SupportIQ agent then uploads the information to a secureIsilon FTP site so it is available for Isilon Technical Support personnel to review. Thesescripts do not affect cluster services or data availability.

Note

The SupportIQ scripts are based on the Isilon isi_gather_info log-gathering tool.

The SupportIQ module is included with the OneFS operating system and does not requireyou to activate a separate license. You must enable and configure the SupportIQ modulebefore SupportIQ can run scripts to gather data. The feature may have been enabledwhen the cluster was first set up, but you can enable or disable SupportIQ through theIsilon web administration interface.

In addition to enabling the SupportIQ module to allow the SupportIQ agent to run scripts,you can enable remote access, which allows Isilon Technical Support personnel tomonitor cluster events and remotely manage your cluster using SSH or the webadministration interface. Remote access helps Isilon Technical Support to quickly identifyand troubleshoot cluster issues. Other diagnostic tools are available for you to use inconjunction with Isilon Technical Support to gather and upload information such aspacket capture metrics.

Note

If you enable remote access, you must also share cluster login credentials with IsilonTechnical Support personnel. Isilon Technical Support personnel remotely access yourcluster only in the context of an open support case and only after receiving yourpermission.



Configuring SupportIQOneFS logs contain data that Isilon Technical Support personnel can securely upload,with your permission, and then analyze to troubleshoot cluster problems. The SupportIQtechnology must be enabled and configured for this process.

When SupportIQ is enabled, Isilon Technical Support personnel can request logs throughscripts that gather cluster data and then upload the data to a secure location. You mustenable and configure the SupportIQ module before SupportIQ can run scripts to gatherdata. The feature may have been enabled when the cluster was first set up.

You can also enable remote access, which allows Isilon Technical Support personnel totroubleshoot your cluster remotely and run additional data-gathering scripts. Remoteaccess is disabled by default. To enable remote SSH access to your cluster, you mustprovide the cluster password to a Technical Support engineer.

Enable and configure SupportIQYou can enable and configure SupportIQ to allow the SupportIQ agent to run scripts thatgather and upload information about your cluster to Isilon Technical Support personnel.Optionally, you can enable remote access to your EMC Isilon cluster.

Procedure

1. Click Cluster Management > General Settings > SupportIQ.

2. In the SupportIQ Settings area, select the Enable SupportIQ check box.

3. For SupportIQ alerts, select an option.

l Send alerts via SupportIQ agent (HTTPS) and by email (SMTP) – SupportIQdelivers notifications to Isilon through the SupportIQ agent over HTTPS and byemail over SMTP.

l Send alerts via SupportIQ agent (HTTPS) – SupportIQ delivers notifications toIsilon only through the SupportIQ agent over HTTPS.

4. Optional: Enable HTTPS proxy support for SupportIQ.

a. Select the HTTPS proxy for SupportIQ check box.

b. In the Proxy host field, type the IP address or fully qualified domain name (FQDN)of the HTTP proxy server.

c. In the Proxy port field, type the number of the port on which the HTTP proxy serverreceives requests.

d. Optional: In the Username field, type the user name for the proxy server.

e. Optional: In the Password field, type the password for the proxy server.

5. Optional: Enable remote access to the cluster.

a. Select the Enable remote access to cluster via SSH and web interface check box.

b. Review the remote-access end user license agreement (EULA) and, if you agree tothe terms and conditions, select the I have read and agree to check box.

6. Click Submit.

A successful configuration is indicated by a message similar to SupportIQsettings have been updated.


Configuring SupportIQ 65

Disable SupportIQYou can disable SupportIQ so the SupportIQ agent does not run scripts to gather andupload data about your EMC Isilon cluster.

Procedure

1. Click Cluster Management > General Settings > SupportIQ.

2. Clear the Enable SupportIQ check box.

3. Click Submit.

SupportIQ scriptsWhen SupportIQ is enabled, Isilon Technical Support personnel can request logs withscripts that gather cluster data and then upload the data. The SupportIQ scripts arelocated in the /usr/local/SupportIQ/Scripts/ directory on each node.

Data-gathering scripts

The following table lists the data-gathering activities that SupportIQ scripts perform.These scripts can be run automatically, at the request of an Isilon Technical Supportrepresentative, to collect information about your cluster's configuration settings andoperations. The SupportIQ agent then uploads the information to a secure Isilon FTP site,so that it is available for Isilon Technical Support personnel to analyze. The SupportIQscripts do not affect cluster services or the availability of your data.

Action Description- -Clean watch folder Clears the contents of /var/crash.

Get application data Collects and uploads information about OneFS applicationprograms.

Generate dashboard filedaily

Generates daily dashboard information.

Generate dashboard filesequence

Generates dashboard information in the sequence that it occurred.

Get ABR data (as builtrecord)

Collects as-built information about hardware.

Get ATA control and GMirrorstatus

Collects system output and invokes a script when it receives anevent that corresponds to a predetermined eventid.

Get cluster data Collects and uploads information about overall clusterconfiguration and operations.

Get cluster events Gets the output of existing critical events and uploads theinformation.

Get cluster status Collects and uploads cluster status details.

Get contact info Extracts contact information and uploads a text file that containsit.

Get contents (var/crash) Uploads the contents of /var/crash.



Action Description- -Get job status Collects and uploads details on a job that is being monitored.

Get domain data Collects and uploads information about the cluster’s ActiveDirectory Services (ADS) domain membership.

Get file system data Collects and uploads information about the state and health of theOneFS /ifs/ file system.

Get IB data Collects and uploads information about the configuration andoperation of the InfiniBand back-end network.

Get logs data Collects and uploads only the most recent cluster log information.

Get messages Collects and uploads active /var/log/messages files.

Get network data Collects and uploads information about cluster-wide and node-specific network configuration settings and operations.

Get NFS clients Runs a command to check if nodes are being used as NFS clients.

Get node data Collects and uploads node-specific configuration, status, andoperational information.

Get protocol data Collects and uploads network status information and configurationsettings for the NFS, SMB, FTP, and HTTP protocols.

Get Pcap client stats Collects and uploads client statistics.

Get readonly status Warns if the chassis is open and uploads a text file of the eventinformation.

Get usage data Collects and uploads current and historical information aboutnode performance and resource usage.

isi_gather_info Collects and uploads all recent cluster log information.

isi_gather_info --incremental

Collects and uploads changes to cluster log information that haveoccurred since the most recent full operation.

isi_gather_info --incremental singlenode

Collects and uploads details for a single node.

Prompts you for the node number.

isi_gather_infosingle node

Collects and uploads changes to cluster log information that haveoccurred since the most recent full operation.

Prompts you for the node number.

Upload the dashboard file Uploads dashboard information to the secure Isilon TechnicalSupport FTP site.


SupportIQ scripts 67

Upgrading OneFSTwo options are available for upgrading the OneFS operating system: a rolling upgrade ora simultaneous upgrade. Before upgrading OneFS softare, a pre-upgrade check must beperformed.

A rolling upgrade individually upgrades and restarts each node in the clustersequentially. During a rolling upgrade, the cluster remains online and continues servingclients with no interruption in service, although some connection resets may occur onSMB clients. Rolling upgrades are performed sequentially by node number, so a rollingupgrade takes longer to complete than a simultaneous upgrade. The final node in theupgrade process is the node that you used to start the upgrade process.

Note

Rolling upgrades are not available for all clusters. For instructions on how to upgrade thecluster operating system, see the OneFS Release Notes.

A simultaneous upgrade installs the new operating system and restarts all nodes in thecluster at the same time. Simultaneous upgrades are faster than rolling upgrades butrequire a temporary interruption of service during the upgrade process. Your data isinaccessible during the time that it takes to complete the upgrade process.

Before beginning either a simultaneous or rolling upgrade, OneFS compares the currentcluster and operating system with the new version to ensure that the cluster meetscertain criteria, such as configuration compatibility (SMB, LDAP, SmartPools), diskavailability, and the absence of critical cluster events. If upgrading puts the cluster atrisk, OneFS warns you, provides information about the risks, and prompts you to confirmwhether to continue the upgrade.

If the cluster does not meet the pre-upgrade criteria, the upgrade does not proceed, andthe unsupported statuses are listed.



CHAPTER 4

Access zones


u Access zones overview..........................................................................................70u Access zone features.............................................................................................70u Managing access zones........................................................................................ 71

Access zones 69

Access zones overviewYou can use access zones to partition cluster configuration into self-contained units.Access zones provide the means for administrators to configure a subset of parametersas a virtual cluster.

OneFS includes a built-in access zone named System that contains all configuredauthentication providers, all available SMB shares, and all available NFS exports. Youcan add additional access zones to the cluster as needed. By default, all cluster IPaddresses connect to the System zone. The System zone is used by administrators formanaging all access zone configuration tasks.

Access zones contain all of the necessary configuration settings to supportauthentication and identity management services in OneFS. NFS exports are added to theSystem zone, but you can configure authentication providers and SMB shares on a zone-by-zone basis. You also can use access zones to partition access to data through IP-address pools. When you configure access zones in this manner, incoming connectionscan be directed to an access zone by a specific IP address within a pool. his restrictsauthentication to that zone and reduces the available SMB shares that can be accessed.

Note

Role-based access, which primarily handles configuration actions, is available throughonly the System zone.

Access zone featuresYou can configure access zones to leverage the following features for your environment.

Feature Description Comment- - -Overlappingshare-namesupport

If multiple SMB share have thesame display name, OneFSsupports the overlapping displaynames if the name appears onlyonce per access zone.

For example, you can assign the name"home" as the display name for a sharein zone A and a different share in zoneB.

Multiple accesszone support

You can create additional accesszones and configure each zonedifferently. Each access zone canbe configured with its own set ofauthentication providers, usermapping rules, and SMB shares.

Multiple access zones are particularlyuseful for server consolidation, forexample when merging multipleWindows file servers that arepotentially joined to different untrustedforests.

Note

NFS users can be authenticated againstonly the System zone.

SMB-protocolaccess auditingon individualaccess zones

You can audit SMB-protocol accesson individual access zones.

For audited zones, you can modify thedefault list of successful and failedprotocol events that are audited.

Access zones


Managing access zonesYou can configure an access zone's settings and add or remove authentication providers,SMB shares, and user mapping rules. If protocol auditing is configured on an accesszone, you can modify the default list of audited events.

To use an access zone, it must be mapped to an IP address pool on the cluster. You candelete any access zone except the built-in System zone.

Create an access zoneWhen you create an access zone, you can add one or more authentication providerinstances, user mapping rules, and SMB shares. You can also create an empty accesszone and configure it later. For more information about settings options, see Access zonesettings.

If you create additional access zones, use them only for data access, and use the Systemzone only for configuration access. Before users can connect to a new access zone, youmust configure your network settings to map an IP address pool to the zone. For moreinformation about settings options, see Access zone settings.

Procedure

1. Click Cluster Management > Access Management > Access Zones.

2. Click Create an access zone.

3. In the Access Zone Name field, type a name for the access zone.

4. Optional: From the Authentication Providers list, select one of the options.

5. Optional: In the User Mapping Rules area, follow these steps for each user mappingrule that you want to add:

a. Click Create a user mapping rule.

The Create a User Mapping Rule form displays.

b. From the Operation list, select one of the operations.

c. Fill in the fields as needed.

Note

Available fields differ depending on the selected operation.

d. Click Add Rule.

Note

Rules are called in the order they are listed. To ensure that each rule getsprocessed, list replacements first and allow/deny rules last. You can change theorder in which a rule is listed by clicking its title bar and dragging it to a newposition.

6. Optional: From the SMB Shares list, select one of the following options:

7. Click Create Access Zone.

After you finish

Before you can use an access zone, you must associate it with an IP address pool.

Access zones

Managing access zones 71

Access zone settingsYou can select from these options when you create an access zone.

Setting Description- -Use allauthenticationproviders

Adds an instance of each available provider to the access zone.

Manually selectauthenticationproviders

Allows you to select one or more provider instances to add to theaccess zone. Follow these steps for each provider instance that youwant to add:

1. Click Add an authentication provider.

2. In the Authentication Provider Type list, select a provider type.A provider type is listed only if an instance of that type exists andis not already in use by the access zone.

3. In the Authentication Provider list, select an available providerinstance.

4. If you are finished adding provider instances, you can change thepriority in which they are called by changing the order in whichthey are listed. To do so, click the title bar of a provider instanceand drag it up or down to a new position in the list.

Append fields froma user

Modifies a token by adding specified fields to it. All appendedidentifiers become members of the additional groups list.

Insert fields froma user

Modifies a token by adding specified fields from another token. Aninserted primary user or group becomes the new primary user or groupin the token and moves the old primary user or group to the additionalidentifiers list. Modifying the primary user leaves the token’susername unchanged. When inserting additional groups from a token,the new groups are added to the existing groups.

Replace a userwith a new user

Replaces a token with the token identified by another user. If anotheruser is not specified, the token is removed from the list and no user isinserted to replace it. If there are no tokens in the list, access isdenied with a "no such user" error.

Removesupplementalgroups from a user

Modifies a token by removing the supplemental groups.

Join two userstogether

Inserts the new token into the list of tokens. If the new token is thesecond user, it is inserted after the existing token; otherwise, it isinserted before the existing token. The insertion point is primarilyrelevant when the existing token is already the first in the list becausethe first token is used to determine the ownership of new systemobjects.

Use no SMB shares Ignores all SMB shares.

Use all SMB shares Adds each available SMB share to the access zone.

Manually selectSMB shares

Allows you to select the SMB shares to add to the access zone. Thefollowing steps are required:

Access zones


Setting Description- -

1. Click Add SMB shares.

2. In the Select SMB Shares dialog box, select the check box foreach SMB share that you want to add to the access zone.

3. To modify the share name that displays when a user connects to

the access zone, click Edit and then, in the Display Name field,type a new name.

Note

You can assign the same name—for example, home, to shares

that are in different access zones.

4. Click Select.

Associate an IP address pool with an access zoneYou can specify which access zone to use according to the IP address that a user isconnecting to.

Procedure

1. Click Cluster Management -> Network Configuration.

2. In the External Network Settings section, under Subnets, click a subnet name (forexample, subnet0).

3. In the IP Address Pools section, click the + icon if necessary to view the settings for apool.

4. Next to the Basic Settings heading, click Edit.

The Configure IP Pool dialog box appears.

5. For the Access zone setting, select the zone to use when connecting through an IPaddress that belongs to this pool.

6. Click Submit.

Modify an access zoneYou can modify the properties of any access zone with one exception: You cannot changethe name of the built-in System zone.

Procedure


2. For the access zone whose settings you want to modify, click View details.

3. For each setting that you want to modify, click Edit, make the change, and then clickSave.

Access zones

Associate an IP address pool with an access zone 73

Delete an access zoneYou can delete any access zone except the built-in System zone. If you delete an accesszone, all associated authentication providers and SMB shares remain available to otherzones.

Procedure


2. Click Delete for the zone that you want to delete.

3. In the confirmation dialog box, click Delete.

Access zones


CHAPTER 5

Authentication and access control


u Authentication and access control overview.......................................................... 76u Data access control............................................................................................... 77u Roles.....................................................................................................................79u Authentication...................................................................................................... 88u Managing access permissions...............................................................................91u Managing roles..................................................................................................... 99u Managing authentication providers..................................................................... 100

Authentication and access control 75

Authentication and access control overviewOneFS supports several methods for ensuring that your cluster remains secure, includingUNIX- and Windows-style permissions for data-level access control. Access zones androle-based administration control access to system configuration settings.

OneFS is designed for a mixed environment that allows you to configure both WindowsAccess Control Lists (ACLs) and standard UNIX permissions on the cluster file system.Windows and UNIX permissions cannot coexist on a single file or directory. However,OneFS uses identity mapping between Windows and UNIX permissions.

Note

In most situations, the default settings are sufficient. You can configure additional accesszones, custom roles, and permissions policies as necessary for your particularenvironment.

Authentication and access control featuresYou can configure settings for the following features for authentication and accesscontrol.

Feature Description Comment- - -Access zones OneFS includes a built-in access zone

named System.By default, new authenticationproviders, SMB shares, and NFSexports are added to the Systemzone. When you create a new IPaddress pool, you must assign it toan access zone.

Authentication Unique user accounts can be a localuser account or user accounts from anActive Directory, LDAP, or NIS.

You can configure access to eachuser account type.

Roles With roles, you can assign privileges tousers and groups. By default, only the"root" and "admin" users can log in tothe command-line interface (CLI)through SSH or the web administrationinterface through HTTP. The root oradmin user can add other users tobuilt-in or custom roles that containthe privileges that are required to login and perform administrativefunctions.

It is good practice to assign users toroles that contain the minimum set ofprivileges that are necessary. Tocreate or assign roles, you must belogged on as a member of theSecurity Administrator role.

Identitymanagement

Identity management enables user-identity integration to provide identicalpermissions to system resources forUnix and Windows users.

All directory services identities for auser can be combined and managedto control access through thesupported protocols to directoriesand files across the cluster.

Mixed-environmentsupport

OneFS is designed for a mixedenvironment, so you can configureboth Windows Access Control Lists

Although Windows and UNIXpermissions cannot coexist on asingle file or directory, OneFS uses



Feature Description Comment- - -

(ACLs) and standard UNIX permissionson the cluster file system.

identity mapping to translatebetween Windows and UNIXpermissions as needed.

Data access controlOneFS supports two types of authorization data on a file: Windows-style access controllists (ACLs) and POSIX mode bits (UNIX permissions). The type of authorization that isused is based on the ACL policies that are set and on the file-creation method.

Access to a file or directory can be governed by either a Windows access control list (ACL)or UNIX mode bits. Regardless of the security model, OneFS enforces access rightsconsistently across access protocols. A user is granted or denied the same rights to a filewhen using SMB for Windows file sharing as when using NFS for UNIX file sharing.

An EMC Isilon cluster includes global policy settings that enable you to customize thedefault ACL and UNIX permissions to best support your environment. Generally, files thatare created over SMB or in a directory that has an ACL receive an ACL; otherwise, OneFSrelies on the POSIX mode bits that define UNIX permissions. In either case, the owner canbe represented by a UNIX identifier (UID or GID) or by its Windows identifier (SID). Theprimary group can be represented by a GID or SID. Although mode bits are present whena file has an ACL, the mode bits are provided only for protocol compatibility and are notused for access checks.

Note

Although you can configure ACL policies to optimize a cluster for UNIX or Windows, youshould do so only if you understand how ACL and UNIX permissions interact.

The OneFS file system installs with UNIX permissions as the default. By using WindowsExplorer or OneFS administrative tools, you can give a file or directory an ACL. In additionto Windows domain users and groups, ACLs in OneFS can include local, NIS, and LDAPusers and groups. After you give a file an ACL, OneFS stops enforcing the file's mode bits,which remain only as an estimate of the effective permissions.

ACLsIn Windows environments, file and directory permissions, referred to as access rights, aredefined in access control lists (ACLs). Although ACLs are more complex than mode bits,ACLs can express much more granular sets of access rules. OneFS uses the ACLprocessing rules commonly associated with Windows ACLs.

A Windows ACL contains zero or more access control entries (ACEs), each of whichrepresents the security identifier (SID) of a user or a group as a trustee. In OneFS, an ACLcan contain ACEs with a UID, GID, or SID as the trustee. Each ACE contains a set of rightsthat allow or deny access to a file or folder. An ACE can optionally contain an inheritanceflag to specify whether the ACE should be inherited by child folders and files.

Note

Instead of the standard three permissions available for mode bits, ACLs have 32 bits offine-grained access rights. Of these, the upper 16 bits are general and apply to all objecttypes. The lower 16 bits vary between files and directories but are defined in a way thatallows most applications to use the same bits for files and directories.


Data access control 77

Rights can be used for granting or denying access for a given trustee. A user's access canbe blocked explicitly through a deny ACE. Access can also be blocked implicitly byensuring that the user does not directly (or indirectly through a group) appear in an ACEthat grants the right in question.

UNIX permissionsIn a UNIX environment, file and directory access is controlled by POSIX mode bits, whichgrant read, write, or execute permissions to the owning user, the owning group, andeveryone else.

OneFS supports the standard UNIX tools for viewing and changing permissions, ls,chmod, and chown. For more information, run the man ls, man chmod, and manchown commands.

All files contain 16 permission bits, which provide information about the file or directorytype and the permissions. The lower 9 bits are grouped as three 3-bit sets, called triples,which contain the read, write, and execute (rwx) permissions for each class of users—owner, group, and other. You can set permissions flags to grant permissions to each ofthese classes.

Unless the user is root, OneFS uses the class to determine whether to grant or denyaccess to the file. The classes are not cumulative; the first class matched is used. It istherefore common to grant permissions in decreasing order.

Mixed-permission environmentsWhen a file operation requests an object’s authorization data (for example, with the ls-l command over NFS or with the Security tab of the Properties dialog box in WindowsExplorer over SMB), OneFS attempts to provide that data in the requested format. In anenvironment that mixes UNIX and Windows systems, some translation may be requiredwhen performing create file, set security, get security, or access operations.

NFS access of Windows-created filesIf a file contains an owning user or group that is a SID, the system attempts to map it to acorresponding UID or GID before returning it to the caller.

In UNIX, authorization data is retrieved by calling stat(2) on a file and examining theowner, group, and mode bits. Over NFSv3, the GETATTR command functions similarly. Thesystem approximates the mode bits and sets them on the file whenever its ACL changes.Mode bit approximations need to be retrieved only to service these calls.

Note

SID-to-UID and SID-to-GID mappings are cached in both the OneFS ID mapper and thestat cache. If a mapping has recently changed, the file might report inaccurateinformation until the file is updated or the cache is flushed.

SMB access of UNIX-created filesNo UID-to-SID or GID-to-SID mappings are performed when creating an ACL for a file; allUIDs and GIDs are converted to SIDs or principals when the ACL is returned.

OneFS uses a two-step process for returning a security descriptor, which contains SIDs forthe owner and primary group of an object:1. The current security descriptor is retrieved from the file. If the file does not have a

discretionary access control list (DACL), a synthetic ACL is constructed from the file’slower 9 mode bits, which are separated into three sets of permission triples—one



each for owner, group, and everyone. For details about mode bits, see "UNIXpermissions."

2. Two access control entries (ACEs) are created for each triple: the allow ACE containsthe corresponding rights that are granted according to the permissions; the deny ACEcontains the corresponding rights that are denied. In both cases, the trustee of theACE corresponds to the file owner, group, or everyone. After all of the ACEs aregenerated, any that are not needed are removed before the synthetic ACL is returned.

RolesYou can permit and limit access to administrative areas of your EMC Isilon cluster on aper-user basis through the use of roles.

OneFS includes built-in administrator roles with predefined sets of privileges that cannotbe modified. The following list describes what you can and cannot do through roles:

u You can assign privileges through role membership.

u You can add any user to a role as long as the user can authenticate to the cluster.

u You can create custom roles and assign privileges to those roles.

u You can add users singly or as groups, including well-known groups.

u You can assign a user as a member of more than one role.

u You can add a group to a role, which grants to all users who are members of thatgroup all of the privileges associated with the role.

u You cannot assign privileges directly to users or groups.

Note

When OneFS is first installed, only users with root- or admin-level can log in and assignusers to roles.

Built-in rolesBuilt-in roles include privileges to perform a set of administrative functions.

The following tables describe each of the built-in roles from most powerful to leastpowerful. The tables include the privileges and read/write access levels (if applicable)that are assigned to each role. You can assign users and groups to built-in roles and toroles that you create.

Table 1 SecurityAdmin role

Description Privileges Read/writeaccess

- - -Administer security configuration on the cluster,including authentication providers, local users andgroups, and role membership.

ISI_PRIV_LOGIN_CONSOLE N/A

ISI_PRIV_LOGIN_PAPI N/A

ISI_PRIV_LOGIN_SSH N/A

ISI_PRIV_AUTH Read/write

ISI_PRIV_ROLE Read/write


Roles 79

Table 2 SystemAdmin role


- - -Administer all aspects of cluster configuration thatare not specifically handled by the SecurityAdminrole.

ISI_PRIV_LOGIN_CONSOLE N/A



ISI_PRIV_SYS_SHUTDOWN N/A

ISI_PRIV_SYS_SUPPORT N/A

ISI_PRIV_SYS_TIME N/A

ISI_PRIV_ANTIVIRUS Read/write

ISI_PRIV_AUDIT Read/write

ISI_PRIV_CLUSTER Read/write

ISI_PRIV_DEVICES Read/write

ISI_PRIV_EVENT Read/write

ISI_PRIV_FTP Read/write

ISI_PRIV_HTTP Read/write

ISI_PRIV_ISCSI Read/write

ISI_PRIV_JOB_ENGINE Read/write

ISI_PRIV_LICENSE Read/write

ISI_PRIV_NDMP Read/write

ISI_PRIV_NETWORK Read/write

ISI_PRIV_NFS Read/write

ISI_PRIV_NTP Read/write

ISI_PRIV_QUOTA Read/write

ISI_PRIV_REMOTE_SUPPORT Read/write

ISI_PRIV_SMARTPOOLS Read/write

ISI_PRIV_SMB Read/write

ISI_PRIV_SNAPSHOT Read/write

ISI_PRIV_STATISTICS Read/write

ISI_PRIV_SYNCIQ Read/write

ISI_PRIV_VCENTER Read/write

ISI_PRIV_NS_TRAVERSE N/A

ISI_PRIV_NS_IFS_ACCESS N/A



Table 3 AuditAdmin role

Description Privileges Read/write access- - -View all system configuration settings. ISI_PRIV_LOGIN_CONSOLE N/A



ISI_PRIV_ANTIVIRUS Read-only

ISI_PRIV_AUDIT Read-only

ISI_PRIV_CLUSTER Read-only

ISI_PRIV_DEVICES Read-only

ISI_PRIV_EVENT Read-only

ISI_PRIV_FTP Read-only

ISI_PRIV_HTTP Read-only

ISI_PRIV_ISCSI Read-only

ISI_PRIV_JOB_ENGINE Read-only

ISI_PRIV_LICENSE Read-only

SI_PRIV_NDMP Read-only

ISI_PRIV_NETWORK Read-only

ISI_PRIV_NFS Read-only

ISI_PRIV_NTP Read-only

ISI_PRIV_QUOTA Read-only

ISI_PRIV_REMOTE_SUPPORT Read-only

ISI_PRIV_SMARTPOOLS Read-only

ISI_PRIV_SMB Read-only

ISI_PRIV_SNAPSHOT Read-only

ISI_PRIV_STATISTICS Read-only

ISI_PRIV_SYNCIQ Read-only

ISI_PRIV_VCENTER Read-only

Table 4 VMwareAdmin role


- - -Administers remotely all aspects of storageneeded by VMware vCenter.


ISI_PRIV_ISCSI Read/write

ISI_PRIV_NETWORK Read/write

ISI_PRIV_SMARTPOOLS Read/write


Built-in roles 81

Table 4 VMwareAdmin role (continued)


- - -ISI_PRIV_SNAPSHOT Read/write

ISI_PRIV_SYNCIQ Read/write

ISI_PRIV_VCENTER Read/write

ISI_PRIV_NS_TRAVERSE N/A

ISI_PRIV_NS_IFS_ACCESS N/A

OneFS privilegesPrivileges in OneFS are assigned through role membership; privileges cannot be assigneddirectly to users and groups.

Table 5 Login privileges

OneFS privilege User right Privilege type- - -ISI_PRIV_LOGIN_CONSOLE Log in from the console Action

ISI_PRIV_LOGIN_PAPI Log in to the Platform APIand the webadministration interface

Action

ISI_PRIV_LOGIN_SSH Log in through SSH Action

Table 6 System privileges

OneFS privilege User right Privilege type- - -ISI_PRIV_SYS_SHUTDOWN Shut down the system Action

ISI_PRIV_SYS_SUPPORT Run cluster diagnostictools

Action

ISI_PRIV_SYS_TIME Change the system time Action

Table 7 Security privileges

OneFS privilege User right Privilege type- - -ISI_PRIV_AUTH Configure external

authentication providersRead/write

ISI_PRIV_ROLE Create new roles andassign privileges

Read/write



Table 8 Configuration privileges

OneFS privilege User right Privilege type- - -ISI_PRIV_ANTIVIRUS Configure antivirus

scanningRead/write

IS_PRIV_AUDIT Configure auditcapabilities

Read/write

ISI_PRIV_CLUSTER Configure cluster identityand general settings

Read/write

ISI_PRIV_DEVICES Create new roles andassign privileges

Read/write

ISI_PRIV_EVENT View and modify systemevents

Read/write

ISI_PRIV_FTP Configure FTP server Read/write

ISI_PRIV_HTTP Configure HTTP server Read/write

ISI_PRIV_ISCSI Configure iSCSI server Read/write

ISI_PRIV_JOB_ENGINE Schedule cluster-widejobs

Read/write

ISI_PRIV_LICENSE Activate OneFS softwarelicenses

Read/write

ISI_PRIV_NDMP Configure NDMP server Read/write

ISI_PRIV_NETWORK Configure networkinterfaces

Read/write

ISI_PRIV_NFS Configure the NFS server Read/write

ISI_PRIV_NTP Configure NTP Read/write

ISI_PRIV_QUOTA Configure file systemquotas

Read/write

ISI_PRIV_REMOTE_SUPPORT

Configure remote support Read/write

ISI_PRIV_SMARTPOOLS Configure storage pools Read/write

ISI_PRIV_SMB Configure the SMB server Read/write

ISI_PRIV_SNAPSHOT Schedule, take, and viewsnapshots

Read/write

ISI_PRIV_SNMP Configure SNMP server Read/write

ISI_PRIV_STATISTICS View file systemperformance statistics

Read/write

ISI_PRIV_SYNCIQ Configure SyncIQ Read/write

ISI_PRIV_VCENTER Configure VMware forvCenter

Read/write


OneFS privileges 83

Table 9 Namespace privileges

OneFS privilege User right Privilege type- - -ISI_PRIV_NS_TRAVERSE Traverse and view

directory metadataAction

ISI_PRIV_NS_IFS_ACCESS Access the /ifs directory

tree through thenamespace REST service

Action

Table 10 Platform API-only privileges

OneFS privilege User right Privilege type- - -ISI_PRIV_EVENT View and modify system

eventsRead/write

ISI_PRIV_LICENSE Activate OneFS softwarelicenses

Read/write

ISI_PRIV_STATISTICS View file systemperformance statistics

Read/write

Command-line interface privilegesYou can perform most tasks granted by a privilege through the command-line interface.

Some OneFS commands require root access; however, if you do not have root access,most of the commands associated with a privilege can be performed through the sudoprogram. The system automatically generates a sudoers file of users based on existingroles.

Prefixing a command with sudo allows you to run commands that require root access. Forexample, if you do not have root access, the following command fails:

isi sync policy list

However, if you are on the sudoers list, the following command succeeds:

sudo isi sync policy list

The following tables list all One FS commands available, the associated privilege or root-access requirement, and whether sudo is required to run the command.

Note

If you are running in compliance mode, additional sudo commands are available.

Table 11 Privileges sorted by CLI command

isi command Privilege Requires sudo- - -isi alert ISI_PRIV_EVENT x

isi audit ISI_PRIV_AUDIT



Table 11 Privileges sorted by CLI command (continued)

isi command Privilege Requires sudo- - -isi auth - excluding isi authrole

ISI_PRIV_AUTH

isi auth role ISI_PRIV_ROLE

isi avscan ISI_PRIV_ANTIVIRUS x

isi batterystatus ISI_PRIV_STATISTICS x

isi config root

isi dedupe - excluding isidedupe stats

ISI_PRIV_JOB_ENGINE

isi dedupe stats ISI_PRIV_STATISTICS

isi devices ISI_PRIV_DEVICES x

isi domain root

isi email ISI_PRIV_CLUSTER x

isi events ISI_PRIV_EVENT x

isi exttools root

isi fc root

isi filepool ISI_PRIV_SMARTPOOLS

isi firmware root

isi ftp ISI_PRIV_FTP x

isi get root

isi hdfs root

isi iscsi ISI_PRIV_ISCSI x

isi job ISI_PRIV_JOB_ENGINE

isi license ISI_PRIV_LICENSE x

isi lun ISI_PRIV_ISCSI x

isi ndmp ISI_PRIV_NDMP x

isi networks ISI_PRIV_NETWORK x

isi nfs ISI_PRIV_NFS

isi perfstat ISI_PRIV_STATISTICS x

isi pkg root

isi quota ISI_PRIV_QUOTA

isi readonly root


Command-line interface privileges 85

Table 11 Privileges sorted by CLI command (continued)

isi command Privilege Requires sudo- - -isi remotesupport ISI_PRIV_REMOTE_SUPPORT

isi servicelight ISI_PRIV_DEVICES x

isi services root

isi set root

isi smartlock root

isi smb ISI_PRIV_SMB

isi snapshot ISI_PRIV_SNAPSHOT

isi snmp ISI_PRIV_SNMP x

isi stat ISI_PRIV_STATISTICS x

isi statistics ISI_PRIV_STATISTICS x

isi status ISI_PRIV_STATISTICS x

isi storagepool ISI_PRIV_SMARTPOOLS

isi sync ISI_PRIV_SYNCIQ

isi tape ISI_PRIV_NDMP x

isi target ISI_PRIV_ISCSI x

isi update root

isi version ISI_PRIV_CLUSTER x

isi worm root

isi zone ISI_PRIV_AUTH

Table 12 CLI commands sorted by privilege

Privilege isi commands Requires sudo- - -ISI_PRIV_ANTIVIRUS isi avscan x

ISI_PRIV_AUDIT isi audit

ISI_PRIV_AUTH isi auth - excluding isi auth role

isi zone

ISI_PRIV_CLUSTER isi email

isi version

x

ISI_PRIV_DEVICES isi devices

isi servicelight

x



Table 12 CLI commands sorted by privilege (continued)

Privilege isi commands Requires sudo- - -ISI_PRIV_EVENT isi alert

isi events

x

ISI_PRIV_FTP isi ftp x

ISI_PRIV_ISCSI isi iscsi

isi lun

isi target

x

ISI_PRIV_JOB_ENGINE isi job

isi dedupe - excluding isi dedupestats

ISI_PRIV_LICENSE isi license x

ISI_PRIV_NDMP isi ndmp

isi tape

x

ISI_PRIV_NETWORK isi networks x

ISI_PRIV_NFS isi nfs

ISI_PRIV_QUOTA isi quota

ISI_PRIV_ROLE isi auth role

ISI_PRIV_REMOTE_SUPPORT isi remotesupport

ISI_PRIV_SMARTPOOLS isi filepool

isi storagepool

ISI_PRIV_SMB isi smb

ISI_PRIV_SNAPSHOT isi snapshot

ISI_PRIV_SNMP isi snmp x

ISI_PRIV_STATISTICS isi batterystatus

isi dedupe stats

isi perfstat

isi stat

isi statistics

isi status

x

ISI_PRIV_SYNCIQ isi sync

root isi config

isi domain

isi exttools

isi fc


Command-line interface privileges 87

Table 12 CLI commands sorted by privilege (continued)

Privilege isi commands Requires sudo- - -

isi firmware

isi get

isi hdfs

isi pkg

isi readonly

isi services

isi set

isi smartlock

isi update

isi worm

AuthenticationOneFS supports local and remote authentication providers to verify that users attemptingto access the cluster are who they claim to be. Anonymous access, which does notrequire authentication, is supported for protocols that allow it.

OneFS supports the concurrent use of multiple authentication provider types, which areanalogous to directory services. For example, OneFS is often configured to authenticateWindows clients with Active Directory and to authenticate UNIX clients with LDAP. It isimportant that you understand their interactions before enabling multiple providers onthe cluster.

Note

OneFS is RFC 2307-compliant.

NIS, designed by Sun Microsystems, can also be used to authenticate users and groupswhen they access the EMC Isilon cluster.

Supported authentication providersOneFS supports local and remote authentication providers to verify that users attemptingto access the cluster are who they claim to be. Anonymous access, which does notrequire authentication, is supported for protocols that allow it.

OneFS supports the concurrent use of multiple authentication provider types, which areanalogous to directory services. For example, OneFS can be configured to authenticateWindows clients with Active Directory and to authenticate UNIX clients with LDAP. NIS,designed by Sun Microsystems, can also be used to authenticate users and groups whenthey access the Isilon cluster. It is important that you understand their interactions beforeenabling multiple provider types on the cluster.

The following table compares features that are available with each of the authenticationproviders that OneFS supports. In the following table, an 'x' indicates that a feature isfully supported by a provider; an asterisk (*) indicates that additional configuration orsupport from another provider is required.



Authenticationprovider

NTLM Kerberos User/groupmanagement

Netgroups UNIXproperties

Windowsproperties

- - - - - - -ActiveDirectory

x x * x

LDAP * x x x *

NIS x x

Local x x x x

File x x x

Authentication provider featuresYou can configure authentication providers for your environment. Authenticationproviders support a mix of the following features.

Feature Description- -Authentication All authentication providers support plain-text authentication.

Some providers can be configured to support NTLM or Kerberosauthentication also.

Users and groups OneFS provides the ability to manage users and groups directly onthe cluster.

Netgroups Used primarily by NFS, netgroups configure access to NFS exports.

UNIX-centric user and groupproperties

Login shell, home directory, UID, and GID. Missing information issupplemented by configuration templates or additionalauthentication providers.

Windows-centric user andgroup properties

NetBIOS domain and SID. Missing information is supplemented byconfiguration templates.

LDAPThe Lightweight Directory Access Protocol (LDAP) is a networking protocol that enablesyou to define, query, and modify directory services and resources.

OneFS can authenticate users and groups against an LDAP repository in order to grantthem access to the cluster. OneFS supports Kerberos authentication for an LDAP provider.

The LDAP service supports the following features:

u Users, groups, and netgroups.

u Configurable LDAP schemas. For example, the ldapsam schema allows NTLMauthentication over the SMB protocol for users with Windows-like attributes.

u Simple bind authentication (with and without SSL).

u Redundancy and load balancing across servers with identical directory data.

u Multiple LDAP provider instances for accessing servers with different user data.

u Encrypted passwords.


Authentication provider features 89

Active DirectoryThe Active Directory directory service is a Microsoft implementation of LightweightDirectory Access Protocol (LDAP), Kerberos, and DNS technologies that can storeinformation about network resources. Active Directory can serve many functions, but theprimary reason for joining the cluster to an Active Directory domain is to perform user andgroup authentication.

When the cluster joins an Active Directory domain, a single Active Directory machineaccount is created. The machine account is used to establish a trust relationship with thedomain and to enable the cluster to authenticate and authorize users in the ActiveDirectory forest. By default, the machine account is named the same as the cluster;however, if the cluster name is more than 15 characters long, the name is hashed anddisplayed after joining the domain.

Whenever possible, a single Active Directory instance should be used when all domainshave a trust relationship. Multiple instances should be used only to grant access tomultiple sets of mutually-untrusted domains.

Note

If you configure an Active Directory provider, Kerberos authentication is providedautomatically.

NISThe Network Information Service (NIS) provides authentication and identity uniformityacross local area networks. OneFS includes a NIS authentication provider that enablesyou to integrate the cluster with your NIS infrastructure.

NIS, designed by Sun Microsystems, can be used to authenticate users and groups whenthey access the cluster. The NIS provider exposes the passwd, group, and netgroup mapsfrom a NIS server. Hostname lookups are also supported. Multiple servers can bespecified for redundancy and load balancing.

Note

NIS is different from NIS+, which OneFS does not support.

File providerA file provider enables you to supply an authoritative third-party source of user and groupinformation to the cluster. A third-party source is useful in UNIX and Linux environmentsthat synchronize /etc/passwd, /etc/group, and etc/netgroup files acrossmultiple servers.

OneFS uses standard BSD /etc/spwd.db and /etc/group database files as thebacking store for the file provider. You generate the spwd.db file by running thepwd_mkdb command in the OneFS command-line interface (CLI). You can script updatesto the database files.

On the Isilon cluster, a file provider uses libcrypt for password hashing. The ModularCrypt Format is parsed to determine the hashing algorithm. The following algorithms aresupported:

u MD5

u Blowfish



u NT-Hash

u SHA-256

u SHA-512

Note

The built-in System file provider includes services to list, manage, and authenticateagainst system accounts such as root, admin, and nobody. It is recommended that youdo not modify the System file provider.

Local providerThe local provider provides authentication and lookup facilities for user accounts thatwere added by an administrator.

Local authentication can be useful when Active Directory, LDAP, or NIS directory servicesare not used, or when a specific user or application needs to access the cluster. Localgroups can include built-in groups and Active Directory groups as members.

In addition to configuring network-based authentication sources, you can also managelocal users and groups by configuring a local password policy for each node in thecluster. OneFS settings specify password complexity, password age and re-use, andpassword-attempt lockout policies.

Managing access permissionsThe internal representation of identities and permissions can contain information fromUNIX sources, Windows sources, or both. Because access protocols can process theinformation from only one of these sources, the system may need to makeapproximations to present the information in a format the protocol can process.

Configure access management settingsDefault access settings include whether to send NTLMv2 responses for SMB connections;the identity type to store on disk; the Windows workgroup name to use when running inlocal mode; and character substitution for spaces encountered in user and group names.

Procedure

1. Click Cluster Management > Access Management > Settings.

2. Configure the following settings as needed.

Send NTLMv2Configures the type of NTLM response that is sent to an SMB client.

On-Disk IdentityControls the preferred identity to store on disk. If OneFS is unable to convert anidentity to the preferred format, it is stored as is. This setting does not affectidentities that are currently stored on disk. Select one of the following settings:

l native: Let OneFS determine the identity to store on disk. This is therecommended setting.

l unix: Always store incoming UNIX identifiers (UIDs and GIDs) on disk.

l sid: Store incoming Windows security identifiers (SIDs) on disk, unless theSID was generated from a UNIX identifier; in that case, convert it back to theUNIX identifier and store it on disk.


Local provider 91

WorkgroupSpecifies the NetBIOS workgroup. The default value is WORKGROUP.

Space ReplacementFor clients that have difficulty parsing spaces in user and group names, specifiesa substitute character.

3. Click Save.

After you finish

If you changed the on-disk identity selection, it is recommended that you run the RepairPermissions job with the 'Convert permissions' repair task to prevent potentialpermissions errors.

Modify ACL policy settingsYou can modify ACL policy settings but the default ACL policy settings are sufficient formost cluster deployments.

CAUTION

Because ACL policies change the behavior of permissions throughout the system, theyshould be modified only as necessary by experienced administrators with advancedknowledge of Windows ACLs. This is especially true for the advanced settings, which areapplied regardless of the cluster's environment.

For UNIX, Windows, or balanced environments, the optimal permission policy settings areselected and cannot be modified. However, you can choose to manually configure thecluster's default permission settings if necessary to support your particular environment.

Note

You must be logged in to the web administration interface to perform this task.

For a description of each setting option, see ACL policy settings options.

Procedure

1. Click Protocols > ACLs > ACL Policies.

2. In the Standard Settings section, under Environment, click to select the setting thatbest describes your environment, or select Configure permission policies manually toconfigure individual permission policies.

3. If you selected the Configure permission policies manually option, configure thesettings as needed.

For more information about these settings, see ACL policy settings options.

4. In the Advanced Settings section, configure the settings as needed.

ACL policy settings optionsYou can configure an ACL policy by choosing from these settings options.

Setting Description- -UNIX only Causes cluster permissions to operate with UNIX semantics, as opposed to

Windows semantics. Enabling this option prevents ACL creation on the system.



Setting Description- -Balanced Causes cluster permissions to operate in a mixed UNIX and Windows

environment. This setting is recommended for most cluster deployments.

Windows only Causes cluster permissions to operate with Windows semantics, as opposed toUNIX semantics. Enabling this option causes the system to return an error onUNIX chmod requests.

Configurepermissionpoliciesmanually

Allows you to configure the individual permissions policy settings available

under Permission Policies.

ACL creationover SMB

Specifies whether to allow or deny creation of ACLs over SMB. Select one of thefollowing options.

l Do not allow the creation of ACLs over Windows File Sharing(SMB): Prevents ACL creation on the cluster.

l Allow the creation of ACLs over SMB: Allows ACL creation on thecluster.

Note

Inheritable ACLs on the system take precedence over this setting: If inheritableACLs are set on a folder, any new files and folders created in that folder willinherit the folder's ACL. Disabling this setting does not remove ACLs currentlyset on files. If you want to clear an existing ACL, run the chmod -b <mode>

<file> command to remove the ACL and set the correct permissions.

chmod on fileswith existingACLs

Controls what happens when a chmod operation is initiated on a file with an

ACL, either locally or over NFS. This setting controls any elements that set UNIXpermissions, including File System Explorer. Enabling this policy setting doesnot change how chmod operations affect files that do not have ACLs. Select

one of the following options.

l Remove the existing ACL and set UNIX permissions instead: Forchmod operations, removes any existing ACL and instead sets the chmodpermissions. Select this option only if you do not need permissions to beset from Windows.

l Remove the existing ACL and create an ACL equivalent to the UNIXpermissions: Stores the UNIX permissions in a Windows ACL. Select thisoption only if you want to remove Windows permissions but do not wantfiles to have synthetic ACLs.

l Remove the existing ACL and create an ACL equivalent to the UNIXpermissions, for all users/groups referenced in old ACL: Stores theUNIX permissions in a Windows ACL. Select this option only if you want toremove Windows permissions but do not want files to have synthetic ACLs.

l Merge the new permissions with the existing ACL: Causes Windowsand UNIX permissions to operate smoothly in a balanced environment bymerging permissions that are applied by chmod with existing ACLs. An

ACE for each identity (owner, group, and everyone) is either modified orcreated, but all other ACEs are unmodified. Inheritable ACEs are also leftunmodified to enable Windows users to continue to inherit appropriate


ACL policy settings options 93


permissions. However, UNIX users can set specific permissions for each ofthose three standard identities.

l Deny permission to modify the ACL: Prevents users from making NFSand local chmod operations. Enable this setting if you do not want to allow

permission sets over NFS. This setting returns an error when an NFS clientattempts to modify the ACL.

l Ignore operation if file has an existing ACL: Prevents an NFS clientfrom making changes to the ACL. This setting does not return an errorwhen a NFS client attempts to modify the ACL. Select this option if youdefined an inheritable ACL on a directory and want to use that ACL forpermissions.

CAUTION

If you try to run the chmod command on the same permissions that are

currently set on a file with an ACL, you may cause the operation to silently fail—The operation appears to be successful, but if you were to examine thepermissions on the cluster, you would notice that the chmod command had

no effect. As a workaround, you can run the chmod command away from the

current permissions and then perform a second chmod command to revert to

the original permissions. For example, if your file shows 755 UNIXpermissions and you want to confirm this number, you could run chmod 700file; chmod 755 file

ACLs createdon directoriesby UNIXchmod

On Windows systems, the access control entries for directories can define fine-grained rules for inheritance; on UNIX, the mode bits are not inherited. MakingACLs that are created on directories by the chmod command inheritable is

more secure for tightly controlled environments but may deny access to someWindows users who would otherwise expect access.Select one of the following options.

l Make them inheritable

l Do not make them inheritable

chown on files

with existingACLs

Changes a file or folder's owning user or group. Select one of the followingoptions.

l Modify the owner and/or group permissions: Causes the chownoperation to perform as it does in UNIX. Enabling this setting modifies anyACEs in the ACL associated with the old and new owner or group.

l Do not modify the ACL: Cause the NFS chown operation to function as it

does in Windows. When a file owner is changed over Windows, nopermissions in the ACL are changed.




Note

Over NFS, the chown operation changes the permissions and the owner or

owning group. For example, consider a file owned by user Joe with "rwx------"(700) permissions, signifying "rwx" permissions for the owner, but nopermissions for anyone else. If you run the chown command to change

ownership of the file to user Bob, the owner permissions are still "rwx" but theynow represent the permissions for Bob, rather than for Joe. In fact, Joe will havelost all of his permissions. This setting does not affect UNIX chown operations

performed on files with UNIX permissions, and it does not affect Windowschown operations, which do not change any permissions.

Access checks(chmod,

chown)

In UNIX environments, only the file owner or superuser has the right to run achmod or chown operation on a file. In Windows environments, you can

implement this policy setting to give users the right to perform chmodoperations, called the "change permissions" right, or the right to performchown operations, called the "take ownership" right.

Note

The "take ownership" right only gives users the ability to take file ownership,not to give ownership away.

Select one of the following options.

l Allow only owners to chmod or chown: Causes chmod and chownaccess checks to operate with UNIX-like behavior.

l Allow owner and users with 'take ownership' right to chown, andowner and users with 'change permissions' right to chmod: Causeschmod and chown access checks to operate with Windows-like behavior.

Treatment of"rwx"permissions

In UNIX environments, "rwx" permissions signify two things: A user or grouphas read, write, and execute permissions; and a user or group has themaximum possible level of permissions.

When you assign UNIX permissions to a file, no ACLs are stored for that file.However, a Windows system processes only ACLs; Windows does not processUNIX permissions. Therefore, when you view a file's permissions on a Windowssystem, the cluster must translate the UNIX permissions into an ACL. This typeof ACL is called a synthetic ACL. Synthetic ACLs are not stored anywhere;instead, they are dynamically generated as needed and then they arediscarded. If a file has UNIX permissions, you may notice synthetic ACLs whenyou run the ls file command on the cluster in order to view a file’s ACLs.

When you generate a synthetic ACL, the cluster maps UNIX permissions toWindows rights. Windows supports a more granular permissions model thanUNIX does, and it specifies rights that cannot easily be mapped from UNIXpermissions. If the cluster maps "rwx" permissions to Windows rights, youmust enable one of the following options. The main difference between "rwx"and "Full Control" is the broader set of permissions with "Full Control".


l Retain 'rwx' permissions: Generates an ACE that provides only read,write, and execute permissions.


ACL policy settings options 95


l Treat 'rwx' permissions as Full Control: Generates an ACE thatprovides the maximum Windows permissions for a user or a group byadding the "change permissions" right, the "take ownership" right, and the"delete" right.

Group ownerinheritance

Operating systems tend to work with group ownership and permissions in twodifferent ways: BSD inherits the group owner from the file's parent folder;Windows and Linux inherit the group owner from the file creator's primarygroup. If you enable a setting that causes the group owner to be inherited fromthe creator's primary group, it can be overridden on a per-folder basis byrunning the chmod command to set the set-gid bit. This inheritance applies

only when the file is created. For more information, see the manual page forthe chmod command.


l When an ACL exists, use Linux and Windows semantics, otherwiseuse BSD semantics: Controls file behavior based on whether the new fileinherits ACLs from its parent folder. If it does, the file uses the creator'sprimary group. If it does not, the file inherits from its parent folder.

l BSD semantics - Inherit group owner from the parent folder: Causesthe group owner to be inherited from the file's parent folder.

l Linux and Windows semantics - Inherit group owner from thecreator's primary group: Causes the group owner to be inherited fromthe file creator's primary group.

chmod (007)on files withexisting ACLs

Specifies whether to remove ACLs when running the chmod (007) command.


l chmod(007) does not remove existing ACL: Sets 007 UNIXpermissions without removing an existing ACL.

l chmod(007) removes existing ACL and sets 007 UNIX permissions:Removes ACLs from files over UNIX file sharing (NFS) and locally on thecluster through the chmod (007) command. If you enable this setting,

be sure to run the chmod command on the file immediately after using

chmod (007) to clear an ACL. In most cases, you do not want to leave

007 permissions on the file.

Ownerpermissions

It is impossible to represent the breadth of a Windows ACL's access rules usinga set of UNIX permissions. Therefore, when a UNIX client requests UNIXpermissions for a file with an ACL over NFS (an action known as a "stat"), itreceives an imperfect approximation of the file's true permissions. By default,executing an ls -lcommand from a UNIX client returns a more open set ofpermissions than the user expects. This permissiveness compensates forapplications that incorrectly inspect the UNIX permissions themselves whendetermining whether to attempt a file-system operation. The purpose of thispolicy setting is to ensure that these applications proceed with the operationto allow the file system to properly determine user access through the ACL.


l Approximate owner mode bits using all possible owner ACEs:Makes the owner permissions appear more permissive than the actualpermissions on the file.




l Approximate owner mode bits using only the ACE with the ownerID: Makes the owner permissions appear more accurate, in that you seeonly the permissions for a particular owner and not the more permissiveset. However, this may cause access-denied problems for UNIX clients.

grouppermissions

Select one of the following options for group permissions:

l Approximate group mode bits using all possible group ACEs: Makesthe group permissions appear more permissive than the actualpermissions on the file.

l Approximate group mode bits using only the ACE with the groupID: Makes the group permissions appear more accurate, in that you seeonly the permissions for a particular group and not the more permissiveset. However, this may cause access-denied problems for UNIX clients.

No "deny"ACEs

The Windows ACL user interface cannot display an ACL if any "deny" ACEs areout of canonical ACL order. However, in order to correctly represent UNIXpermissions, deny ACEs may be required to be out of canonical ACL order.


l Remove “deny” ACEs from synthetic ACLs: Does not include "deny"ACEs when generating synthetic ACLs. This setting can cause ACLs to bemore permissive than the equivalent mode bits.

l Do not modify synthetic ACLs and mode bit approximations:Specifies to not modify synthetic ACL generation; “deny” ACEs will begenerated when necessary.

CAUTION

This option can lead to permissions being reordered, permanentlydenying access if a Windows user or an application performs an ACL get,an ACL modification, and an ACL set (known as a "roundtrip") to and fromWindows.

Access check(utimes)

You can control who can change utimes, which are the access andmodification times of a file, by selecting one of the following options.

l Allow only owners to change utimes to client-specific times (POSIXcompliant): Allows only owners to change utimes, which complies withthe POSIX standard—an approach that is probably familiar toadministrators of UNIX systems.

l Allow owners and users with ‘write’ access to change utimes toclient-specific times: Allows owners as well as users with write access tomodify utimes—a less restrictive approach that is probably familiar toadministrators of Windows systems.

Update cluster permissionsYou can update file permissions or ownership by running the Repair Permissions job. Toprevent permissions issues that can occur after changing the on-disk identity, run this


Update cluster permissions 97

job with the 'convert permissions' task to ensure that the changes are fully propagatedthroughout the cluster.

Procedure

1. Click Protocols > ACLs > Repair Permissions Job.

2. Optional: From the Priority list, select the priority level at which to run the job inrelation to other jobs.

3. Optional: From the Impact policy list, select an impact policy for the job to follow.

4. From the Repair task list, select one of the following methods for updatingpermissions:

Options Description

Convertpermissions

For each file and directory in the specified Path to repairdirectory, converts the owner, group, and access control list(ACL) to the target on-disk identity.

Clonepermissions

Applies the permissions settings for the directory specifiedby the Template Directory setting to the Path to repairdirectory.

Inheritpermissions

Recursively applies the ACL of the directory that is specifiedby the Template Directory setting to each file andsubdirectory in the specified Path to repair directory,according to standard inheritance rules.

The remaining settings differ depending on the selected repair task.

5. In the Path to repair field, type or browse to the directory in /ifs whose permissionsyou want to repair.

6. Optional: In the Template Directory field, type or browse to the directory in /ifs thatyou want to copy permissions from. This setting applies to only the Clonepermissions and Inherit permissions repair tasks.

7. Optional: From the Target list, select the preferred on-disk identity type to apply. Thissetting applies to only the Convert permissions repair task.

Options Description

Use default systemtype

Applies the system's default identity type.

Use native type If a user or group does not have an authoritativeUNIX identifier (UID or GID), applies the Windowsidentity type (SID).

Use UNIX type Applies the UNIX identity type.

Use SID (Windows)type

Applies the Windows identity type.

8. Optional: From the Access Zone list, select an access zone to use for ID mapping. Thissetting applies to only the Convert permissions repair task.



Managing rolesYou can view, add, or remove members of any role. Except for built-in roles, whoseprivileges you cannot modify, you can add or remove OneFS privileges on a role-by-rolebasis.

Note

Roles take both users and groups as members. If a group is added to a role, all users whoare members of that group are assigned the privileges associated with the role. Similarly,members of multiple roles are assigned the combined privileges of each role.

View rolesYou can view information about built-in and custom roles.

This procedure must be performed through the command-line interface (CLI).

Procedure


2. At the command prompt, run one of the following commands.

l To view a basic list of all roles on the cluster, run:

isi auth roles listl To view detailed information about each role on the cluster, including member and

privilege lists, run:

isi auth roles list --verbosel To view detailed information about a single role, run the following command,

where <role> is the name of the role:

isi auth roles view <role>

View privilegesYou can view user privileges.

This procedure must be performed through the command-line interface (CLI). You canview a list of your privileges or the privileges of another user using the followingcommands:

Procedure

1. At the command prompt, run one of the following commands.

l To view a list of all privileges:isi auth privileges --verbose

l To view a list of your privileges:isi auth id

l To view a list of privileges for another user, where <user> specifies the user byname:isi auth mapping token <user>


Managing roles 99

Create a custom roleTo create a custom role, you must first create an empty role and then add privileges andmembers to the role.


Procedure


2. At the command prompt, run the following command, where <name> is the name thatyou want to assign to the role and --description <string> specifies an optionaldescription:

isi auth roles create <name> [--description <string>]

After you finish

Add privileges and members to the role by running the isi auth roles modifycommand.

Modify a roleYou can modify the description and the user or group membership of any role, includingbuilt-in roles. However, you cannot modify the name or privileges that are assigned tobuilt-in roles.


Procedure


2. At the command prompt, run the following command, where <role> is the role nameand <options> are optional parameters:

isi auth roles modify <role> [<options>]

Delete a custom roleDeleting a role does not affect the privileges or users that are assigned to it. Built-in rolescannot be deleted.


Procedure


2. At the command prompt, run the following command, where <role> is the name of therole that you want to delete:

isi auth roles delete <role>3. At the confirmation prompt, type y.

Managing authentication providersYou can configure one or more LDAP, Active Directory, NIS, and file providers. A localprovider is created automatically when you create an access zone, which allows you tocreate a configuration for each access zone so it has its own list of local users that can



authenticate to it. You also can create a password policy for each local provider toenforce password complexity.

Configure an LDAP providerBy default, when you configure an LDAP provider, it is automatically added to the Systemaccess zone.

Procedure

1. Click Cluster Management > Access Management > LDAP.

2. Click Add an LDAP provider.

3. In the LDAP Provider Name field, type a name for the provider.

4. In the Servers field, type one or more valid LDAP server URIs, one per line, in theformat ldaps://server:port (secure LDAP) or ldap://server:port (non-secure LDAP).

Note

If you do not specify a port, the default port is used. The default port for non-secureLDAP (ldap://) is 389; for secure LDAP (ldaps://) it is 636. If you specify non-secureLDAP, the bind password is transmitted to the server in clear text.

5. Optional: Configure the following settings as needed.

Load balance serversSelect the check box to connect to a random server, or clear the check box toconnect according to the order in which the servers are listed in the Servers field.

Base Distinguished NameType the distinguished name (DN) of the entry at which to start LDAP searches.Base DNs can include cn (Common Name), l (Locality), dc (Domain Component),ou (Organizational Unit), or other components. For example, dc=emc,dc=com is abase DN for emc.com.

Bind toType the distinguished name of the entry at which to bind to the LDAP server.

PasswordSpecify the password to use when binding to the LDAP server. Use of thispassword does not require a secure connection; if the connection is not usingTransport Layer Security (TLS), the password is sent in clear text.

6. Optional: To modify the default settings for user, group, and netgroup queries, clickDefault Query Settings.

7. Optional: To modify the settings for user queries and home directory provisioning,click User Query Settings.

8. Optional: To modify the settings for group queries, click Group Query Settings.

9. Optional: To modify the settings for netgroup queries, click Netgroup Query Settings.

10.Optional: To modify the default LDAP attributes that contain user information or tomodify LDAP security settings, click Advanced LDAP Settings.

11.Click Add LDAP provider.


Configure an LDAP provider 101

Managing LDAP providersYou can view, modify, and delete LDAP providers or you can stop using an LDAP providerby removing it from all access zones that are using it.

Modify an LDAP providerYou can modify any setting for an LDAP provider except its name. You must specify atleast one server for the provider to be enabled.

Procedure


2. In the list of LDAP providers, click View details for the provider whose settings youwant to modify.


4. Optional: Click Close.

Delete an LDAP providerWhen you delete an LDAP provider, it is removed from all access zones. As an alternative,you can stop using an LDAP provider by removing it from each access zone that containsit so that the provider remains available for future use.

Procedure


2. Click Delete for the provider that you want to delete.


Configure Kerberos settingsOneFS includes a Kerberos configuration file for Active Directory in addition to the globalKerberos configuration file.

This procedure is available only through the OneFS command-line interface (CLI).

Note

Most settings require modification only if you are using a Kerberos Key DistributionCenter (KDC) other than Active Directory—for example, if you are using an MIT KDC for NFSversion 3 or version 4 authentication.

Procedure


2. Run the isi auth krb5 command with the add, modify, or delete sub-command to specify which entries to modify in the Kerberos configuration file.

3. Write the changes to the file by running the isi auth krb5 write command.

Note

By default, changes are written to the global Kerberos configuration file, /etc/krb5.conf. To update the Kerberos configuration file for Active Directory, include

the --path option to specify the /etc/likewise-krb5-ad.conf file.



LDAP query settingsYou can configure the entry point and depth at which to search for LDAP users, groups,and netgroups. You can also configure the settings for user home directory provisioning.

Note


Distinguished NameSpecifies the base distinguished name (base DN) of the entry at which to start LDAPsearches for user, group, or netgroup objects. Base DNs can include cn (CommonName), l (Locality), dc (Domain Component), ou (Organizational Unit), or othercomponents. For example, dc=emc,dc=com is a base DN for emc.com.

Search ScopeSpecifies the depth from the base DN at which to perform LDAP searches.

The following values are valid:

defaultApplies the search scope that is defined in the default query settings. Thisoption is not available for the default query search scope.

baseSearches only the entry at the base DN.

onelevelSearches all entries exactly one level below the base DN.

subtreeSearches the base DN and all entries below it.

childrenSearches all entries below the base DN, excluding the base DN itself.

Search TimeoutSpecifies the number of seconds after which to stop retrying and fail a search. Thedefault value is 100. This setting is available only in the default query settings.

Query FilterSpecifies the LDAP filter for user, group, or netgroup objects. This setting is notavailable in the default query settings.

Authenticate users from this providerSpecifies whether to allow the provider to respond to authentication requests. Thissetting is available only in the user query settings.

Home Directory NamingSpecifies the path to use as a template for naming home directories. The path mustbegin with /ifs and can contain variables, such as %U, that are expanded togenerate the home directory path for the user. This setting is available only in theuser query settings.


Managing LDAP providers 103

Create home directories on first loginSpecifies whether to create a home directory the first time a user logs in, if a homedirectory does not already exist for the user. This setting is available only in the userquery settings.

UNIX ShellSpecifies the path to the user's login shell, for users who access the file systemthrough SSH. This setting is available only in the user query settings.

LDAP advanced settingsYou can configure LDAP security settings and specify the LDAP attributes that containuser information.

Note


Name AttributeSpecifies the LDAP attribute that contains UIDs, which are used as login names. Thedefault value is uid.

Common Name AttributeSpecifies the LDAP attribute that contains common names (CNs). The default value iscn.

Email AttributeSpecifies the LDAP attribute that contains email addresses. The default value ismail.

GECOS Field AttributeSpecifies the LDAP attribute that contains GECOS fields. The default value is gecos.

UID AttributeSpecifies the LDAP attribute that contains UID numbers. The default value isuidNumber.

GID AttributeSpecifies the LDAP attribute that contains GIDs. The default value is gidNumber.

Home Directory AttributeSpecifies the LDAP attribute that contains home directories. The default value ishomeDirectory.

UNIX Shell AttributeSpecifies the LDAP attribute that contains UNIX login shells. The default value isloginShell.

Netgroup Members AttributeSpecifies the LDAP attribute that contains netgroup members. The default value ismemberNisNetgroup.

Netgroup Triple AttributeSpecifies the LDAP attribute that contains netgroup triples. The default value isnisNetgroupTriple.



Group Members AttributeSpecifies the LDAP attribute that contains group members. The default value ismemberUid.

Unique Group Members AttributeSpecifies the LDAP attribute that contains unique group members. This attribute isused to determine which groups a user belongs to if the LDAP server is queried bythe user’s DN instead of the user’s name. This setting has no default value.

UNIX Password AttributeSpecifies the LDAP attribute that contains UNIX passwords. This setting has nodefault value.

Windows Password AttributeSpecifies the LDAP attribute that contains Windows passwords. The default value isntpasswdhash.

Certificate Authority FileSpecifies the full path to the root certificates file.

Require secure connection for passwordsSpecifies whether to require a Transport Layer Security (TLS) connection.

Ignore TLS ErrorsContinues over a secure connection even if identity checks fail.

Configure an Active Directory providerYou can configure one or more Active Directory providers, each of which must be joined toa separate Active Directory domain. By default, when you configure an Active Directoryprovider, it is automatically added to the System access zone.

Note

If you migrate users to a new or different Active Directory domain, you must re-set the ACLdomain information after you configure the new provider. Third-party tools can be used,such as Microsoft SubInACL.

Procedure

1. Click Cluster Management > Access Management > Active Directory.

2. Click Join a domain.

3. In the Domain Name field, type a fully qualified Active Directory domain name.

The domain name will also be used as the provider name.

4. In the User field, type the username of an account that is authorized to join the ActiveDirectory domain.

5. In the Password field, type the password of the user account.

6. Optional: In the Organizational Unit field, type the name of the organizational unit(OU) to connect to on the Active Directory server. Specify the OU in the form OuName orOuName1/SubName2.

7. Optional: In the Machine Account field, type the name of the machine account.


Configure an Active Directory provider 105

Note

If you specified an OU to connect to, the domain join will fail if the machine accountdoes not reside in the OU.

8. Optional: To enable Active Directory authentication for NFS, select the Enable SecureNFS check box.

If you enable this setting, OneFS registers NFS service principal names (SPNs) duringthe domain join.

9. Optional: To configure advanced settings, click Advanced Active Directory Settings.

10.Click Join.

Managing Active Directory providersYou can view, modify, and delete Active Directory providers. OneFS includes a Kerberosconfiguration file for Active Directory in addition to the global Kerberos configuration file,both of which you can configure through the command-line interface.

Modify an Active Directory providerYou can modify the advanced settings for an Active Directory provider.

Procedure


2. In the list of Active Directory providers, click View details for the provider whosesettings you want to modify.

3. Click Advanced Active Directory Settings.


5. Optional: Click Close.

Delete an Active Directory providerWhen you delete an Active Directory provider, you disconnect the cluster from the ActiveDirectory domain that is associated with the provider, disrupting service for users whoare accessing it. After you leave an Active Directory domain, users can no longer accessthe domain from the cluster.

Procedure


2. In the Active Directory Providers table, click Leave for the domain you want to leave.

3. In the confirmation dialog box, click Leave.

Configure Kerberos settingsOneFS includes a Kerberos configuration file for Active Directory in addition to the globalKerberos configuration file.

This procedure is available only through the OneFS command-line interface (CLI).



Note

Most settings require modification only if you are using a Kerberos Key DistributionCenter (KDC) other than Active Directory—for example, if you are using an MIT KDC for NFSversion 3 or version 4 authentication.

Procedure


2. Run the isi auth krb5 command with the add, modify, or delete sub-command to specify which entries to modify in the Kerberos configuration file.

3. Write the changes to the file by running the isi auth krb5 write command.

Note

By default, changes are written to the global Kerberos configuration file, /etc/krb5.conf. To update the Kerberos configuration file for Active Directory, include

the --path option to specify the /etc/likewise-krb5-ad.conf file.

Active Directory provider settingsYou can view or modify the advanced settings for an Active Directory provider.

Setting Description- -Services For UNIX Specifies whether to support RFC 2307 attributes for domain

controllers. RFC 2307 is required for Windows UNIX Integration andServices For UNIX technologies.

Map to primary domain Enables the lookup of unqualified user names in the primary domain.If this setting is not enabled, the primary domain must be specifiedfor each authentication operation.

Ignore Trusted Domains Ignores all trusted domains.

Trusted Domains Specifies trusted domains to include if the Ignore TrustedDomains setting is enabled.

Domains to Ignore Specifies trusted domains to ignore even if the Ignore TrustedDomains setting is disabled.

Offline Alerts Sends an alert if the domain goes offline.

Enhanced Privacy Encrypts communication to and from the domain controller.

Home Directory Naming Specifies the path to use as a template for naming home directories.The path must begin with /ifs and can contain variables, such as

%U, that are expanded to generate the home directory path for theuser.

Create Home Directory Creates a home directory the first time a user logs in, if a homedirectory does not already exist for the user.

UNIX Shell Specifies the path to the login shell to use if the Active Directoryserver does not provide login-shell information. This setting appliesonly to users who access the file system through SSH.


Managing Active Directory providers 107

Setting Description- -Lookup User Looks up Active Directory users in all other providers before

allocating a UID.

Match Users withLowercase

Normalizes Active Directory user names to lowercase before lookup.

Auto-assign UIDs Enables UID allocation for unmapped Active Directory users.

Lookup Group Looks up Active Directory groups in all other providers beforeallocating a GID.

Match Groups withLowercase

Normalizes Active Directory group names to lowercase before lookup.

Auto-assign GIDs Enables GID allocation for unmapped Active Directory groups.

Make UID/GIDassignments for users andgroups in these specificdomains

Restricts user and group lookups to the specified domains.

Managing NIS providersYou can view and modify NIS providers or delete providers that are no longer needed. Asan alternative to deleting a NIS provider, you can remove it from any access zones thatare using it.

Configure an NIS providerBy default, when you configure an NIS provider it is automatically added to the Systemaccess zone.

Procedure

1. Click Cluster Management > Access Management > NIS.

2. Click Add a NIS provider.

3. In the NIS Provider Name field, type a name for the provider.

4. In the Servers field, type one or more valid NIS server IP addresses, host names, orfully qualified domain names (FQDNs), separated by commas.

Note

If the Load balance servers option is not selected, servers are accessed in the order inwhich they are listed.

5. Optional: Configure the Load balance servers setting:

l To connect to a random server, select the check box.

l To connect according to the order in which the servers are listed in the Serversfield, clear the check box.

6. Optional: Click Default Query Settings and then configure the following settings:

NIS DomainSpecifies the NIS domain name.



Search TimeoutSpecifies the number of seconds after which to stop retrying and fail a search.The default value is 100.

Retry FrequencySpecifies the timeout period in seconds after which a request will be retried. Thedefault value is 5.

7. Optional: Click User Query Settings and then configure the following settings:

Authenticate users from this providerSpecifies whether to allow the provider to respond to authentication requests.

Home Directory NamingSpecifies the path to use as a template for naming home directories. The pathmust begin with /ifs and can contain variables, such as %U, that are expandedto generate the home directory path for the user.

Create home directories on first loginSpecifies whether to create a home directory the first time a user logs in, if ahome directory does not already exist for the user.

UNIX ShellSpecifies the path to the user's login shell, for users who access the file systemthrough SSH.

8. Optional: Click Host Name Query Settings and then configure the Resolve Hostssetting:

l To enable host resolution, select the check box.

l To disable host resolution, clear the check box.

9. Click Add NIS provider.

Modify an NIS providerYou can modify any setting for an NIS provider except its name. You must specify at leastone server for the provider to be enabled.

Procedure


2. In the list of NIS providers, click View details for the provider whose settings you wantto modify.


4. Click Close.

Delete an NIS providerWhen you delete an NIS provider, it is removed from all access zones. As an alternative,you can stop using an NIS provider by removing it from each access zone that contains itso that the provider remains available for future use.

Procedure



Managing NIS providers 109

2. Click Delete for the provider that you want to delete.


Configuring file providersYou can configure one or more file providers, each with its own combination ofreplacement files, for each access zone. Password database files, which are also calleduser database files, must be in binary format.

Configure a file providerYou can specify replacement files for any combination of users, groups, and netgroups.

Procedure

1. Click Cluster Management > Access Management > File Provider.

2. Click Add a file provider.

3. In the File Provider Name field, type a name for the file provider.

4. Optional: Specify one or more of the following replacement files.

l To specify a user replacement file, in the Users File field, type or browse to thelocation of the spwd.db file.

l To specify a group replacement file, in the Groups File field, type or browse to thelocation of the group file.

l To specify a netgroup replacement file, in the Netgroups File field, type or browseto the location of the netgroup file.

5. Optional: To enable this provider to authenticate users, select the Authenticate usersfrom this provider check box.

6. Optional: In the Home Directory Naming field, type the path to use as a template fornaming home directories. The path must begin with /ifs and may contain variables,such as %U, that are expanded to generate the home directory path for the user.

7. Optional: To automatically create a home directory the first time a user logs in, selectthe Create home directories on first login check box.

This setting applies only if a home directory does not already exist for the user.

8. Optional: From the UNIX Shell list, select the login shell for users who access the filesystem through SSH.

9. Click Add File Provider.

Generate a password fileA file provider requires the password database file to be in binary format. To generate abinary password file, you must run the pwd_mkdb command in the OneFS command-lineinterface (CLI).

Procedure


2. Run the following command, where -d <directory> specifies the location in which tostore the spwd.db file and <file> specifies the location of the source password file:

pwd_mkdb -d <directory> <file>



Note

If you omit the -d option, the file is created in the /etc directory. For full commandusage guidelines, view the manual ("man") page by running the man pwd_mkdbcommand.

The following command generates an spwd.db file in the /ifs directory from apassword file that is located at /ifs/test.passwd:

pwd_mkdb -d /ifs /ifs/test.passwd

Managing file providersEach file provider pulls directly from up to three replacement database files: a group filethat uses the same format as /etc/group; a netgroups file; and a binary password file,spwd.db, which provides fast access to the data in a file that uses the /etc/master.passwd format. You must copy the replacement files to the cluster andreference them by their directory path.

Note

If the replacement files are located outside the /ifs directory tree, you must manuallydistribute them to every node in the cluster. Changes that are made to the systemprovider's files are automatically distributed across the cluster.

Modify a file providerYou can modify any setting for a file provider, with the exception that you cannot renamethe System file provider.

Procedure


2. In the File Providers table, click View details for the provider whose settings you wantto modify.


4. Click Close.

Delete a file providerTo stop using a file provider, you can clear all of its replacement file settings or you canpermanently delete the provider.

Procedure


2. In the File Providers table, click Delete for the provider that you want to delete.



Managing file providers 111

Password file formatThe file provider uses a binary password database file, spwd.db. You can generate abinary password file from a master.passwd-formatted file by running the pwd_mkdbcommand.

The master.passwd file contains ten colon-separated fields, as shown in the followingexample:

admin:*:10:10::0:0:Web UI Administrator:/ifs/home/admin:/bin/zsh

The fields are defined below in the order in which they appear in the file.

Note

UNIX systems often define the passwd format as a subset of these fields, omitting theClass, Change, and Expiry fields. To convert a file from passwd to master.passwdformat, add :0:0: between the GID field and the Gecos field.

UsernameThe user name. This field is case-sensitive. OneFS does not set a limit on the length;however, many applications truncate the name to 16 characters.

PasswordThe user’s encrypted password. If authentication is not required for the user, anasterisk (*) can be substituted for a password. The asterisk character is guaranteedto not match any password.

UIDThe UNIX user identifier. This value must be a number in the range 0-4294967294that is not reserved or already assigned to a user. Compatibility issues will occur ifthis value conflicts with an existing account's UID.

GIDThe group identifier of the user’s primary group. All users are a member of at leastone group, which is used for access checks and can also be used when creatingfiles.

ClassThis field is not supported by OneFS and should be left empty.

ChangeOneFS does not support changing the passwords of users in the file provider. Thisfield is ignored.

ExpiryOneFS does not support the expiration of user accounts in the file provider. This fieldis ignored.

GecosThis field can store a variety of information but is usually used to store the user’s fullname.

HomeThe absolute path to the user’s home directory, beginning at /ifs.



ShellThe absolute path to the user’s shell. If this field is set to /sbin/nologin, the useris denied command-line access.

Group file formatThe file provider uses a group file in the format of the /etc/group file that exists onmost UNIX systems.

The group file consists of one or more lines containing four colon-separated fields, asshown in the following example:

admin:*:10:root,admin

The fields are defined below in the order in which they appear in the file.

Group nameThe name of the group. This field is case-sensitive. Although OneFS does not set alimit on the length of the group name, many applications truncate the name to 16characters.

PasswordThis field is not supported by OneFS and should contain an asterisk (*).

GIDThe UNIX group identifier. Valid values are any number in the range 0-4294967294that is not reserved or already assigned to a group. Compatibility issues will occur ifthis value conflicts with an existing group's GID.

Group membersA comma-delimited list of user names.

Netgroup file formatA netgroup file consists of one or more netgroups, each of which can contain members.Members of a netgroup can be hosts, users, or domains that are specified in a membertriple. A netgroup can also contain another netgroup.

Each entry in a netgroup file consists of the netgroup name, followed by a space-delimited set of member triples and nested netgroup names. If you specify a nestednetgroup, it must be defined on a separate line in the file.

A member triple takes the form (host, user, domain), where host is a machinename, user is a user name, and domain is a domain name. Any combination is validexcept an empty triple (,,).

The following sample file contains two netgroups. The 'rootgrp' netgroup contains fourhosts: two hosts are defined in member triples and two hosts are contained in the nested'othergrp' netgroup, which is defined on the second line.

rootgrp (myserver, root, somedomain.com) (otherserver, root, somedomain.com) othergrpothergrp (other-win,, somedomain.com) (other-linux,, somedomain.com)

Note

A new line signifies a new netgroup. You can continue a long netgroup entry to the nextline by typing a backslash character (\) in the right-most position of the first line.


Managing file providers 113

Create a local userEach access zone includes a local provider that allows you to create and manage localusers and groups. When creating a local user account, you can configure its name,password, home directory, UNIX user identifier (UID), UNIX login shell, and groupmemberships.

Procedure

1. Click Cluster Management > Access Management > Users.

2. From the Select a zone list, select an access zone (for example, System).

3. From the Select a provider list, select the local provider for the zone (for example,LOCAL:System).

4. Click Create a user.

5. In the Username field, type a username for the account.

6. In the Password field, type a password for the account.

7. Optional: Configure the following additional settings as needed.

l Allow password to expire: Select this check box to specify that the password isallowed to expire.

l UID: If this setting is left blank, the system automatically allocates a UID for theaccount. This is the recommended setting.

Note

You cannot assign a UID that is in use by another local user account.

l Full Name: Type a full name for the user.

l Email Address: Type an email address for the account.

l Primary Group: Click Select group to specify the owner group.

l Additional Groups: Specify any additional groups to make this user a member of.

l Home Directory: Type the path to the user's home directory. If you do not specify apath, a directory is automatically created at /ifs/home/<Username>.

l UNIX Shell: This setting applies only to users who access the file system throughSSH. From the list, click the shell that you want. By default, the /bin/zsh shell isselected.

l Enabled: Select this check box to allow the user to authenticate against the localdatabase for SSH, FTP, HTTP, and Windows file sharing through SMB. This settingis not used for UNIX file sharing through NFS.

l Account Expires: Optionally select one of the following options:

– Never expires: Click to specify that this account does not have an expirationdate.

– Account expires on: Click to display the Expiration date field, and then type thedate in the format mm/dd/yyyy.

l Prompt password change: Select this check box to prompt for a password changethe next time the user logs in.

8. Click Create User.



Create a local groupIn the local provider of an access zone, you can create groups and assign members tothem.

Procedure

1. Click Cluster Management > Access Management > Groups.


3. From the Select a provider list that appears, select the local provider for the zone (forexample, LOCAL:System).

4. Click the Create a group link.

5. In the Group Name box, type a name for the group.

6. Optional: To override automatic allocation of the UNIX group identifier (GID), in theGID box, type a numerical value.

Note

You cannot assign a GID that is in use by another group. It is recommended that youleave this field blank to allow the system to automatically generate the GID.

7. Optional: Follow these steps for each member that you want to add the group:

a. For the Members setting, click Add user. The Select a User dialog box appears.

b. For the Search for setting, select either Users or Well-known SIDs.

c. If you selected Users, specify values for the following fields:

– Username: Type all or part of a user name, or leave the field blank to return allusers. Wildcard characters are accepted.

– Access Zone: Select the access zone that contains the authentication providerthat you want to search.

– Provider: Select an authentication provider.

d. Click Search.

e. In the Search Results table, select a user and then click Select.

The dialog box closes.

8. Click Create.

Managing local users and groupsAlthough you can view the users and groups of any authentication provider, you cancreate, modify, and delete users and groups in the local provider only.

View a list of users or groups by providerYou can view the users and groups of any authentication provider.

Procedure

1. Click Cluster Management > Access Management.

2. Click one of the following, depending on what you want to view:


Create a local group 115

Option Description

Users Select this tab to view all users by provider.

Groups Select this tab to view all groups by provider.


4. From the Select a provider list, select the local provider for the access zone (forexample, LOCAL:System).

Modify a local userYou can modify any setting for a local user account except the user name.

Procedure




4. In the list of users, click View details for the local user whose settings you want tomodify.


6. Click Close.

Modify a local groupYou can add or remove members from a local group.

Procedure




4. In the list of groups, click View details for the local group whose settings you want tomodify.

5. For the Members setting, click Edit.

6. Add or remove the users that you want, and then click Save.

7. Click Close.

Delete a local userA deleted user can no longer access the cluster through the command-line interface, webadministration interface, or file access protocol. When you delete a local user account, itshome directory remains in place.

Procedure






4. Click Delete for the user that you want to delete.


Delete a local groupYou can delete a local group even if members are assigned to it; deleting a group doesnot affect the members of that group.

Procedure




4. Click Delete for the group that you want to delete.


Configure a local password policyYou can configure a local password policy for a local provider.

This procedure cannot be performed through the web administration interface.

Note

You must configure a separate password policy for each access zone. Each access zonein the cluster contains a separate instance of the local provider, which allows eachaccess zone to have its own list of local users who can authenticate. Passwordcomplexity is configured for each local provider, not for each user.

Procedure


2. Optional: To view the current password settings, run the following command:

isi auth local view system3. Run the isi auth local modify command, choosing from local password policy

settings options.

The --password-complexity parameter must be specified for each setting, asdemonstrated in the following example:isi auth local modify system --password-complexity=lowercase \ --password-complexity=uppercase --password-complexity=numeric \ --password-complexity=symbol

The following command is an example of how to configure a local password policy for alocal provider. isi auth local modify provider-name=<provider-name> \ --min-password-length=15 \ --lockout-duration=15m \ --lockout-window=1m \ --lockout-threshold=5 \ --add-password-complexity=uppercase \ --add-password-complexity=numeric

Local password policy settingsYou can configure local password policy settings and specify the default for each settingthrough the isi auth local modify command. Password complexity increases the


Managing local users and groups 117

number of possible passwords that an attacker must check before the correct passwordcan be guessed.

Setting Description Comments- - -min-password-length Minimum password length

in characters.Long passwords are best. Theminimum length should not be solong that users have a difficult timeentering or remembering thepassword.

password-complexity A list of cases that a newpassword must contain. Bydefault, the list is empty.

You can specify as many as fourcases. The following cases are valid:

l uppercase

l lowercase

l numeric

l symbol (excluding # and @)

min-password-age The minimum passwordage. You can set this valueusing characters for units;for example, 4W for 4weeks, 2d for 2 Days.

A minimum password age can beused to ensure that a user cannotenter a temporary password and thenimmediately change it to the previouspassword. Attempts to check or set apassword before the time expires aredenied.

max-password-age The maximum passwordage. You can set this valueusing characters for units;for example, 4W for 4weeks, 2d for 2 Days.

Attempts to login after a passwordexpires forces a password change. Ifa password change dialog cannot bepresented, the user is not allowed tologin.

password-history-length

The number of historicalpasswords to keep. Newpasswords are checkedagainst this list andrejected if the password isalready present. The maxhistory length is 24.

To avoid recycling of passwords, youcan specify the number of previouspasswords to remember. If a newpassword matches a rememberedprevious password, it is rejected.

lockout-duration The length of time inseconds that an account islocked after a configurablenumber of bad passwordsare entered.

After an account is locked, it isunavailable from all sources until it isunlocked. OneFS provides twoconfigurable options to avoidadministrator interaction for everylocked account:

l Specify how much time mustelapse before the account isunlocked.

l Automatically reset the incorrect-password counter after aspecified time, in seconds.

lockout-threshold The number of incorrectpassword attempts beforean account is locked. A

After an account is locked, it isunavailable from all sources until it isunlocked.



Setting Description Comments- - -

value of zero disablesaccount lockout.

lockout-window The time that elapsesbefore the incorrectpassword attempts count isreset.

If the configured number of incorrectpassword attempts is reached, theaccount is locked and lockout-duration determines the length oftime that the account is locked. Avalue of zero disables the window.


Managing local users and groups 119

CHAPTER 6

Identity management

This section includes the following topics:

u Identity management.......................................................................................... 122u Identity types...................................................................................................... 122u Access tokens..................................................................................................... 123u Access token generation..................................................................................... 124

Identity management 121

Identity managementIn environments with several different types of directory services, OneFS maps the usersand groups from the separate services to provide a single unified identity on the EMCIsilon cluster and uniform access control to files and directories, regardless of theincoming protocol. This process is called identity mapping.

Isilon clusters are frequently deployed in multiprotocol environments with multiple typesof directory services, such as Active Directory and LDAP. When a user with accounts inmultiple directory services logs in to an Isilon cluster, OneFS combines the user’sidentities and privileges from all the directory services into a native access token. Youcan configure OneFS settings to include a list of rules for token manipulation to controluser identity and privileges. For example, you can set a user mapping rule to merge anActive Directory identity and an LDAP identity into a single token that works for access tofiles stored over both SMB and NFS. The token can include groups from Active Directoryand LDAP. The mapping rules that you create can solve identity problems bymanipulating access tokens in many ways, including the following examples:

u Authenticate a user with Active Directory but give the user a UNIX identity.

u Select a primary group from competing choices in Active Directory or LDAP.

u Disallow login of users that do not exist in both Active Directory and LDAP.

For more information about identity management, see the white paper Managingidentities with the Isilon OneFS user mapping service (white paper) at EMC OnlineSupport (https://support.emc.com).

Identity typesOneFS supports three primary identity types, each of which can be stored directly on thefile system. These types are user identifier and group identifier for UNIX, and securityidentifier for Windows.

When you log on to an Isilon cluster, the user mapper expands your identity to includeyour other identities from all the directory services, including Active Directory, LDAP, andNIS. After OneFS maps your identities across the directory services, it generates anaccess token that includes the identity information associated with your accounts. Atoken includes the following identifiers:

u A UNIX user identifier (UID) and a group identifier (GID). A UID or GID is a 32-bitnumber with a maximum value of 4,294,967,295.

u A security identifier (SID) for a Windows user account. A SID is a series of authoritiesand sub-authorities ending with a 32-bit relative identifier (RID). Most SIDs have theform S-1-5-21-A-B-C-<RID> , where A, B, and C are specific to a domain orcomputer and <RID> denotes the object in the domain.

u A primary group SID for a Windows group account.

u A list of supplemental identities, including all groups in which the user is a member.

The token also contains privileges that stem from administrative role-based accesscontrol.

On an Isilon cluster, a file contains permissions, which appear as an access control list(ACL). The ACL controls access to directories, files, and other securable system objects.

When a user tries to access a file, OneFS compares the identities in the user’s accesstoken with the file’s ACL. OneFS grants access when the file’s ACL includes an accesscontrol entry (ACE) that allows the identity in the token to access the file and that does

Identity management


not include an ACE that denies the identity access. OneFS compares the access token ofa user with the ACL of a file.

Note

For more information about access control lists, including a description of thepermissions and how they correspond to POSIX mode bits, see the white paper titledEMC Isilon multiprotocol data access with a unified security model on the EMC OnlineSupport web site (https://support.emc.com).

When a name is provided as an identifier, it is converted into the corresponding user orgroup object and the correct identity type. There are various ways that a name can beentered or displayed:

u UNIX assumes unique case-sensitive namespaces for users and groups. For example,"Name" and "name" represent different objects.

u Windows provides a single, case-insensitive namespace for all objects and alsospecifies a prefix to target an Active Directory domain (for example, domain\name).

u Kerberos and NFSv4 define principals, which require names to be formatted the sameway as email addresses (for example, [email protected]).

Multiple names can reference the same object. For example, given the name "support"and the domain "example.com", support, EXAMPLE\support, and [email protected] all names for a single object in Active Directory.

Access tokensAn access token is created when the user first makes a request for access.

Access tokens represent who a user is when performing actions on the cluster and supplythe primary owner and group identities to use during file creation. Access tokens are alsocompared against the ACL or mode bits during authorization checks.

During user authorization, OneFS compares the access token, which is generated duringthe initial connection, with the authorization data on the file. All user and identitymapping occurs during token generation; no mapping takes place during permissionsevaluation.

An access token includes all UIDs, GIDs, and SIDs for an identity, in addition to all OneFSprivileges. OneFS exclusively uses the information in the token to determine whether auser has access to a resource. It is important that the token contains the correct list ofUIDs, GIDs, and SIDs.

An access token is created from one of the following sources:

Source Authorization method- -Username SMB impersonate user

Kerberized NFSv3

Kerberized NFSv4

mountd root mapping

HTTP

FTP

Privilege Attribute Certificate (PAC) SMB NTLM

Active Directory Kerberos

Identity management

Access tokens 123

Source Authorization method- -User identifier (UID) NFS AUTH_SYS mapping

Access token generationFor most protocols, the access token is generated from the username or from theauthorization data retrieved during authentication.

The process of token generation and user mapping is described below:

1. Using the initial identity, the user is looked up in all configured authenticationproviders in the access zone, in the order in which they are listed, until a match isfound. An exception to this behavior occurs if the AD provider is configured to callother providers, such as LDAP or NIS. The user identity and group list are retrievedfrom the authenticating provider. Any SIDs, UIDs, or GIDs are added to the initialtoken.

2. All identities in the token are queried in the ID mapper. All SIDs are converted to theirequivalent UID/GID and vice versa. These ID mappings are also added to the accesstoken.

3. If the username matches any user mapping rules, the rules are processed in orderand the token is updated accordingly. (For details about user mapping rules, see"User mapping.")

The default on-disk identity is calculated using the final token and the global setting.These identities are used for newly created files.

ID mappingThe ID mapping service maps Windows SIDs to UNIX UIDs and, conversely, to controlaccess consistently across protocols. Administrators with advanced knowledge of UNIXand Windows identities can modify the default settings that determine how identities aremapped in the system.

Note

Identity (ID) mapping and user mapping are different services, despite the similarity innames.

During authentication, the ID mapping service associates Windows identifiers with UNIXidentifiers. When a user connects to a cluster over NFS, the ID mapping service maps theuser’s UID and GID to a SID for access to files that another user stored over SMB. In thesame way, when a user connects to the cluster over SMB with a SID, the service maps itto a UID and GID for access to files stored over NFS by a UNIX client. By default, the IDmapping service matches accounts with the same name.

Mappings are stored in a cluster-distributed database called the ID mapper. Whenretrieving a mapping from the database, the ID mapper takes a source and target identitytype as input. If a mapping already exists between the specified source and therequested type, that mapping is returned; otherwise, a new mapping is created. Eachmapping is stored in the ID mapper database as a one-way relationship from the sourceto the target identity type. Two-way mappings are stored as complementary one-waymappings.

Identity management


Mapping Windows IDs to UNIX IDsIf a caller requests a SID-to-UID or SID-to-GID mapping, OneFS must first locate the ActiveDirectory user or group that is associated with the SID.

After locating the Active Directory user or group, OneFS applies the following rules in theorder listed to create two mappings, one in each direction:

1. If the object has an associated UID or GID through an external mapping, create amapping from the SID.

2. If a mapping for the SID already exists in the ID mapper database, use that mapping.

3. Determine whether a lookup of the user or group is necessary in an external source,according to the following conditions:

l The user or group is in the primary domain or one of the listed lookup domains.

l Lookup is enabled for users or groups.

4. If a lookup is necessary, perform these steps:

a. By default, normalize the user or group name to lowercase.

b. Search all authentication providers except Active Directory for a matching user orgroup object by name.

c. If an object is found, use the associated UID or GID to create an external mapping.

5. Allocate an automatic mapping from the configured range.

Mapping UNIX IDs to Windows IDsOneFS creates temporary UID-to-SID and GID-to-SID mappings only if the caller requests amapping that does not already exist. The “UNIX SIDs” that result from these mappingsare never stored on disk.

UIDs and GIDs have a set of pre-defined mappings to and from SIDs.

If a UID-to-SID or GID-to-SID mapping is requested, a temporary UNIX SID is generated inthe format S-1-22-1-<UID> or S-1-22-2-<GID> by applying the following rules:

u For UIDs, generate a UNIX SID with a domain of S-1-22-1 and a resource ID (RID)matching the UID. For example, the UNIX SID for UID 600 is S-1-22-1-600.

u For GIDs, generate a UNIX SID with a domain of S-1-22-2 and a RID matching theGID. For example, the UNIX SID for GID 800 is S-1-22-2-800.

User mapping across systems and identitiesUser mapping provides a way to control permissions by specifying a user's securityidentifiers, user identifiers, and group identifiers. OneFS uses the identifiers to check fileor group ownership.

With the user mapping service, you can apply rules to modify which user identity OneFSuses, add supplemental user identities, and modify a user's group membership. TheOneFS user mapper provides a way to control the permissions given to users byspecifying user and group identifiers (SIDs, UIDs, and GIDs) for a user. The user mappingservice combines a user’s identities from different directory services into a single accesstoken and then modifies it according to the rules that you set.

Identity management

User mapping across systems and identities 125

Note

You can configure mapping rules when you create an access zone. OneFS maps usersonly during login or protocol access.

If you do not configure rules, a user who authenticates with one directory service receivesfull access to the identity information in other directory services when the account namesare the same. For example, a user who authenticates with an Active Directory domain asDesktop\jane automatically receives permissions for the corresponding UNIX useraccount for jane from LDAP or NIS.

In the most common scenario, OneFS is connected to two directory services, ActiveDirectory and LDAP. In such a case, the default mapping provides a user with a UID fromLDAP and a SID from the default group in Active Directory. The user's groups come fromActive Directory and LDAP, with the LDAP groups added to the list. To pull groups fromLDAP, the mapping service queries the memberUid. The user’s home directory, gecos,and shell come from Active Directory.

User mapping guidelinesYou can create and configure user mapping rules in each access zone, following theseguidelines.

By default, every mapping rule is processed. This behavior allows multiple rules to beapplied, but can present problems when applying a “deny all" rule such as "deny allunknown users." Additionally, replacement rules may interact with rules that containwildcard characters.

To minimize complexity when configuring multiple mapping rules, it is recommended thatyou group rules by type and organize them in the following order:

1. Replacements: Any user renaming should be processed first to ensure that allinstances of the name are replaced.

2. Joins: After the names are set by any replacement operations, use join, add, andinsert rules to add extra identifiers.

3. Allow/deny: All processing must be stopped before a default deny rule can beapplied. To do this, create a rule that matches allowed users but does nothing (suchas an add operator with no field options) and has the break option. After enumeratingthe allowed users, a catchall deny may be placed at the end to replace anybodyunmatched with an empty user.

Within each group of rules, put explicit rules before rules involving wildcard characters;otherwise, the explicit rules might be skipped.

Elements of user-mapping rulesYou combine operators with user names to create a user-mapping rule.

The following elements affect how the user mapper applies a rule:

u The operator, which determines the operation that a rule performs

u Fields for usernames

u Options

u A parameter

u Wildcards

Identity management


Mapping rule operatorsThe operator determines what a mapping rule does.

You can create user-mapping rules through either the web-administration interface,where the operators are spelled out in a list, or from the command-line interface.

When you create a mapping rule with the OneFS command-line interface (CLI), you mustspecify an operator with a symbol. The operator affects the direction in which themapping service processes a rule. For more information about creating a mapping rule,see the white paper Managing identities with the Isilon OneFS user mapping service. Thefollowing table describes the operators that you can use in a mapping rule.

A rule can contain only one operator.

Operator Web interface CLI Direction Description- - - - -append Append fields

from a user++ Left-to-right Modifies an access token by adding fields to

it. The mapping service appends the fieldsthat are specified in the list of options (user,group, groups) to the first identity in therule. The fields are copied from the secondidentity in the rule. All appended identifiersbecome members of the additional groupslist. An append rule without an optionperforms only a lookup operation; you mustinclude an option to alter a token.

insert Insert fieldsfrom a user

+= Left-to-right Modifies an existing access token by addingfields to it. Fields specified in the optionslist (user, group, groups) are copied fromthe new identity and inserted into theidentity in the token. When the rule inserts aprimary user or primary group, it become thenew primary user and primary group in thetoken. The previous primary user andprimary group move to the additionalidentifiers list. Modifying the primary userleaves the token’s username unchanged.When inserting the additional groups froman identity, the service adds the new groupsto the existing groups.

replace Replace oneuser with adifferent user

=> Left-to-right Removes the token and replaces it with thenew token that is identified by the secondusername. If the second username is empty,the mapping service removes the firstusername in the token, leaving nousername. If a token contains no username,OneFS denies access with a no suchuser error.

removegroups

Removesupplementalgroups from auser

-- Unary Modifies a token by removing thesupplemental groups.

Identity management


Operator Web interface CLI Direction Description- - - - -join Join two users

together&= Bidirectional Inserts the new identity into the token. If the

new identity is the second user, themapping service inserts it after the existingidentity; otherwise, the service inserts itbefore the existing identity. The location ofthe insertion point is relevant when theexisting identity is already the first in the listbecause OneFS uses the first identity todetermine the ownership of new file systemobjects.

Mapping rule optionsMapping rules can contain options that target the fields of an access token.

A field represents an aspect of a cross-domain access token, such as the primary UID andprimary user SID from a user that you select. You can see some of the fields in the OneFSweb adminstration interface. User in the web administration interface is the same asusername. You can also see fields in an access token by running the command isiauth mapping token. For more information about running this command, see Viewon-disk identity.

When you create a rule, you can add an option to manipulate how OneFS combinesaspects of two identities into a single token. For example, an option can force OneFS toappend the supplement groups to a token.

A token includes the following fields that you can manipulate with user mapping rules:

u username

u unix_name

u primary_uid

u primary_user_sid

u primary_gid

u primary_group_sid

u additional_ids (includes supplemental groups)

Options control how a rule combines identity information in a token. The break option isthe exception: It stops OneFS from processing additional rules.

Although several options can apply to a rule, not all options apply to all operators. Thefollowing table describes the effect of each option and the operators that they work with.

Option Operator Description- - -user insert, append Copies the primary UID

and primary user SID, ifthey exist, to the token.

groups insert, append Copies the primary GIDand primary group SID, ifthey exist, to the token.

groups insert, append Copies all the additionalidentifiers to the token.(The additional identifiers

Identity management


Option Operator Description- - -

exclude the primary UID,the primary GID, theprimary user SID, and theprimary group SID.)

default_user all operators except remove groups If the mapping servicefails to find the seconduser in a rule, the servicetries to find the usernameof the default user. Thename of the default usercannot include wildcards.When you set the optionfor the default user in arule with the command-line interface, you mustset it with an underscore:default_user.

break all operators Stops the mapping servicefrom applying rules thatfollow the insertion pointof the break option. Themapping servicegenerates the final tokenat the point of the break.

User-mapping best practicesYou can follow best practices to simplify user mapping.

Best practice Comments- -Use Active Directory withRFC 2307 and WindowsServices for UNIX

Use Microsoft Active Directory with Windows Services for UNIX andRFC 2307 attributes to manage Linux, UNIX, and Windows systems.Integrating UNIX and Linux systems with Active Directory centralizesidentity management and eases interoperability, reducing the needfor user mapping rules. Make sure your domain controllers arerunning Windows Server 2003 or later.

Employ a consistentusername strategy

The simplest configurations name users consistently, so that eachUNIX user corresponds to a similarly named Windows user. Such aconvention allows rules with wildcards to match names and mapthem without explicitly specifying each pair of accounts.

Do not use overlapping IDranges

In networks with multiple identity sources, such as LDAP and ActiveDirectory with RFC 2307 attributes, you should ensure that UID andGID ranges do not overlap. It is also important that the range fromwhich OneFS automatically allocates UIDs and GIDs does not overlapwith any other ID range. The range from which OneFS automaticallyallocates a UID and GID is 1,000,000 to 2,000,000. If UIDs and GIDsoverlap across two or more directory services, some users might gainaccess to other users’ directories and files.

Identity management


Best practice Comments- -Avoid common UIDs andGIDs

You should not use well-known UIDs and GIDs in your ID rangesbecause they are reserved for system accounts. UIDs and GIDs below1000 are reserved for system accounts; do not assign them to usersor groups.

Do not use user principalnames in mapping rules

You cannot use a user principal name(UPN) in a user mapping rule. Auser principal name is an Active Directory domain and username thatare combined into an Internet-style name with an @ symbol, like anemail address: [email protected]. If you include a UPN in a rule,the mapping service ignores it and may return an error.

Group rules by type andorder them

The system processes every mapping rule by default, which canpresent problems when you apply a rule to deny all unknown usersaccess. In addition, replacement rules may interact with rules thatcontain wildcard characters. To minimize complexity, it isrecommended that you group rules by type and organize them in thefollowing order:

1. Place the rules that replace an identity first to ensure that OneFSreplaces all instances of the identity.

2. Set join, add, and insert rules second.

3. Set rules that allow or deny access last.

4. Within each group of rules, put explicit rules before rules withwildcards; otherwise, the explicit rules might be skipped.

Add the LDAP or NISprimary group to thesupplemental groups

When an Isilon cluster is connected to Active Directory and LDAP, abest practice is to add the LDAP primary group to the list ofsupplemental groups. This lets OneFS honor group permissions onfiles created over NFS or migrated from other UNIX storage systems.The same practice is advised when an Isilon cluster is connected toboth Active Directory and NIS.

On-disk identityAfter the user mapper resolves a user's identities, OneFS determines an authoritativeidentifier for it, which is the preferred on-disk identity.

OnesFS stores either UNIX or Windows identities in file metadata on disk. On-disk identitytypes are UNIX, SID, and native. Identities are set when a file is created or a file's accesscontrol data is modified. Almost all protocols require some level of mapping to operatecorrectly, so choosing the preferred identity to store on disk is important. You canconfigure OneFS to store either the UNIX or the Windows identity, or you can allow OneFSto determine the optimal identity to store.

On-disk identity types are UNIX, SID, and native. Although you can change the type of on-disk identity, the native identity is best for a network with UNIX and Windows systems. Innative mode, setting the UID as the on-disk identity improves NFS performance.

Note

When you upgrade from a version of OneFS that is older than 7.0, the on-disk identity isset to UNIX. The SID on-disk identity is for a homogeneous network of Windows systemsmanaged only with Active Directory. On new installations, the on-disk identity is set tonative.

Identity management


The native on-disk identity type allows the OneFS authentication daemon to select thecorrect identity to store on disk by checking for the identity mapping types in thefollowing order:

Order Mappingtype

Description

- - -1 Algorithmic

mappingA SID that matches S-1-22-1-UID or S-1-22-2-GID in the internal IDmapping database is converted back to the corresponding UNIX identityand the UID and GID are set as the on-disk identity.

2 Externalmapping

A user with an explicit UID and GID defined in a directory service (such asActive Directory with RFC 2307 attributes, LDAP, NIS, or the OneFS fileprovider or local provider) has the UNIX identity set as the on-diskidentity.

3 Persistentmapping

Mappings are stored persistently in the identity mapper database. Anidentity with a persistent mapping in the identity mapper database usesthe destination of that mapping as the on-disk identity, which occursprimarily with manual ID mappings. For example, if there is an IDmapping of GID:10000 to S-1-5-32-545, a request for the on-disk storageof GID:10000 returns S-1-5-32-545.

4 No mapping If a user lacks a UID or GID even after querying the other directoryservices and identity databases, its SID is set as the on-disk identity. Inaddition, to make sure a user can access files over NFS, OneFS allocatesa UID and GID from a preset range of 1,000,000 to 2,000,000. In nativemode, a UID or GID that OneFS generates is never set as the on-diskidentity.

Note

If you change the on-disk identity, you should run the repairpermissions job.

Identity management

On-disk identity 131

CHAPTER 7

Auditing


u Auditing overview................................................................................................134u Protocol audit events.......................................................................................... 134u Supported event types........................................................................................ 134u Supported audit tools......................................................................................... 135u Enable system configuration auditing..................................................................136u Enable protocol access auditing..........................................................................136u Auditing settings................................................................................................. 137u Integrating with the EMC Common Event Enabler.................................................137

Auditing 133

Auditing overviewYou can audit system configuration and SMB protocol activity on the Isilon cluster. Allaudit data is stored and protected in the cluster file system and organized in files calledaudit topics.

You can export SMB audit data to Varonis DatAdvantage or other third-party vendors thatsupport the EMC Common Event Enabler (CEE) framework. You can view systemconfiguration activity on each node through a command-line tool.

You can enable or disable system configuration auditing; no additional configuration isrequired. If you enable configuration auditing, all configuration events that are handledby the API—including writes, modifications, and deletions—are tracked and recorded inthe config audit topic.

You can enable and configure protocol auditing for one or more access zones in the Isiloncluster. If you enable protocol auditing for an access zone, file-access events through theSMB protocol are recorded in the protocol audit topic. The protocol audit topic isconsumable by auditing applications that support the EMC Common Event Enabler (CEE),such as Varonis DatAdvantage for Windows. By default, OneFS logs only the events thatare handled by Varonis, but you can specify which events to log in each access zone. Forexample, you might want to audit the default set of protocol events in the System accesszone but audit only successful attempts to delete files in a different access zone.

Protocol audit eventsBy default, audited access zones track only events that are used by VaronisDatAdvantage—including successful and failed attempts to access files or directories.

Although recent versions of Varonis DatAdvantage do not directly audit read and writeattempts, the intention to read or write is captured by the access bits for a create event.

The names of generated events are loosely based on the Windows I/O request packet(IRP) model in which all operations begin with a create event to obtain a file handle. Alldelete, rename, read, write, or set_security events must be preceded by a create event. Aclose event marks when the client is finished with the file handle that was produced by acreate event.

These internally-stored events are translated to events that are forwarded through CEE toVaronis DatAdvantage. The CEE export facilities on OneFS perform this mapping.

It is important to note that different SMB clients issue different requests, and that oneversion of Windows or Mac OS X using SMB may differ from another. It is also importantto note that different versions of an application such as Microsoft Word or Explorer mightmake very different SMB protocol requests. For example, a client with a MicrosoftExplorer window open might occasionally generate many events if an automatic ormanual refresh of that window occurs. Because programs issue requests with the logged-in user's credentials, examining file system audit events under the assumption that theywere purposeful user activities may yield unexpected results.

Supported event typesYou can view or modify the event types that are audited in an access zone. By default,OneFS audits only the event types that are supported by Varonis DatAdvantage.

The following event types are configured by default on each audited access zone:

Auditing


Event name Example protocol activity- -create Create a file or directory

Open a file, directory, or share

Mount a share

Delete a file

rename Rename a file or directory

delete Delete a file or directory

set_security Attempt to modify file or directory permissions

The following event types are available for forwarding through CEE but are unsupportedby Varonis DatAdvantage:

Event name Example protocol activity- -read The first read request on an open file handle

write The first write request on an open file handle

close The client is finished with an open file handle

get_security The client reads security information for an open file handle

The following protocol audit events are not exported through CEE and are unsupported byVaronis DatAdvantage:

Event name Example protocol activity- -logon SMB session create request by a client

logoff SMB session logoff

tree_connect SMB first attempt to access a share

Supported audit toolsYou can configure OneFS to send protocol auditing logs to servers that support the EMCCommon Event Enabler (CEE).

CEE has been tested and verified to work with the following applications.

Note

It is recommended that you install and configure third-party auditing applications beforeyou enable the OneFS auditing feature. Otherwise, the backlog consumed by the tool maybe so large that results may be stale for a prolonged time.

Application Supported features Audit events- - -Varonis DatAdvantage for Windows Usable Access Auditing

Recommendations, Analytics, and Modeling

Data Owner Identification and Involvement

create

delete

rename

Auditing

Supported audit tools 135

Application Supported features Audit events- - -

set_security

Enable system configuration auditingYou can enable or disable the auditing of system configuration changes. No additionalsettings are available.

Note


Procedure

1. Click Cluster Management > Auditing.

2. In the Settings area, select the Enable Configuration Change Auditing checkbox.

3. Click Save Changes.

Enable protocol access auditingYou can audit SMB protocol access on a per-access zone basis and optionally forward thegenerated events to the EMC Common Event Enabler (CEE) for export to third-partyproducts.

Note


Procedure

1. Click Cluster Management > Auditing.

2. In the Settings area, select the Enable Protocol Access Auditing checkbox.

3. In the Audited Zones area, click Add Zones.

4. In the Select Zones dialog box, select the checkbox for one or more access zones, andthen click Add Zones.

5. Optional: In the Event Forwarding area, specify one or more CEE servers to forwardlogged events to.

a. In the CEE Server URIs field, type the URI of each CEE server in the CEE server pool.

The OneFS CEE export service uses round robin load-balancing when exportingevents to multiple CEE servers. Valid URIs start with http:// and include the portnumber and path to the CEE server if necessary.

http://example.com:12228/ceeb. In the Storage Cluster Name field, specify the name of the storage cluster to use

when forwarding protocol events.

Auditing


This value is typically the SmartConnect zone name. This setting is required only ifneeded by your third-party audit application.


Results

The following protocol events, which are the only events supported by VaronisDatAdvantage, are collected for audited access zones by default: create, delete,rename, and set_security. You can modify the set of events that are audited in anaccess zone by running the isi zone zones modify command in the command-lineinterface. Because each audited event consumes system resources, it is recommendedthat you only configure zones for events that are needed by your auditing application.

Auditing settingsYou can view or modify basic settings for configuration change auditing and protocolaccess auditing.

Enable Configuration Change AuditingAudits requests that are made through the API for system configuration changes.

Enable Protocol Access AuditingAudits requests that are made through the SMB protocol to access data.

Audited ZonesSpecifies one or more access zones to audit. This setting applies only to protocolaccess auditing.

CEE Server URIsSpecifies one or more CEE server URIs where audit events will be forwarded. TheOneFS CEE export service uses round robin load-balancing when exporting events tomultiple CEE servers. This setting applies only to protocol access auditing.

Storage Cluster NameSpecifies the name of the storage cluster to use when forwarding protocol events—typically, the SmartConnect zone name. This setting is required only if needed byyour third-party audit application.

Integrating with the EMC Common Event EnablerOneFS integration with the EMC Common Event Enabler (CEE) allows third-party auditingapplications such as Varonis DatAdvantage to collect and analyze SMB protocol auditinglogs.

OneFS supports the Common Event Publishing Agent (CEPA) component of CEE forWindows. For integration with OneFS, you must install and configure CEE for Windows ona supported Windows client.

Note


Auditing

Auditing settings 137

Install CEE for WindowsTo integrate CEE with OneFS, you must first install CEE on a computer that is running theWindows operating system.

Before you begin

Be prepared to extract files from the .iso file, described in the steps below. If you arenot familiar with the process, consider choosing one of the following methods:

u Install WinRAR or another suitable archival program that can open .iso files as anarchive, and copy the files.

u Burn the image to a CD-ROM, and then copy the files.

u Install SlySoft Virtual CloneDrive, which allows you to mount an ISO image as a drivethat you can copy files from.

Note

You should install a minimum of two servers.

Procedure

1. Download the CEE framework software from EMC Online Support:

a. In a web browser, go to https://support.emc.com/search/.

b. In the Search Support field, type Common Event Enabler for Windows, andthen click the Search icon.

c. Click Common Event Enabler <Version> for Windows, where <Version> is 6.0.0 orlater, and then follow the instructions to open or save the .iso file.

2. From the .iso file, extract the 32-bit or 64-bit EMC_CEE_Pack executable file thatyou need.

After the extraction completes, the EMC Common Event Enabler installation wizardopens.

3. Click Next to proceed to the License Agreement page.

4. Select the I accept... option to accept the terms of the license agreement, and thenclick Next.

5. On the Customer Information page, type your user name and organization, select yourinstallation preference, and then click Next.

6. On the Setup Type page, select Complete, and then click Next.

7. Click Install to begin the installation.

The Installing EMC Common Event Enabler page displays the progress of theinstallation. When the installation is complete, the InstallShield Wizard Completedpage appears.

8. Click Finish to exit the wizard.

9. Restart the system.

Auditing


Configure CEE for WindowsAfter you install CEE for Windows on a client computer, you must configure additionalsettings through the Windows Registry Editor (regedit.exe).

Procedure

1. Open the Windows Registry Editor.

2. Configure the following registry keys:

Setting Registry location Key Value- - - -CEE HTTPlisten port

[HKEY_LOCAL_MACHINE\SOFTWARE\EMC\CEE\Configuration]

HttpPort 12228

Enableauditremoteendpoints

[HKEY_LOCAL_MACHINE\SOFTWARE\EMC\CEE\CEPP\Audit\Configuration]

Enabled 1

Auditremoteendpoints

[HKEY_LOCAL_MACHINE\SOFTWARE\EMC\CEE\CEPP\Audit\Configuration]

EndPoint <EndPoint>

Note

l The HttpPort value must match the port in the CEE URIs that you specify duringOneFS protocol audit configuration.

l The EndPoint value must be in the format <EndPoint_Name>@<IP_Address>. You canspecify multiple endpoints by separating each value with a semicolon (;).

Specify a single remote endpoint:[HKEY_LOCAL_MACHINE\SOFTWARE\EMC\CEE\CEPP\Audit\Configuration] EndPoint [email protected] multiple remote endpoints:[HKEY_LOCAL_MACHINE\SOFTWARE\EMC\CEE\CEPP\Audit\Configuration] EndPoint [email protected];[email protected]

3. Close the Windows Registry Editor.

Auditing

Configure CEE for Windows 139

CHAPTER 8

File sharing


u File sharing overview...........................................................................................142u SMB.................................................................................................................... 142u NFS..................................................................................................................... 143u HTTP and HTTPS.................................................................................................. 143u FTP...................................................................................................................... 144u Mixed protocol environments.............................................................................. 144u Write caching with SmartCache........................................................................... 144u Managing the SMB service.................................................................................. 146u Managing SMB shares.........................................................................................148u Managing the NFS service................................................................................... 152u Managing NFS exports.........................................................................................156u Enable and configure FTP file sharing.................................................................. 159u Enable and configure HTTP.................................................................................. 159u Home directories.................................................................................................160

File sharing 141

File sharing overviewMulti-protocol support is built into the OneFS operating system, enabling a single file ordirectory to be accessed through SMB for Windows file sharing, NFS for UNIX file sharing,secure shell (SSH), FTP, and HTTP. By default, only the SMB and NFS protocols areenabled.

OneFS creates the /ifs directory, which is the root directory for all file system data onthe cluster. The /ifs directory is configured as an SMB share and an NFS export bydefault. You can create additional shares and exports within the /ifs directory tree.

Note

It is recommended that you do not save data to the root /ifs file path but in directories

below /ifs. The design of your data storage structure should be planned carefully. A

well-designed directory optimizes cluster performance and cluster administration.

You can set Windows- and UNIX-based permissions on OneFS files and directories. Userswho have the required permissions and administrative privileges can create, modify, andread data on the cluster through one or more of the supported file sharing protocols.

u SMB. Allows Microsoft Windows and Mac OS X clients to access files that are storedon the cluster.

u NFS. Allows UNIX, Linux, Mac OS X, Solaris, and other UNIX-based clients to accessfiles that are stored on the cluster.

u HTTP and HTTPS (with optional DAV). Allows clients to access files that are stored onthe cluster through a web browser.

u FTP. Allows any client that is equipped with an FTP client program to access files thatare stored on the cluster through the FTP protocol.

SMBOneFS includes a configurable SMB service to create and manage SMB shares. SMBshares provide Windows clients network access to file system resources on the cluster.You can grant permissions to users and groups to carry out operations such as reading,writing, and setting access permissions on SMB shares.

The /ifs directory is configured as an SMB share and is enabled by default. OneFSsupports both user and anonymous security modes. If the user security mode is enabled,users who connect to a share from an SMB client must provide a valid user name withproper credentials.

The SMB protocol uses security identifiers (SIDs) for authorization data. All identities areconverted to SIDs during retrieval and are converted back to their on-disk representationbefore they are stored on the cluster.

When a file or directory is created, OneFS checks the access control list (ACL) of its parentdirectory. If the ACL contains any inheritable access control entries (ACEs), a new ACL isgenerated from those ACEs. Otherwise, OneFS creates an ACL from the combined file anddirectory create mask and create mode settings.

OneFS supports the following SMB clients:

File sharing


SMB version Supported operating systems- -1 Windows 2000 or later

Windows XP or later

Mac OS X 10.5 or later

2 Windows Vista or later

Windows Server 2008 or later

2.1 Windows 7 or later

Windows Server 2008 R2 or later

Overlapping display names for SMB sharesOneFS supports overlapping display names for SMB shares if the display name appearsonly once per access zone.

All SMB shares belong to a global list of shares and require unique SMB share names. Bydefault, users see the SMB share name when connecting to the EMC Isilon cluster;however, you can configure a display name for the SMB share that users see instead.

Display names must be unique within a zone; therefore, if you would like more than oneSMB share to display the same name you must add each share to a separate accesszone. For example, you can assign the "Home" as the display name for an SMB share inzoneA and also assign it to a different share in zoneB.

NFSOneFS includes a configurable NFS service to create and manage NFS exports, whichprovide UNIX clients network access to file system resources on the cluster.

The Isilon cluster supports NIS and LDAP authentication providers for NFS file sharing.OneFS supports asynchronous and synchronous communication over NFS.

HTTP and HTTPSOneFS includes a configurable HTTP service, which is used to request files that are storedon the cluster and to interact with the web administration interface.

OneFS supports both HTTP and its secure variant, HTTPS. Each node in the cluster runs aninstance of the Apache HTTP Server to provide HTTP access. You can configure the HTTPservice to run in different modes.

Both HTTP and HTTPS are supported for file transfer, but only HTTPS is supported forPlatform API calls. The HTTPS-only requirement includes the web administration interface.In addition, OneFS supports a form of the web-based DAV (WebDAV) protocol thatenables users to modify and manage files on remote web servers. OneFS performsdistributed authoring, but does not support versioning and does not perform securitychecks. You can enable DAV in the web administration interface.

File sharing

Overlapping display names for SMB shares 143

FTPOneFS includes a secure FTP service called vsftpd, which stands for Very Secure FTPDaemon, that you can configure for standard FTP and FTPS file transfers.

Mixed protocol environmentsThe /ifs directory is the root directory for all file system data in the cluster, serving asan SMB share, an NFS export, and a document root directory. You can create additionalshares and exports within the /ifs directory tree. You can configure your OneFS clusterto use SMB or NFS exclusively. You can also enable HTTP, FTP, and SSH.

Access rights are consistently enforced across access protocols on all security models. Auser is granted or denied the same rights to a file whether using SMB or NFS. Clustersrunning OneFS support a set of global policy settings that enable you to customize thedefault access control list (ACL) and UNIX permissions settings.

OneFS is configured with standard UNIX permissions on the file tree. Through WindowsExplorer or OneFS administrative tools, you can give any file or directory an ACL. Inaddition to Windows domain users and groups, ACLs in OneFS can include local, NIS, andLDAP users and groups. After a file is given an ACL, the mode bits are no longer enforcedand exist only as an estimate of the effective permissions.

Note

It is recommended that you configure ACL and UNIX permissions only if you fullyunderstand how they interact with one another.

Write caching with SmartCacheWrite caching accelerates the process of writing data to the cluster. OneFS includes awrite-caching feature called SmartCache, which is enabled by default for all files anddirectories.

If write caching is enabled, OneFS writes data to a write-back cache instead ofimmediately writing the data to disk. OneFS can write the data to disk at a time that ismore convenient.

Note

It is recommended that you keep write caching enabled. You should also enable writecaching for all file pool policies.

OneFS interprets writes to the cluster as either synchronous or asynchronous, dependingon a client's specifications. The impacts and risks of write caching depend on whatprotocols clients use to write to the cluster, and whether the writes are interpreted assynchronous or asynchronous. If you disable write caching, client specifications areignored and all writes are performed synchronously.

The following table explains how clients' specifications are interpreted, according to theprotocol.

File sharing


Protocol Synchronous Asynchronous- - -NFS The stable field is set to data_sync or

file_sync.

The stable field is set to unstable.

SMB The write-through flag has been

applied.

The write-through flag has not been

applied.

iSCSI The write-cache enabled (WCE) setting isset to false.

The WCE setting is set to true.

Write caching for asynchronous writesWriting to the cluster asynchronously with write caching is the fastest method of writingdata to your cluster.

Write caching for asynchronous writes requires fewer cluster resources than write cachingfor synchronous writes, and will improve overall cluster performance for most workflows.However, there is some risk of data loss with asynchronous writes.

The following table describes the risk of data loss for each protocol when write cachingfor asynchronous writes is enabled:

Protocol Risk- -NFS If a node fails, no data will be lost except in the unlikely event that a client of that

node also crashes before it can reconnect to the cluster. In that situation,asynchronous writes that have not been committed to disk will be lost.

SMB If a node fails, asynchronous writes that have not been committed to disk will be lost.

iSCSICAUTION

If a node fails, asynchronous writes that have not been committed can causeinconsistencies in any file system that is laid out on the LUN, rendering the filesystem unusable.

It is recommended that you do not disable write caching, regardless of the protocol thatyou are writing with. If you are writing to the cluster with asynchronous writes, and youdecide that the risks of data loss are too great, it is recommended that you configure yourclients to use synchronous writes, rather than disable write caching.

Write caching for synchronous writesWrite caching for synchronous writes costs cluster resources, including a negligibleamount of storage space. Although it is not as fast as write caching with asynchronouswrites, unless cluster resources are extremely limited, write caching with synchronouswrites is faster than writing to the cluster without write caching.

Write caching does not affect the integrity of synchronous writes; if a cluster or a nodefails, none of the data in the write-back cache for synchronous writes is lost.

File sharing

Write caching for asynchronous writes 145

Managing the SMB serviceYou can enable or disable the SMB service and configure global settings for the SMBservice, including default settings that are applied to new SMB shares.

Configure SMB file sharingGlobal configuration settings for SMB include snapshot directory settings and SMB sharesettings.

The global SMB share settings are the same as the settings for individual SMB shares. Tochange the advanced settings for an individual share, click SMB Shares.

CAUTION

Modifying the advanced settings could result in operational failures. Be aware of thepotential consequences before committing changes to these settings.

Procedure

1. Click Protocols > Windows Sharing (SMB) > SMB Settings.

2. For the SMB service setting, select Enabled.

3. To configure advanced SMB server settings, click SMB Server Settings.

4. To configure advanced SMB share settings, click SMB Share Settings.

5. Click Save.

Limit access to /ifs share for the Everyone accountYou should limit SMB share access on the root /ifs directory for the Everyone account.

It is a best practice to limit the Everyone account to read-only access on the root /ifsshare.

Procedure

1. Click Protocols > Windows Sharing (SMB) > SMB Shares, and then click View detailsfor the /ifs share.

2. Click Edit next to the Users & Groups option.

3. In the User/Group accounts list, click Edit next to the Everyone account.

4. Click Specify Permission Level and then select from the following options to definethe permissions that you want to allow the Everyone account:

l Full Control

l Read-Write

l Read

5. Click Save.

File sharing


Snapshots directory settingsYou can view and configure the settings that control the snapshots directories in SMB.

CAUTION

These settings affect the behavior of the SMB service. Changes to these settings canaffect all current and future SMB shares.

Setting Setting value- -Visible at Root Specifies whether to make the .snapshot directory visible at the root

of the share. The default value is Yes.

Accessible at Root Specifies whether to make the .snapshot directory accessible at theroot of the share. The default value is Yes.

Visible in Subdirectories Specifies whether to make the .snapshot directory visible insubdirectories of the share root. The default value is No.

Accessible inSubdirectories

Specifies whether to make the .snapshot directory accessible insubdirectories of the share root. The default value is Yes.

File and directory permission settingsYou can view and configure the default source permissions and UNIX create mask/modebits that are applied when a file or directory is created in an SMB share.

Note

Changes that are made from the SMB Settings tab override the default settings for allSMB shares.

If the mask and mode bits match the default values, a green check mark next to a settingappears, indicating that the specified read (R), write (W), or execute (X) permission isenabled at the user, group, or "other" level. The "other" level includes all users who arenot listed as the owner of the share, and are not part of the group level that the filebelongs to.

Setting Setting value- -Create Permissions Sets the default source permissions to apply when a file or directory is

created. The default value is Default ACL.

Create Mask (Dir) Specifies UNIX mode bits that are removed when a directory is created,restricting permissions. Mask bits are applied before mode bits are applied.

Create Mode (Dir) Specifies UNIX mode bits that are added when a directory is created,enabling permissions. Mode bits are applied after mask bits are applied.

Create Mask (File) Specifies UNIX mode bits that are removed when a file is created, restrictingpermissions. Mask bits are applied before mode bits are applied.

Create Mode (File) Specifies UNIX mode bits that are added when a file is created, enablingpermissions. Mode bits are applied after mask bits are applied.

File sharing

Snapshots directory settings 147

SMB performance settingsYou can view and configure the change notify and oplocks performance settings of anSMB share.

Note


Setting Setting value- -Change Notify Configures notification of clients when files or directories change. This helps

prevent clients from seeing stale content, but requires server resources. Thedefault value is Norecurse.

Oplocks Indicates whether an opportunistic lock (oplock) request is allowed. An oplockallows clients to provide performance improvements by using locally-cachedinformation. The default value is Yes.

SMB security settingsYou can view and configure the Impersonate Guest, Impersonate User, and NTFS ACLsecurity settings of an SMB share.

Note


Setting Setting value- -Impersonate Guest Determines guest access to a share. The default value is Never.

Impersonate User Allows all file access to be performed as a specific user. This must be a fullyqualified user name. The default value is No value.

NTFS ACL Allows ACLs to be stored and edited from SMB clients. The default value isYes.

Managing SMB sharesYou can configure the rules and other settings that govern the interaction between yourWindows network and individual SMB shares on the cluster.

OneFS supports %U, %D, %Z, %L, %0, %1, %2, and %3 variable expansion andautomatic provisioning of user home directories.

You can configure the users and groups that are associated with an SMB share, and viewor modify their share-level permissions.

File sharing


Note

It is recommended that you configure advanced SMB share settings only if you have asolid understanding of the SMB protocol.

Create an SMB shareWhen you create an SMB share, you can override the default permissions, performance,and access settings. You can configure SMB home directory provisioning by includingexpansion variables in the share path to automatically create and redirect users to theirown home directories.

Procedure

1. Click Protocols > Windows Sharing (SMB) > SMB Shares.

2. Click Add a share.

3. In the Share Name field, type a name for the share.

Share names can contain up to 80 characters, and can only contain alphanumericcharacters, hyphens, and spaces.

4. Optional: In the Description field, type a comment about the share.

A description is optional, but can be helpful if you are managing multiple shares. Thisfield is limited to 255 characters.

5. In the Directory to be Shared field, type the full path of the share, beginning with /ifs, or click Browse to locate the share.

Note

You can specify one or more of the following variables in the directory path but youmust select the Allow Variable Expansion check box or the string is interpretedliterally by the system.

Variable Expansion- -%D NetBIOS domain name.

%U User name—for example, user_001.

%Z Zone name—for example, System.

%L Host name of the cluster, normalized to lowercase.

%0 First character of the user name.

%1 Second character of the user name.

%2 Third character of the user name.

For example, if a user is in a domain named DOMAIN and has a username of user_1,the path /ifs/home/%D/%U expands to /ifs/home/DOMAIN/user_1.

6. Apply the initial ACL settings for the directory. You can modify these settings later.

l To apply a default ACL to the shared directory, click Apply Windows default ACLs.

File sharing

Create an SMB share 149

Note

If the Auto-Create Directories setting is enabled, OneFS creates an ACL with theequivalent of UNIX 700 mode bit permissions for any directory that is createdautomatically.

l To maintain the existing permissions on the shared directory, click Do not changeexisting permissions.

7. Optional: Configure home directory provisioning settings.

l To expand path variables such as %U in the share directory path, select AllowVariable Expansion.

l To automatically create home directories when users access the share for the firsttime, select Auto-Create Directories. This option is available only if the AllowVariable Expansion option is enabled.

8. Optional: Apply advanced SMB share settings if needed.

9. Click Create.

After you finish

Note

SMB shares are currently created with read-only permissions by default. To enable accessto a share, you must modify the share settings to allow users to write to the share. Thisfunctionality is available only through the OneFS command-line interface. For example,the following command allows the well-known user Everyone full permissions to a sharenamed HOMEDIR:isi smb shares permission modify HOMEDIR --wellknown Everyone \--permission-type allow --permission full

Modify SMB share permissions, performance, or securityYou can modify the permissions, performance, and access settings for individual SMBshares.

You can configure SMB home directory provisioning by using directory path, orexpansion, variables to automatically create and redirect users to their own homedirectories.

Note

Any changes made to these settings will only affect the settings for this share. If you needto make changes to the global default values, that can be done from the SMB Settingstab.

Procedure


2. From the list of SMB shares, locate the share you want to modify and then click Viewdetails.


4. To modify the settings for file and directory permissions, performance, or security,click Advanced SMB Share Settings.

File sharing


Add a user or group to an SMB shareFor each SMB share, you can add share-level permissions for specific users and groups.

Procedure

1. Click Protocols > Windows Sharing (SMB) > SMB Shares, and then click View detailsfor the share you want to add a user or group to.

2. Click Edit next to the Users & Groups option.

The User/Group permission list for the share appears.

3. Click Add a User or Group. Then select the option you want to search for.

Options Description

Users Enter the username you want to search for in the text field, andthen click Search.

Groups Enter the group you want to search for in the text field, and thenclick Search.

Well-known SIDs Skip to step 5.

4. From the Access Zone list, select the access zone you want to search.

5. From the Provider list, select the authentication provider you want to search. Onlyproviders that are currently configured and enabled on the cluster are listed.

6. Click Search.

The results of the search appear in the Search Results box.

7. In the search results, click the user, group, or SID that you want to add to the SMBshare and then click Select.

8. By default, the access rights of the new account are set to "Deny All". To enable a useror group to access the share, follow these additional steps:

a. Next to the user or group account you added, click Edit.

b. Select the permission level you want to assign to the user or group. The choicesare Run as Root or specific permission levels: Full Control, Read-Write, or Read.

9. Click Save.

Configure overlapping share display namesYou can specify the same display name for multiple SMB shares by assigning the sharesto separate access zones.

Procedure

1. Click Cluster Management > Access Management > Access Zone.

2. From the Access Zones list, click View details for the zone that you want to modify.

3. In the SMB Shares area, click Edit.

4. In the drop-down menu, select Manually select SMB shares.

The SMB Shares area expands to display the SMB Shares in this Zone list.

5. Click Add SMB shares.

The Select SMB Shares dialog box appears.

File sharing

Add a user or group to an SMB share 151

6. Select the SMB share you want to add to this zone and click Select.

7. In the SMB Shares in this Zone list, click Edit to modify the share.

8. Type the overlapping display name you want in the Display Name field and click Save.

9. In the SMB Shares area, click Save.

10.Repeat this procedure for each SMB share that will use the same display name.

After you finish

You must associate an IP address pool with each access zone. The IP address the userconnects through specifies which zone the user is allowed to access.

Configure multi-protocol home directory accessFor users who will access this share through FTP or SSH, you can ensure that their homedirectory path is the same whether they connect through SMB or they log in through FTPor SSH.

This command directs the SMB share to use the home directory template that is specifiedin the user's authentication provider. This procedure is available only through thecommand-line interface.

Procedure


2. Run the following command, where <homedir_share> is the name of the SMB share:isi smb share modify <homedir_share> --path=""

Delete an SMB shareYou can delete SMB shares that are no longer needed.

Unused SMB shares do not hinder cluster performance. If you delete an SMB share, theshare path is deleted but the directory it referenced still exists. If you create a new sharewith the same path as the share that was deleted, the directory that the previous sharereferenced will be accessible again through the new share.

Procedure


2. From the list of SMB shares, select the share that you want to delete.

Note

You can delete all of the shares on the cluster by selecting the Name/Path option, andthen selecting Delete from the drop-down menu.

3. Click Delete.

4. In the confirmation dialog box, click Delete to confirm the deletion.

Managing the NFS serviceYou can enable or disable the NFS service and specify the NFS versions to support. NFSsettings are applied across all nodes in the cluster.

Configure NFS file sharingYou can enable or disable the NFS service, set the lock protection level, and set thesecurity type. These settings are applied across all nodes in the cluster. You can change

File sharing


the settings for individual NFS exports as you create them, or edit the settings forindividual exports as needed.

Procedure

1. Click Protocols > UNIX Sharing (NFS) > NFS Settings.

2. Enable or disable the NFS service and version support settings:

l NFS Service

l NFSv2 Support

l NFSv3 Support

l NFSv4 Support

3. Select the Lock Protection Level setting.

4. Click the Reload Cached Configuration button.

The cached NFS export settings are reloaded to ensure that changes to DNS or NIS areapplied.

5. In the Users/Groups Mapping menu, click Custom Default. A box containing thesettings for Map to User Credentials and Also map these user groups appears.

a. To limit access by mapping root users or all users to a specific user or group, fromthe Root users list, click Specific username and then type the user names in thetext field. A user is any user available in one of the configured authorizationproviders.

b. To map users to groups, select the Also map these users to groups check box,click Specific user group(s), and then type the group names in the text field.

6. Select the security type.

The default setting is UNIX.

7. Click Save.

Create a root-squashing rule for the default NFS exportYou should create a root-squashing rule for the default NFS export.

Procedure


2. From the list of NFS exports, click View details for the default export.

3. In the Users/Group Mapping menu, click Use Custom. A box containing the settingsfor Map to User Credentials and Also map these user groups appears.

Option Description

Click Specific username and then type the usernames iselect the Also map these users to groupscheck box, click Specific user group(s), and thentype the group names in the text field.

Limit access by mapping rootusers or all users to a specificuser or group, from the Rootusers list.

Select the Also map these users to groups checkbox, click Specific user group(s), and then type thegroup names in the text field.

Map users to groups.

4. Select the security type.

The default setting is UNIX.

File sharing

Create a root-squashing rule for the default NFS export 153

5. Click Save.

NFS service settingsThe NFS service settings are the global settings that determine how the NFS file sharingservice operates. These settings include versions of NFS to support, the lock protectionlevel, NFS exports configuration, user/group mappings, and security types.

Setting Description- -Service Enables or disables the NFS service. This setting is enabled by default.

NFSv2 support Enables or disables support for NFSv2. This setting is enabled by default.

NFSv3 support Enables or disables support for NFSv3. This setting is enabled by default.

NFSv4 support Enables or disables support for NFSv4. This setting is disabled by default.

Lock protection level Determines the number of node failures that can happen before a lock maybe lost. The default value is +2.

NFS performance settingsThe NFS performance settings are global settings that affect the performance of NFSexports, such as maximum file size and read/write transfer settings.

Setting Description- -Block size The block size reported to NFSv2+ clients. The default value is 8192.

Commit asynchronously If set to yes, allows NFSv3 and NFSv4 COMMIT operations to beasynchronous. The default value is No.

Directory read transfer The preferred directory read transfer size reported to NFSv3 and NFSv4clients. The default value is 131072.

Read transfer max The maximum read transfer size reported to NFSv3 and NFSv4 clients.The default value is 1048576.

Read transfer multiple The recommended read transfer size multiple reported to NFSv3 andNFSv4 clients. The default value is 512.

Read transfer preferred The preferred read transfer size reported to NFSv3 and NFSv4 clients.

Readdirplus prefetch The number of file nodes to be prefetched on readdir. The default valueis 10.

Setattr asynchronous If set to yes, performs set attribute operations asynchronously. The

default value is No.

Write datasync action The action to perform for DATASYNC writes. The default value isDATASYNC.

Write datasync reply The reply to send for DATASYNC writes. The default value is DATASYNC.

Write filesync action The action to perform for FILESYNC writes. The default value isFILESYNC.

File sharing


Setting Description- -Write filesync reply The reply to send for FILESYNC writes. The default value is FILESYNC.

Write transfer max The maximum write transfer size reported to NFSv3 and NFSv4 clients.The default value is 1048576.

Write transfer multiple The recommended write transfer size reported to NFSv3 and NFSv4clients. The default value is 512.

Write transfer preferred The preferred write transfer size reported to NFSv3 and NFSv4 clients.The default value is 524288.

Write unstable action The action to perform for UNSTABLE writes. The default value isUNSTABLE.

Write unstable reply The reply to send for UNSTABLE writes. The default value is UNSTABLE.

NFS client compatibility settingsThe NFS client compatibility settings are global settings that affect the customization ofNFS exports. These settings include the maximum file size, enabling readdirplus, and 32-bit file IDs.

Setting Setting value- -Max file size Specifies the maximum file size to allow. The default value is

9223372036854776000.

Readdirplus enable Enables readdirplus. The default value is yes.

Return 32 bit file IDs Returns 32-bit file IDs.

NFS export behavior settingsThe NFS export behavior settings are global settings that control options such as whethernon-root users can set file times, the general encoding settings of an export, whether tolook up UIDs (incoming user identifiers), or set the server clock granularity.

Setting Setting value- -Can set time Permits non-root users to set file times. The default value is Yes.

Encoding Overrides the general encoding settings the cluster has for the export. Thedefault value is DEFAULT.

Map Lookup UID Looks up incoming user identifiers (UIDs) in the local authentication database.The default value is No.

Symlinks Enables symlink support for the export. The default value is Yes.

Time delta Sets the server clock granularity. The default value is 1e-9.

File sharing

NFS client compatibility settings 155

Managing NFS exportsThe default /ifs export is configured to allow UNIX clients to mount any subdirectory.You can view and modify NFS export settings, and you can delete NFS exports that are nolonger needed.

You must define all mount points for a given export host as a single export rule, which isthe collection of options and constraints that govern the access of an export to the filesystem. To add a mount point for an export host that appears in the list of export rules,you must modify that entry rather than add a new one.

You can apply individual host rules to each export, or you can specify all hosts, whicheliminates the need to create multiple rules for the same host. To prevent problems whensetting up new exports, be sure to delete export rules for directories that have beenremoved from the file system.

Note

Changes to the advanced settings affect all current and future NFS exports that usedefault settings, and may impact the availability of the NFS file sharing service. Do notmake changes to these settings unless you have experience working with NFS. It isrecommended that you change the default values for individual NFS exports as you createthem, or edit the settings of existing exports.

View and configure default NFS export settingsDefault settings apply to all current and future NFS exports. For each setting, you cancustomize the default value or select the factory default value. The factory default valuecannot be modified.

Note

Modifying the global default values is not recommended. You can override the settingsfor NFS exports as you create them, or modify the settings for existing exports.

Procedure


2. Click the NFS export settings menu.

3. For each setting that you want to modify, click System Default in the list of optionsand select Custom Default.

Note

If a confirmation dialog box appears, click Continue.

4. Make your changes to the information in the setting value text field.

5. When you are finished modifying settings, click Save.

Create an NFS exportOneFS does not restrict the number of NFS exports that you can create.

Procedure

1. Click Protocols > UNIX Sharing (NFS) > NFS Export.

File sharing


2. Click Add an Export.

3. Optional: In the Description field, type a comment that describes the export.

4. Optional: Specify which clients are allowed to access the export.

You can specify a client by host name, IP address, subnet, or netgroup. You canspecify multiple clients in each field by typing one entry per line.

Note

If no clients are specified, all clients are allowed to access the export. If you add thesame client to more than one list and the client is entered in the same format for eachentry, the client is normalized to a single list in the following order of priority: RootClients, Always Read-Write Clients, Always Read-Only Clients,

Clients.

Setting Description

Clients Specifies one or more clients to be allowed access to the export.Access level is controlled through export permissions.

AlwaysRead-WriteClients

Specifies one or more clients to be allowed read/write access to theexport regardless of the export's access-restriction setting. Equivalentto adding a client to the Clients list with the Restrict access to read-only setting cleared.

AlwaysRead-OnlyClients

Specifies one or more clients to be allowed read-only access to theexport regardless of the export's access-restriction setting. Equivalentto adding a client to the Clients list with the Restrict access to read-only setting selected.

Root Clients Specifies one or more clients to be mapped as root for the export.This setting is equivalent to adding a client to the Clients list andmapping root users to the root username.

5. For the Directory Paths setting, type or browse to the directory that you want to export.

You can add more directory paths by clicking Add another directory path.

6. Specify export permissions:

l Restrict actions to read-only.

l Enable mount access to subdirectories. Allow subdirectories below the path(s) tobe mounted.

7. Specify User/Group mapping.

If you select the Use custom option, you can limit access by mapping root users or allusers to a specific user and group ID. For root squash, map root users to the username nobody.

8. Specify Security Type(s).

If you select the Use custom option, you can select one or more of the followingsecurity types:

l UNIX (system)

l Kerberos5

l Kerberos5 Integrity

l Kerberos5 Privacy

File sharing

Create an NFS export 157

9. Configure Advanced NFS Export Settings.

10.Click Save.

Modify an NFS exportYou can modify the settings for individual NFS exports.

Procedure

1. Click Protocols > UNIX Sharing (NFS) > NFS Export.

2. From the list of NFS exports, click View details for the export that you want to modify.


4. Click Close.

Delete an NFS exportYou can delete NFS exports that are no longer needed.

Note

You can delete all the exports on a cluster by selecting the Export ID/Path option, andthen selecting Delete from the drop-down menu.

Procedure

1. Click Protocols > UNIX Sharing (NFS) > NFS Export

2. From the list of NFS exports, click the check box for the export that you want to delete.

3. Click Delete.

4. In the confirmation dialog box, click Delete to confirm the deletion.

Check NFS exports for errorsYou can check for errors in NFS exports.

This procedure is available only through the CLI.

Procedure

1. Run the isi nfs exports check command.

In the following example output, no errors were found:ID Message--------------------Total: 0

In the following example output, export 1 contains a path that does not currentlyexist:

ID Message-----------------------------------1 '/ifs/test' is not a directory-----------------------------------Total: 1

File sharing


Enable and configure FTP file sharingYou can set the FTP service to allow any node in the cluster to respond to FTP requeststhrough a standard user account.

You can enable the transfer of files between remote FTP servers and enable anonymousFTP service on the root by creating a local user named anonymous or ftp.

When configuring FTP access, make sure that the specified FTP root is the home directoryof the user who logs in. For example, the FTP root for local user jsmith should be ifs/home/jsmith.

Procedure

1. Click Protocols > FTP Settings.

2. Click Enable.

3. Select one or more of the following settings:

Option Description

Server-to-servertransfers

Enables the transfer of files between two remote FTP servers.This setting is disabled by default.

Anonymousaccess

Enables users with "anonymous" or "ftp" as the user name toaccess files and directories without requiring authentication.This setting is disabled by default.

Local access Enables local users to access files and directories with theirlocal user name and password, allowing them to upload filesdirectly through the file system. This setting is enabled bydefault.

4. Click Submit.

Enable and configure HTTPYou can configure HTTP and DAV to enable users to edit and manage files collaborativelyacross remote web servers.

This procedure is available only through the web administration interface.

Procedure

1. Click Protocols > HTTP Settings.

2. From the Service options, select one of the following settings:

Option Description

Enable HTTP Allows HTTP access for cluster administration and browsingcontent on the cluster.

Disable HTTP andredirect to the webinterface

Allows only administrative access to the web administrationinterface. This is the default setting.

File sharing

Enable and configure FTP file sharing 159

Option Description

Disable HTTP entirely Closes the HTTP port used for file access. Users can continueto access the web administration interface by specifying theport number in the URL. The default port is 8080.

3. In the Document root directory field, type or click Browse to navigate to an existingdirectory in /ifs, or click File System Explorer to create a new directory and set itspermissions.

Note

The HTTP server runs as the daemon user and group. To properly enforce accesscontrols, you must grant the daemon user or group read access to all files under thedocument root, and allow the HTTP server to traverse the document root.

4. In the Server hostname field, type the HTTP server name. The server hostname mustbe a fully-qualified, SmartConnect zone name and valid DNS name. The name mustbegin with a letter and contain only letters, numbers, and hyphens (-).

5. In the Administrator email address field, type an email address to display as theprimary contact for issues that occur while serving files.

6. From the Active Directory Authentication list, select an authentication setting:

Option Description

Off Disables HTTP authentication.

Basic Authentication Only Enables HTTP basic authentication. User credentialsare sent in plain text.

Integrated AuthenticationOnly

Enables HTTP authentication via NTLM, Kerberos, orboth.

Integrated and BasicAuthentication

Enables both basic and integrated authentication.

Basic Authentication withAccess Controls

Enables HTTP authentication via NTLM andKerberos, and enables the Apache web server toperform access checks.

Integrated and Basic Authwith Access Controls

Enables HTTP basic authentication and integratedauthentication, and enables access checks via theApache web server.

7. Click the Enable DAV check box. This allows multiple users to manage and modify filescollaboratively across remote web servers.

8. Click the Disable access logging check box.

9. Click Submit.

Home directoriesWhen you create a local user, OneFS automatically creates a home directory for the user.OneFS also supports dynamic home directory provisioning for users who access thecluster by connecting to an SMB share or by logging in through FTP or SSH. Regardless of

File sharing


the method by which a home directory was created, you can configure access to thehome directory through a combination of SMB, SSH, and FTP.

Home directory permissionsA user's home directory can be set up with a Windows ACL or with POSIX mode bits,which are then converted into a synthetic ACL. The method by which a home directory iscreated determines the initial permissions that are set on the home directory.

When you create a local user, the user's home directory is created with mode bits bydefault.

For users who authenticate against external sources, home directories can bedynamically created at login time. If a home directory is created during a login throughSSH or FTP, it is set up with mode bits; if a home directory is created during an SMBconnection, it receives either mode bits or an ACL. For example, if an LDAP user first logsin through SSH or FTP, the user's home directory is created with mode bits. However, ifthe same user first connects through an SMB share, the home directory is created withthe permissions indicated by the configured SMB settings. If the "inherited path ACL"setting is enabled, an ACL is generated; otherwise, mode bits are used.

Note

Because SMB sends an NT password hash to authenticate SMB users, only users fromauthentication providers that can handle NT hashes can log in over SMB. These providersinclude the local provider, Active Directory, and LDAP with Samba extensions enabled.File, NIS, and non-Samba LDAP users cannot log in over SMB.

Home directory creation through SMBYou can create an SMB share that includes expansion variables in the share path,enabling users to access their home directories by connecting to the share. You canenable dynamic creation of home directories that do not exist at SMB connection time.

By default, an SMB share's directory path is created with a synthetic ACL based on modebits. You can enable the "inheritable ACL" setting on a share to specify that, if the parentdirectory has an inheritable ACL, it will be inherited on the share path.

Note

Share permissions are checked when files are accessed, before the underlying filesystem permissions are checked. Either of these permissions can prevent access to thefile or directory.

Create home directories with expansion variablesYou can create SMB share home directories by using expansion variables, asdemonstrated in this example.

When using SMB, home directories are accessed through shares. A special homedirectory share can be set up with a path that uses a special variable expansion syntax toallow a user to connect to his or her home directory by connecting to this share.

Note

Home directory share paths must begin with /ifs/ and must be within the root path of

the access zone in which the home directory SMB share is created.

File sharing

Home directory permissions 161

In the following example, on the cluster the --allow-variable-expansion optionis enabled to indicate that %U should be expanded to the user name (user411 in thisexample). The --auto-create-directory option is enabled to indicate that thedirectory should be created if it does not exist. When user411 connects to the share withthe net use command, the user's home directory is created at /ifs/home/user411.# isi smb shares create HOMEDIR --path=/ifs/home/%U \ --allow-variable-expansion=yes --auto-create-directory=yes# isi smb shares permission modify HOMEDIR --wellknown Everyone \ --permission-type allow --permission full# isi smb shares view HOMEDIR Share Name: HOMEDIR Path: /ifs/home/%U Description: Client-side Caching Policy: manualAutomatically expand user names or domain names: TrueAutomatically create home directories for users: True Browsable: TruePermissions:Account Account Type Run as Root Permission Type Permission------------------------------------------------------------Everyone wellknown False allow full ------------------------------------------------------------Total: 1...On the user's Windows client, m: is connected to /ifs/home/user411 through theHOMEDIR share.# net use m: \\cluster.company.com\HOMEDIR /u:user411

Create home directories with the --inheritable-path-ACL optionYou can enable the inheritable ACL option on a share to specify that it is to beinherited on the share path if the parent directory has an inheritable ACL.

By default, an SMB share's directory path is created with a synthetic ACL based on modebits. You can enable the --inheritable-path-ACL option to use the inheritable ACLon all directories that are created, either at share creation time or for those dynamicallyprovisioned when connecting to that share.

In this example, the --inheritable-path-ACL is enabled on the cluster todynamically provision a user home directory at first connection to a share on the cluster:# isi SMB shares create HOMEDIR_ACL --path=/ifs/home/%U \ --allow-variable-expansion=yes --auto-create-directory=yes \ --inheritable-path-ACL=yes# isi SMB shares permission modify HOMEDIR_ACL \ --wellknown Everyone \ --permission-type allow --permission fullOn the user's Windows client:# net use q: \\cluster.company.com\HOMEDIR_ACL /u:user411The result on the cluster:# cd /ifs/home/user411# ls -lde .drwx------ + 2 user411 <your-company> Users 0 Oct 19 16:23 ./ OWNER: user:user411 GROUP: group:<your-company> Users CONTROL:dacl_auto_inherited,dacl_protected 0: user:user411 allow dir_gen_all,object_inherit,container_inherit

File sharing


Create special home directories with the SMB share %U variableYou can used the SMB share %U variable to create a home directory SMB share thatmaps to a user name.

When you use an SMB share %U variable so that it maps to a user name, it is typicallywith a share path that includes the %U expansion variable. When a user attempts toconnect to a share matching the login name and it does not exist, he or she connects tothe %U share instead and is directed to the expanded path for the %U share.

The following example creates a share that matches the authenticated user login name.For example, user Zachary will connect to /ifs/home/zachary on the cluster when he triesto connect to share zachary in the System zone.# isi smb share create %U /ifs/home/%U \ --allow-variable-expansion=yes --auto-create-directory=yes \ --zone=System In this example from a Windows client, when the net use command is run on m:, Zacharysees the contents of his /ifs/home/zachary directory:# net use m: \\cluster.ip\zachary /u:zachary # cd m: # dir In this example, another user, Claudia, sees the directory contents of /ifs/home/claudia:# net use m: \\cluster.ip\claudia /u:claudia # cd m: # dir If Claudia tries to access Zachary's share, however, she cannot connect because it doesnot exist for her.

Note

If another SMB share exists that matches the user's name, then the user connects to theexplicitly named share rather than the %U share.

Authenticating SMB usersYou can authenticate SMB users from authentication providers that can handle NThashes.

SMB sends an NT password hash to authenticate SMB users, so only users fromauthentication providers that can handle NT hashes can log in over SMB. The followingOneFS-supported authentication providers can handle NT hashes:

u Active Directory

u Local

u LDAPSAM (LDAP with Samba extensions enabled)

Note

File, NIS, and non-SAM LDAP providers cannot handle NT hashes, so users from thoseauthentication providers cannot log in over SMB.

Home directory creation through SSH and FTPFor users who access the cluster through SSH or FTP, you can configure home directorysupport by modifying authentication provider settings.

The following authentication provider settings determine how home directories are setup.

File sharing

Home directory creation through SSH and FTP 163

Home Directory NamingSpecifies the path to use as a template for naming home directories. The path mustbegin with /ifs and may contain variables, such as %U, that are expanded togenerate the home directory path for the user.

Create home directories on first loginSpecifies whether to create a home directory the first time a user logs in, if a homedirectory does not already exist for the user.

UNIX ShellSpecifies the path to the user's login shell. This setting applies only to users whoaccess the file system through SSH.

Set SSH/FTP home directory creation optionsYou can configure home directory support for a user who accesses the cluster throughSSH or FTP by specifying authentication provider options.

The following authentication provider options affect home directory creation:

u Create Home Directory. A boolean value that indicates whether to create thehome directory if it does not exist.

u Home Directory Template. The template path name for the user's homedirectory, and may contain special variables beginning with '%' that are expanded togenerate the home directory path for the user. The path name must begin with /ifs/.

u Login Shell. The default login shell for the user. The user's login shell may alsobe provided by the authentication provider.

Note

A user must have the ISI_PRIV_LOGIN_SSH privilege to log in to a node through SSH.

The following example demonstrates setting these options for an Active Directoryauthentication provider.# isi auth ads listName Authentication Status DC Name Site ---------------------------------------------------------YOUR.DOMAIN.NAME.COM Yes online - SEA---------------------------------------------------------Total: 1

# isi auth ads modify YOUR.DOMAIN.NAME.COM \ --home-directory-template=/ifs/home/ADS/%D/%U \ --create-home-directory=yes# isi auth ads view -v YOUR.DOMAIN.NAME.COM Name: YOUR.DOMAIN.NAME.COM NetBIOS Domain: YOUR ... Create Home Directory: Yes Home Directory Template: /ifs/home/ADS/%D/%U Login Shell: /bin/sh

# id YOUR\\user_100uid=1000008(<your-domain>\user_100) gid=1000000(<your-domain>\domain users) \groups=1000000(<your-domain>\domain users),1000024(<your-domain>\c1t),1545(Users)The information is verified from an external Unix node, as seen in the result of thefollowing command:# ssh <your-domain>\\[email protected]

File sharing


After logging in, the user above will be in directory /ifs/home/ADS/<your-domain>/user_100, which will be created if it did not previously exist.

Set the SSH or FTP login shellYou can use the Login Shell option to set the default login shell for the user.

By default, the Login Shell option, if provided, overrides any login-shell informationprovided by the authentication provider, except with Active Directory, in which case itsimply represents the default login shell if the Active Directory server does not providelogin-shell information.

In the following example, the login shell for all local users is set to /bin/bash# isi auth local modify System --login-shell=/bin/bash This example sets the default login shell for all Active Directory users in <your-domain>to /bin/bash:# isi auth ads modify YOUR.DOMAIN.NAME.COM --login-shell=/bin/bash

Set SSH/FTP home directory permissionsYou can specify home directory permissions for a home directory accessed through SSHor FTP by setting the Home Directory Umask option.

When a user's home directory is created at login through SSH or FTP, it is created usingPOSIX mode bits. The permissions setting on a user's home directory is affected by theHome Directory Umask setting of the user's authentication zone.

This example shows how the Home Directory Umask setting can be seen:# isi zone zones view System...Home Directory Umask: 0077You can modify the umask option for a zone with the --home-directory-umask option,specifying an octal number as the umask.The following example demonstrates how to allow a group/others write/executepermission in a home directory. In this example, the user's home directory is created withmode bits 0755 masked by the umask field, set to the value of 022. So by default, auser's home directory is created with mode bits 0700 (equivalent to (0755 & ~(077)) :# isi zone zones modify <zone-name> --home-directory-umask=022

Provision home directories with dot filesYou can provision home directories with dot files.

The skeleton directory, which is located at /usr/share/skel by default, contains a setof files that are copied to the user's home directory when a local user is created or when auser home directory is dynamically created during login. Files in the skeleton directorythat begin with dot. are renamed to remove the dot prefix when they are copied to theuser's home directory. For example, dot.cshrc is copied to the user's home directoryas .cshrc. This format enables dot files in the skeleton directory to be viewable throughthe command-line interface without requiring the ls -a command.

For SMB shares that might use home directories that were provisioned with dot files, youcan set an option to prevent users who connect to the share through SMB from viewingthe dot files.

The following command is used to determine the default skeleton directory zone:

# isi zone zones view System Name: System... Skeleton Directory: /usr/share/skel

File sharing

Home directory creation through SSH and FTP 165

The default skeleton directory, /usr/share/skel, can be modified in anauthentication zone with the --skeleton-directory=<path> option, as shown inthis example:

# isi zone zones modify System --skeleton-directory=/usr/share/skel2# isi zone zones view System Name: System... Skeleton Directory: /usr/share/skel2

Home directory creation in a mixed environmentIf a user will log in through both SMB and SSH, it is recommended that you set up thehome directory so that the path template is the same in the SMB share and eachauthentication provider against which the user is authenticating through SSH.

As an alternative to setting the home directory path in each authentication provider thatmight be used, you can configure an SMB share to use the path that is specified in theuser's authentication provider by setting the share path as an empty string ("").

Interactions between ACLs and mode bitsHome directory set up is determined by several factors, including how users authenticateand the options that specify home directory creation.

A user's home directory may be set up with either ACLs or POSIX mode bits, which areconverted into a synthetic ACL. The directory of a local user is created when the local useris created, and is set up with POSIX mode bits by default. Directories can be dynamicallyprovisioned at log in for users who authenticate against external sources, and in somecases users who authenticate against the File provider. In this case, their home directoryis created according to how they first log in.

For example, if an LDAP user first logs in through SSH or FTP and his or her home directoryis created, it is created with POSIX mode bits. If that same user first connects through anSMB home directory share, the home directory is created as dictated by the SMB optionsettings. If --inherited-path-acl=yes, ACLs are generated. Otherwise, POSIX modebits are used.

Interactions with dot-file provisioningA user's home directory may or may not contain dot files, depending on the method offirst access to the cluster.

If a home directory is first created through SSH access, it has a configuration file namethat begins with a '.' (for example, .login or .cshrc); if the home directory is createdthrough SMB, it may not. If you are planning an SMB share that uses home directoriesthat have been created either as local users or through home directory creation over SSHor FTP, you may want to consider the --hide-dot-files=yes option, which can be set so thatSMB users do not see any files that begin with a "."

If a user can log in to the cluster through both SMB and SSH or FTP and the user's homedirectory is dynamically created on first access, then the user's dot files are set up only ifSSH or FTP is used to first access the cluster. If SMB is used for first access, the dot filesare not copied, but they can be copied manually from the path set in the SkeletonDirectory field of the user's access zone.

The following command modifies an SMB share HOMEDIR to hide any files that beginwith ".":# isi smb modify share HOMEDIR --hide-dot-files=yes

File sharing


This command reveals the path set in the Skeleton Directory field of the user'saccess zone:isi zone zones view <ZONE-NAME>" | grep Skeleton

Default home directory settings in authentication providersThe default settings that affect how home directories are set up differ based on theauthentication provider that the user authenticates against.

Authenticationprovider

Home directory naming Home directorycreation

UNIX loginshell

- - - -Local /ifs/home/%U Enabled /bin/sh

File None Disabled None

Active Directory /ifs/home/%D/%U

Note

If available, providerinformation overrides thisvalue.

Disabled /bin/sh

LDAP None Disabled None

NIS None Disabled None

Supported expansion variablesYou can include expansion variables in an SMB share path or in an authenticationprovider's home directory template.

OneFS supports the following expansion variables. You can improve performance andreduce the number of shares to be managed when you configure shares with expansionvariables. For example, you can use the %U variable for a share rather than create a sharefor each user. When a %U is used in the name so that each user's path is different,security is still ensured because each user can view and access only his or her homedirectory.

Note

When you create an SMB share through the web administration interface, you must selectthe Allow Variable Expansion check box or the string is interpreted literally by the system.

Variable Value Description- - -%U User name (for example,

user_001)Expands to the user name to allow different usersto use different home directories. This variable istypically included at the end of the path. Forexample, for a user named user1, the path /ifs/home/%U is mapped to /ifs/home/user1.

%D NetBIOS domain name (forexample, YORK forYORK.EAST.EXAMPLE.COM)

Expands to the user's domain name, based on theauthentication provider:

File sharing

Default home directory settings in authentication providers 167

Variable Value Description- - -

l For Active Directory users, %D expands to theActive Directory NetBIOS name.

l For local users, %D expands to the clustername in uppercase characters. For example,for a cluster named cluster1, %D expands toCLUSTER1.

l For users in the System file provider, %Dexpands to UNIX_USERS.

l For users in other file providers, %D expandsto FILE_USERS.

l For LDAP users, %D expands to LDAP_USERS.

l For NIS users, %D expands to NIS_USERS.

%Z Zone name (for example,ZoneABC)

Expands to the access zone name. If multiplezones are activated, this variable is useful fordifferentiating users in separate zones. Forexample, for a user named user1 in the Systemzone, the path /ifs/home/%Z/%U is mapped

to /ifs/home/System/user1.

%L Host name (cluster host name inlowercase)

Expands to the host name of the cluster,normalized to lowercase. Limited use.

%0 First character of the user name Expands to the first character of the user name.

%1 Second character of the username

Expands to the second character of the user name.

%2 Third character of the user name Expands to the third character of the user name.

Note

If the user name includes fewer than three characters, the %0, %1, and %2 variableswrap around. For example, for a user named ab, the variable maps to a, b, and a. For auser named a, all three variables map to a.

Domain variables in home directory provisioningYou can use domain variables to specify authentication providers when provisioninghome directories.

The domain variable is typically used for Active Directory users, but it has a value set thatcan be used for other authentication providers. It expands to the following for the variousauthentication providers.

Authenticated user Expands to- -Active Directory User %D = Active Directory NetBIOS name (for example, YORK for provider

YORK.EAST.EXAMPLE.COM)

Local User If cluster is called "cluster-name":%D = CLUSTER-NAME (uppercase)

File sharing


Authenticated user Expands to- -File User %D = UNIX_USERS (for System file provider)

%D = FILE_USERS (for all other file providers)

LDAP User %D = LDAP_USERS (for all LDAP authentication providers)

NIS User %D = NIS_USERS (for all NIS authentication providers)

File sharing

Domain variables in home directory provisioning 169

CHAPTER 9

Snapshots


u Snapshots overview............................................................................................ 172u Data protection with SnapshotIQ.........................................................................172u Snapshot disk-space usage................................................................................ 172u Snapshot schedules............................................................................................173u Snapshot aliases................................................................................................ 173u File and directory restoration...............................................................................173u Snapshot best practices......................................................................................174u Best practices for creating snapshot schedules...................................................174u File clones...........................................................................................................175u Snapshot locks................................................................................................... 176u Snapshot reserve................................................................................................ 177u SnapshotIQ license functionality.........................................................................177u Creating snapshots with SnapshotIQ...................................................................177u Managing snapshots .......................................................................................... 182u Restoring snapshot data..................................................................................... 184u Managing snapshot schedules............................................................................186u Managing with snapshot locks............................................................................ 187u Configure SnapshotIQ settings............................................................................ 189u Set the snapshot reserve.....................................................................................190

Snapshots 171

Snapshots overviewA OneFS snapshot is a logical pointer to data that is stored on a cluster at a specific pointin time.

A snapshot references a directory on a cluster, including all data stored in the directoryand its subdirectories. If the data referenced by a snapshot is modified, the snapshotstores a physical copy of the data that was modified. Snapshots are created according touser specifications or are automatically generated by OneFS to facilitate systemoperations.

To create and manage snapshots, you must activate a SnapshotIQ license on the cluster.Some applications must generate snapshots to function but do not require you toactivate a SnapshotIQ license; by default, these snapshots are automatically deletedwhen OneFS no longer needs them. However, if you activate a SnapshotIQ license, youcan retain these snapshots. You can view snapshots generated by other modules withoutactivating a SnapshotIQ license.

You can identify and locate snapshots by name or ID. A snapshot name is specified by auser and assigned to the virtual directory that contains the snapshot. A snapshot ID is anumerical identifier that OneFS automatically assigns to a snapshot.

Data protection with SnapshotIQYou can create snapshots to protect data with the SnapshotIQ software module.Snapshots protect data against accidental deletion and modification by enabling you torestore deleted and modified files. To use SnapshotIQ, you must activate a SnapshotIQlicense on the cluster.

Snapshots are less costly than backing up your data on a separate physical storagedevice in terms of both time and storage consumption. The time required to move data toanother physical device depends on the amount of data being moved, whereassnapshots are always created almost instantaneously regardless of the amount of datareferenced by the snapshot. Also, because snapshots are available locally, end-users canoften restore their data without requiring assistance from a system administrator.Snapshots require less space than a remote backup because unaltered data isreferenced rather than recreated.

Snapshots do not protect against hardware or file-system issues. Snapshots referencedata that is stored on a cluster, and if the data on the cluster becomes unavailable, thesnapshots will also be unavailable. Because of this, it is recommended that you back upyour data to separate physical devices in addition to creating snapshots.

Snapshot disk-space usageThe amount of disk space that a snapshot consumes depends on both the amount ofdata stored by the snapshot and the amount of data the snapshot references from othersnapshots.

Immediately after OneFS creates a snapshot, the snapshot consumes a negligibleamount of disk space. The snapshot does not consume additional disk space unless thedata referenced by the snapshot is modified. If the data that a snapshot references ismodified, the snapshot stores read-only copies of the original data. A snapshotconsumes only the space that is necessary to restore the contents a directory to the stateit was in when the snapshot was taken.

Snapshots


To reduce disk-space usage, snapshots that reference the same directory reference eachother, with older snapshots referencing newer snapshots. If a file is deleted, and severalsnapshots reference the file, a single snapshot stores a copy the file, and the othersnapshots reference the file from the snapshot that stored the copy. The reported size ofa snapshot reflects only the amount of data stored by the snapshot and does not includethe amount of data referenced by the snapshot.

Because snapshots do not consume a set amount of storage space, there is no available-space requirement for creating a snapshot. The size of a snapshot grows according tohow the data referenced by the snapshot is modified. A cluster cannot contain more than20,000 snapshots.

Snapshot schedulesYou can automatically generate snapshots according to a snapshot schedule.

With snapshot schedules, you can periodically generate snapshots of a directory withouthaving to manually create a snapshot every time. You can also assign an expirationperiod that determines when SnapshotIQ deletes each automatically generatedsnapshot.

Snapshot aliasesA snapshot alias is an optional, alternative name for a snapshot. If a snapshot isassigned an alias, and that alias is later assigned to another snapshot, OneFSautomatically removes the alias from the old snapshot and then assigns the alias to thenew snapshot.

Snapshot aliases are most useful when you specify them in a snapshot schedule. Whenyou specify an alias in a snapshot schedule, SnapshotIQ assigns the alias to the mostrecently generated snapshot, enabling you to quickly identify the most recent snapshotgenerated according to a schedule.

OneFS uses snapshot aliases internally to identify the most recent snapshot generated byOneFS operations.

File and directory restorationYou can restore the files and directories that are referenced by a snapshot by copyingdata from the snapshot, cloning a file from the snapshot, or reverting the entire snapshot.

Copying a file from a snapshot creates an additional copy of the file, which will roughlydouble the amount of storage space consumed. Even if you delete the original file fromthe non-snapshot directory, the copy of the file remains in the snapshot.

Unlike copying a file from a snapshot, which immediately consumes additional space onthe cluster, cloning a file from a snapshot does not consume any additional space on thecluster unless the clone or cloned file is modified.

Reverting a snapshot replaces the contents of a directory with the data stored in thesnapshot. Before a snapshot is reverted, SnapshotIQ creates a snapshot of the directorythat is being replaced, which enables you to undo the snapshot revert later. Reverting asnapshot can be useful if you want to undo a large number of changes that you made tofiles and directories. If new files or directories have been created in a directory since asnapshot of the directory was created, those files and directories are deleted when thesnapshot is reverted.

Snapshots

Snapshot schedules 173

Snapshot best practicesConsider the following snapshot best practices when working with a large number ofsnapshots.

It is recommended that you do not create more than 1,000 snapshots of a single directoryto avoid performance degradation. If you create a snapshot of a root directory, thatsnapshot counts towards the total number of snapshots for any subdirectories of the rootdirectory. For example, if you create 500 snapshots of /ifs/data and 500 snapshotsof /ifs/data/media, you have created 1000 snapshots of /ifs/data/media.Avoid creating snapshots of directories that are already referenced by other snapshots.

It is recommended that you do not create more than 1000 hard links per file in asnapshot to avoid performance degradation. Always attempt to keep directory paths asshallow as possible. The deeper the depth of directories referenced by snapshots, thegreater the performance degradation.

Creating snapshots of directories higher on a directory tree will increase the amount oftime it takes to modify the data referenced by the snapshot and require more clusterresources to manage the snapshot and the directory. However, creating snapshots ofdirectories lower on directories trees will require more snapshot schedules, which can bedifficult to manage. It is recommended that you do not create snapshots of /ifsor /ifs/data.

You can create up to 20,000 snapshots on a cluster at a time. If you create a largenumber of snapshots, you might not be able to manage snapshots through the OneFSweb administration interface. However, you can manage any number of snapshotsthrough the OneFS command-line interface.

Note

It is recommended that you do not disable the snapshot delete job. Disabling thesnapshot delete job prevents unused disk space from being freed and can also causeperformance degradation.

Best practices for creating snapshot schedulesSnapshot schedule configurations can be categorized by how they delete snapshots:ordered deletions and unordered deletions.

An ordered deletion is the deletion of the oldest snapshot of a directory. An unordereddeletion is the deletion of a snapshot that is not the oldest snapshot of a directory.Unordered deletions take approximately twice as long to complete and consume morecluster resources than ordered deletions. However, unordered deletions can save spaceby retaining a smaller total number of snapshots.

The benefits of unordered deletions versus ordered deletions depend on how often thedata referenced by the snapshots is modified. If the data is modified frequently,unordered deletions will save space. However, if data remains unmodified, unordereddeletions will most likely not save space, and it is recommended that you performordered deletions to free cluster resources.

To implement ordered deletions, assign the same duration period for all snapshots of adirectory. The snapshots can be created by one or multiple snapshot schedules. Alwaysensure that no more than 1000 snapshots of a directory are created.

Snapshots


To implement unordered snapshot deletions, create several snapshot schedules for asingle directory, and then assign different snapshot duration periods for each schedule.Ensure that all snapshots are created at the same time when possible.

The following tables describe snapshot schedules that follow snapshot best practices:

Table 13 Snapshot schedule configurations

Deletiontype

Snapshotfrequency

Snapshot time Snapshotexpiration

Max snapshotsretained

- - - - -Ordereddeletion(for mostlystaticdata)

Every hour Beginning at 12:00AM Ending at 11:59AM

1 month 720

Unordereddeletion(forfrequentlymodifieddata)

Every other hour Beginning at 12:00AM Ending at 11:59PM

1 day 27

Every day At 12:00 AM 1 week

Every week Saturday at 12:00AM

1 month

Every month The first Saturday ofthe month at 12:00AM

3 months

File clonesSnapshotIQ enables you to create file clones that share blocks with existing files in orderto save space on the cluster. A file clone usually consumes less space and takes lesstime to create than a file copy. Although you can clone files from snapshots, clones areprimarily used internally by OneFS.

The blocks that are shared between a clone and cloned file are contained in a hidden filecalled a shadow store. Immediately after a clone is created, all data originally containedin the cloned file is transferred to a shadow store. Because both files reference all blocksfrom the shadow store, the two files consume no more space than the original file; theclone does not take up any additional space on the cluster. However, if the cloned file orclone is modified, the file and clone will share only blocks that are common to both ofthem, and the modified, unshared blocks will occupy additional space on the cluster.

Over time, the shared blocks contained in the shadow store might become useless ifneither the file nor clone references the blocks. The cluster routinely deletes blocks thatare no longer needed. You can force the cluster to delete unused blocks at any time byrunning the shadow store delete job.

Clones cannot contain alternate data streams (ADS). If you clone a file that containsalternate data streams, the clone will not contain the alternate data streams.

Snapshots

File clones 175

Shadow store considerationsShadow stores are hidden files that are referenced by cloned and deduplicated files. Filesthat reference shadow stores behave differently than other files.

u Reading shadow-store references might be slower than reading data directly.Specifically, reading non-cached shadow-store references is slower than reading non-cached data. Reading cached shadow-store references takes no more time thanreading cached data.

u When files that reference shadow stores are replicated to another Isilon cluster orbacked up to a Network Data Management Protocol (NDMP) backup device, theshadow stores are not transferred to the target Isilon cluster or backup device. Thefiles are transferred as if they contained the data that they reference from shadowstores. On the target Isilon cluster or backup device, the files consume the sameamount of space as if they had not referenced shadow stores.

u When OneFS creates a shadow store, OneFS assigns the shadow store to a storagepool of a file that references the shadow store. If you delete the storage pool that ashadow store resides on, the shadow store is moved to a pool occupied by anotherfile that references the shadow store.

u OneFS does not delete a shadow store block immediately after the last reference tothe block is deleted. Instead, OneFS waits until the ShadowStoreDelete job is run todelete the unreferenced block. If a large number of unreferenced blocks exist on thecluster, OneFS might report a negative deduplication savings until theShadowStoreDelete job is run.

u Shadow stores are protected at least as much as the most protected file thatreferences it. For example, if one file that references a shadow store resides in astorage pool with +2 protection and another file that references the shadow storeresides in a storage pool with +3 protection, the shadow store is protected at +3.

u Quotas account for files that reference shadow stores as if the files contained thedata referenced from shadow stores; from the perspective of a quota, shadow storereferences do not exist. However, if a quota includes data protection overhead, thequota does not account for the data protection overhead of shadow stores.

iSCSI LUN clonesOneFS enables you to create clones of iSCSI logical units (LUNs) that share blocks withexisting LUNs in order to save space on the cluster. Internally, OneFS creates iSCSI LUNclones by creating file clones.

Snapshot locksA snapshot lock prevents a snapshot from being deleted. If a snapshot has one or morelocks applied to it, the snapshot cannot be deleted and is referred to as a lockedsnapshot. If the duration period of a locked snapshot expires, OneFS will not delete thesnapshot until all locks on the snapshot have been deleted.

OneFS applies snapshot locks to ensure that snapshots generated by OneFS applicationsare not deleted prematurely. For this reason, it is recommended that you do not deletesnapshot locks or modify the duration period of snapshot locks.

A limited number of locks can be applied to a snapshot at a time. If you create snapshotlocks, the limit for a snapshot might be reached, and OneFS could be unable to apply asnapshot lock when necessary. For this reason, it is recommended that you do not createsnapshot locks.

Snapshots


Snapshot reserveThe snapshot reserve enables you to set aside a minimum percentage of the clusterstorage capacity specifically for snapshots. If specified, all other OneFS operations areunable to access the percentage of cluster capacity that is reserved for snapshots.

Note

The snapshot reserve does not limit the amount of space that snapshots can consume onthe cluster. Snapshots can consume a greater percentage of storage capacity specified bythe snapshot reserve. It is recommended that you do not specify a snapshot reserve.

SnapshotIQ license functionalityYou can create snapshots only if you activate a SnapshotIQ license on a cluster. However,you can view snapshots and snapshot locks that are created for internal use by OneFSwithout activating a SnapshotIQ license.

The following table describes what snapshot functionality is available depending onwhether the SnapshotIQ license is active:

Inactive Active- - -Create snapshots andsnapshot schedules

No Yes

Configure SnapshotIQsettings

No Yes

View snapshot schedules Yes Yes

Delete snapshots Yes Yes

Access snapshot data Yes Yes

View snapshots Yes Yes

If you a SnapshotIQ license becomes inactive, you will no longer be able to create newsnapshots, all snapshot schedules will be disabled, and you will not be able to modifysnapshots or snapshot settings. However, you will still be able to delete snapshots andaccess data contained in snapshots.

Creating snapshots with SnapshotIQTo create snapshots, you must configure the SnapshotIQ licence on the cluster. You cancreate snapshots either by creating a snapshot schedule or manually generating anindividual snapshot.

Manual snapshots are useful if you want to create a snapshot immediately, or at a timethat is not specified in a snapshot schedule. For example, if you plan to make changes toyour file system, but are unsure of the consequences, you can capture the current state ofthe file system in a snapshot before you make the change.

Before creating snapshots, consider that reverting a snapshot requires that a SnapRevertdomain exist for the directory that is being reverted. If you intend on reverting snapshotsfor a directory, it is recommended that you create SnapRevert domains for those

Snapshots

Snapshot reserve 177

directories while the directories are empty. Creating a domain for a directory that containsless data takes less time.

Create a SnapRevert domainBefore you can revert a snapshot that contains a directory, you must create a SnapRevertdomain for the directory. It is recommended that you create SnapRevert domains for adirectory while the directory is empty.

The root path of the SnapRevert domain must be the same root path of the snapshot. Forexample, a domain with a root path of /ifs/data/media cannot be used to revert asnapshot with a root path of /ifs/data/media/archive. To revert /ifs/data/media/archive, you must create a SnapRevert domain with a root path of /ifs/data/media/archive.

Procedure

1. Click Cluster Management > Job Operations > Job Types.

2. In the Job Types area, in the DomainMark row, from the Actions column, select StartJob.

3. In the Domain Root Path field, type the path of a snapshot root directory.

4. From the Type of domain list, select SnapRevert.

5. Ensure that the Delete this domain check box is cleared.

6. Click Start Job.

Create a snapshot scheduleYou can create a snapshot schedule to continuously generate snapshots of directories.

Procedure

1. Click Data Protection > SnapshotIQ > Snapshot Schedules.

2. Click Create a snapshot schedule.

3. Optional: In the Create a Snapshot Schedule area, in the Schedule Name field, type aname for the snapshot schedule.

4. Optional: In the Naming pattern for Generated Snapshots field, type a namingpattern. Each snapshot generated according to this schedule is assigned a namebased on the pattern.

For example, the following naming pattern is valid:

WeeklyBackup_%m-%d-%Y_%H:%M

The example produces names similar to the following:

WeeklyBackup_07-13-2012_14:215. In the Directory Path field, specify the directory that you want to be contained in

snapshots that are generated according to this schedule.

6. Specify how often you want to generate snapshots according to the schedule.

Options Description

Generate snapshots every day, or skipgenerating snapshots for a specifiednumber of days.

From the Snapshot Frequency list, selectDaily, and specify how often you want togenerate snapshots.

Snapshots


Options Description

Generate snapshots on specific days ofthe week, and optionally skipgenerating snapshots for a specifiednumber of weeks.

From the Snapshot Frequency list, selectWeekly and specify how often you want togenerate snapshots.

Generate snapshots on specific days ofthe month, and optionally skipgenerating snapshots for a specifiednumber of months.

From the Snapshot Frequency list, selectMonthly and specify how often you wantto generate snapshots.

Generate snapshots on specific days ofthe year.

From the Snapshot Frequency list, selectYearly and specify how often you want togenerate snapshots.

Note

A snapshot schedule cannot span multiple days. For example, you cannot specify tobegin generating snapshots at 5:00 PM Monday and end at 5:00 AM Tuesday. Tocontinuously generate snapshots for a period greater than a day, you must create twosnapshot schedules. For example, to generate snapshots from 5:00 PM Monday to5:00 AM Tuesday, create one schedule that generates snapshots from 5:00 PM to11:59 PM on Monday, and another schedule that generates snapshots from 12:00 AMto 5:00 AM on Tuesday.

7. Optional: To assign an alternative name to the most recent snapshot generated by theschedule, specify a snapshot alias.

a. Next to Create an Alias, click Yes.

b. To modify the default snapshot alias name, in the Alias Name field, type analternative name for the snapshot.

8. Optional: To specify a length of time that snapshots generated according to theschedule exist on the cluster before they are automatically deleted by OneFS, specifyan expiration period.

a. Next to Snapshot Expiration, click Snapshots expire.

b. Next to Snapshots expire, specify how long you want to retain the snapshotsgenerated according to the schedule.

9. Click Create.

Create a snapshotYou can create a snapshot of a directory.

Procedure

1. Click Data Protection > SnapshotIQ > Summary.

2. Click Capture a new snapshot.

3. Optional: In the Capture a Snapshot area, in the Snapshot Name field, type a name.

4. In the Directory Path field, specify the directory that you want the snapshot to contain.

5. Optional: To create an alternative name for the snapshot, specify a snapshot alias.

a. Next to Create an Alias, click Yes.

Snapshots

Create a snapshot 179

b. To modify the default snapshot alias name, in the Alias Name field, type analternative name for the snapshot.

6. Optional: To assign a time that OneFS will automatically delete the snapshot, specifyan expiration period.

a. Next to Snapshot Expiration, click Snapshot Expires on.

b. In the calendar, specify the day that you want the snapshot to be automaticallydeleted.

7. Click Capture.

Snapshot naming patternsIf you schedule snapshots to be automatically generated, either according to a snapshotschedule or a replication policy, you must assign a snapshot naming pattern thatdetermines how the snapshots are named. Snapshot naming patterns contain variablesthat include information about how and when the snapshot was created.

The following variables can be included in a snapshot naming pattern:

Variable Description- -%A The day of the week.

%a The abbreviated day of the week. For example, if the snapshot is generatedon a Sunday, %a is replaced with Sun.

%B The name of the month.

%b The abbreviated name of the month. For example, if the snapshot isgenerated in September, %b is replaced with Sep.

%C The first two digits of the year. For example, if the snapshot is created in2012, %C is replaced with 20.

%c The time and day. This variable is equivalent to specifying %a %b %e %T%Y.

%d The two digit day of the month.

%e The day of the month. A single-digit day is preceded by a blank space.

%F The date. This variable is equivalent to specifying %Y-%m-%d

%G The year. This variable is equivalent to specifying %Y. However, if the

snapshot is created in a week that has less than four days in the current year,the year that contains the majority of the days of the week is displayed. Thefirst day of the week is calculated as Monday. For example, if a snapshot iscreated on Sunday, January 1, 2017, %G is replaced with 2016, because only

one day of that week is in 2017.

%g The abbreviated year. This variable is equivalent to specifying %y. However, if

the snapshot was created in a week that has less than four days in thecurrent year, the year that contains the majority of the days of the week isdisplayed. The first day of the week is calculated as Monday. For example, if asnapshot is created on Sunday, January 1, 2017, %g is replaced with 16,

because only one day of that week is in 2017.

Snapshots


Variable Description- -%H The hour. The hour is represented on the 24-hour clock. Single-digit hours are

preceded by a zero. For example, if a snapshot is created at 1:45 AM, %H is

replaced with 01.

%h The abbreviated name of the month. This variable is equivalent to specifying%b.

%I The hour represented on the 12-hour clock. Single-digit hours are precededby a zero. For example, if a snapshot is created at 1:45 AM, %I is replaced

with 01.

%j The numeric day of the year. For example, if a snapshot is created onFebruary 1, %j is replaced with 32.

%k The hour represented on the 24-hour clock. Single-digit hours are precededby a blank space.

%l The hour represented on the 12-hour clock. Single-digit hours are precededby a blank space. For example, if a snapshot is created at 1:45 AM, %I is

replaced with 1.

%M The two-digit minute.

%m The two-digit month.

%p AM or PM.

%{PolicyName} The name of the replication policy that the snapshot was created for. Thisvariable is valid only if you are specifying a snapshot naming pattern for areplication policy.

%R The time. This variable is equivalent to specifying %H:%M.

%r The time. This variable is equivalent to specifying %I:%M:%S %p.

%S The two-digit second.

%s The second represented in UNIX or POSIX time.

%{SrcCluster} The name of the source cluster of the replication policy that the snapshot wascreated for. This variable is valid only if you are specifying a snapshot namingpattern for a replication policy.

%T The time. This variable is equivalent to specifying %H:%M:%S

%U The two-digit numerical week of the year. Numbers range from 00 to 53. The

first day of the week is calculated as Sunday.

%u The numerical day of the week. Numbers range from 1 to 7. The first day of

the week is calculated as Monday. For example, if a snapshot is created onSunday, %u is replaced with 7.

%V The two-digit numerical week of the year that the snapshot was created in.Numbers range from 01 to 53. The first day of the week is calculated as

Monday. If the week of January 1 is four or more days in length, then thatweek is counted as the first week of the year.

Snapshots

Snapshot naming patterns 181

Variable Description- -%v The day that the snapshot was created. This variable is equivalent to

specifying %e-%b-%Y.

%W The two-digit numerical week of the year that the snapshot was created in.Numbers range from 00 to 53. The first day of the week is calculated as

Monday.

%w The numerical day of the week that the snapshot was created on. Numbersrange from 0 to 6. The first day of the week is calculated as Sunday. For

example, if the snapshot was created on Sunday, %w is replaced with 0.

%X The time that the snapshot was created. This variable is equivalent tospecifying %H:%M:%S.

%Y The year that the snapshot was created in.

%y The last two digits of the year that the snapshot was created in. For example,if the snapshot was created in 2012, %y is replaced with 12.

%Z The time zone that the snapshot was created in.

%z The offset from coordinated universal time (UTC) of the time zone that thesnapshot was created in. If preceded by a plus sign, the time zone is east ofUTC. If preceded by a minus sign, the time zone is west of UTC.

%+ The time and date that the snapshot was created. This variable is equivalentto specifying %a %b %e %X %Z %Y.

%% Escapes a percent sign. "100%%" is replaced with 100%.

Managing snapshotsYou can delete and view snapshots. You can also modify the name, duration period, andalias of an existing snapshot. However, you cannot modify the data contained in asnapshot; the data contained in a snapshot is read-only.

Reducing snapshot disk-space usageIf multiple snapshots contain the same directories, deleting one of the snapshots mightnot free the entire amount of space that the system reports as the size of the snapshot.The size of a snapshot is the maximum amount of data that might be freed if thesnapshot is deleted.

Deleting a snapshot frees only the space that is taken up exclusively by that snapshot. Iftwo snapshots reference the same stored data, that data is not freed until bothsnapshots are deleted. Remember that snapshots store data contained in allsubdirectories of the root directory; if snapshot_one contains /ifs/data/, andsnapshot_two contains /ifs/data/dir, the two snapshots most likely share data.

If you delete a directory, and then re-create it, a snapshot containing the directory storesthe entire re-created directory, even if the files in that directory are never modified.

Deleting multiple snapshots that contain the same directories is more likely to free datathan deleting multiple snapshots that contain different directories.

Snapshots


If multiple snapshots contain the same directories, deleting older snapshots is morelikely to free disk-space than deleting newer snapshots.

Snapshots that are assigned expiration dates are automatically marked for deletion bythe snapshot daemon. If the daemon is disabled, snapshots will not be automaticallydeleted by the system. It is recommended that you do not disable the snapshot daemon.

Delete snapshotsYou can delete a snapshot if you no longer want to access the data contained in thesnapshot.

OneFS frees disk space occupied by deleted snapshots when the snapshot delete job isrun. Also, if you delete a snapshot that contains clones or cloned files, data in a shadowstore might no longer be referenced by files on the cluster; OneFS deletes unreferenceddata in a shadow store when the shadow store delete job is run. OneFS routinely runsboth the shadow store delete and snapshot delete jobs. However, you can also manuallyrun the jobs at any time.

Procedure

1. Click Data Protection > SnapshotIQ > Snapshots.

2. Specify the snapshots that you want to delete.

a. For each snapshot you want to delete, in the Saved File System Snapshots table,in the row of a snapshot, select the check box.

b. From the Select an action list, select Delete.

c. In the confirmation dialog box, click Delete.

3. Optional: To increase the speed at which deleted snapshot data is freed on thecluster, run the snapshot delete job.

a. Click Cluster Management > Operations.

b. In the Running Jobs area, click Start Job.

c. From the Job list, select SnapshotDelete.

d. Click Start.

4. Optional: To increase the speed at which deleted data shared between deduplicatedand cloned files is freed on the cluster, run the shadow store delete job.

Run the shadow store delete job only after you run the snapshot delete job.

a. Click Cluster Management > Operations.

b. In the Running Jobs area, click Start Job.

c. From the Job list, select ShadowStoreDelete.

d. Click Start.

Modify snapshot attributesYou can modify the name and expiration date of a snapshot.

Procedure

1. Click File System Management > SnapshotIQ > Snapshots.

2. In the Saved File System Snapshots table, in the row of a snapshot, click ViewDetails.

Snapshots

Delete snapshots 183

3. In the Snapshot Details area, modify snapshot attributes.

4. Next to each snapshot attribute that you modified, click Save.

Modify a snapshot aliasYou can modify the alias of a snapshot to assign an alternative name for the snapshot.

Procedure


2. Above the Saved File System Snapshots table, click View snapshot aliases.

3. In the Snapshot Aliases table, in the row of an alias, click View details.

4. In the Snapshot Alias Details pane, in the Alias Name area, click Edit.

5. In the Alias Name field, type a new alias name.

6. Click Save.

View snapshotsYou can view all snapshots.

Procedure


2. In the Saved File System Snapshots table, view snapshots.

Snapshot informationYou can view information about snapshots, including the total amount of spaceconsumed by all snapshots.

The following information is displayed in the Saved Snapshots area:

SnapshotIQ StatusIndicates whether a SnapshotIQ license has been activated on the cluster.

Total Number of Saved SnapshotsIndicates the total number of snapshots that exist on the cluster.

Total Number of Snapshots Pending DeletionIndicates the total number of snapshots that were deleted on the cluster since thelast snapshot delete job was run. The space consumed by the deleted snapshots isnot freed until the snapshot delete job is run again.

Total Number of Snapshot AliasesIndicates the total number of snapshot aliases that exist on the cluster.

Capacity Used by Saved SnapshotsIndicates the total amount of space consumed by all snapshots.

Restoring snapshot dataYou can restore snapshot data through various methods. You can revert a snapshot oraccess snapshot data through the snapshots directory.

From the snapshots directory, you can either clone a file or copy a directory or a file. Thesnapshots directory can be accessed through Windows Explorer or a UNIX command line.

Snapshots


You can disable and enable access to the snapshots directory for any of these methodsthrough snapshots settings.

Revert a snapshotYou can revert a directory back to the state it was in when a snapshot was taken.

Before you begin

u Create a SnapRevert domain for the directory.

u Create a snapshot of a directory.

Procedure

1. Click Cluster Management > Operations > Operations Summary.

2. In the Running Jobs area, click Start job.

3. From the Job list, select SnapRevert.

4. Optional: To specify a priority for the job, from the Priority list, select a priority.

Lower values indicate a higher priority. If you do not specify a priority, the job isassigned the default snapshot revert priority.

5. Optional: To specify the amount of cluster resources the job is allowed to consume,from the Impact policy list, select an impact policy.

If you do not specify a policy, the job is assigned the default snapshot revert policy.

6. In the Snapshot field, type the name or ID of the snapshot that you want to revert, andthen click Start.

Restore a file or directory using Windows ExplorerIf the Microsoft Shadow Copy Client is installed on your computer, you can use it torestore files and directories that are stored in snapshots.

Note

You can access up to 64 snapshots of a directory through Windows explorer, starting withthe most recent snapshot. To access more than 64 snapshots for a directory, access thecluster through a UNIX command line.

Procedure

1. In Windows Explorer, navigate to the directory that you want to restore or the directorythat contains the file that you want to restore.

2. Right-click the folder, and then click Properties.

3. In the Properties window, click the Previous Versions tab.

4. Select the version of the folder that you want to restore or the version of the folderthat contains the version of the file that you want to restore.

5. Restore the version of the file or directory.

l To restore all files in the selected directory, click Restore.

l To copy the selected directory to another location, click Copy and then specify alocation to copy the directory to.

l To restore a specific file, click Open, and then copy the file into the originaldirectory, replacing the existing copy with the snapshot version.

Snapshots

Revert a snapshot 185

Restore a file or directory through a UNIX command lineYou can restore a file or directory through a UNIX command line.

Procedure

1. Open a connection to the cluster through a UNIX command line.

2. Optional: To view the contents of the snapshot you want to restore a file or directoryfrom, run the ls command for a subdirectory of the snapshots root directory.

For example, the following command displays the contents of the /archivedirectory contained in Snapshot2012Jun04:

ls /ifs/.snapshot/Snapshot2012Jun04/archive3. Copy the file or directory by using the cp command.

For example, the following command creates a copy of file1:

cp /ifs/.snapshot/Snapshot2012Jun04/archive/file1 \/ifs/archive/file1_copy

Clone a file from a snapshotYou can clone a file from a snapshot. This procedure is available only through thecommand-line interface (CLI).

Procedure

1. Open a secure shell (SSH) connection to any node in the cluster and log in.

2. To view the contents of the snapshot you want to restore a file or directory from, runthe ls command for a subdirectory of the snapshots root directory.

For example, the following command displays the contents of the /archivedirectory contained in Snapshot2012Jun04:

ls /ifs/.snapshot/Snapshot2012Jun04/archive3. Clone a file from the snapshot by running the cp command with the -c option.

For example, the following command clones test.txt from Snapshot2012Jun04:

cp -c /ifs/.snapshot/Snapshot2012Jun04/archive/test.txt \/ifs/archive/test_clone.text

Managing snapshot schedulesYou can modify, delete, and view snapshot schedules.

Modify a snapshot scheduleYou can modify a snapshot schedule. Any changes to a snapshot schedule are appliedonly to snapshots generated after the modifications are made. Existing snapshots are notaffected by schedule modifications.

If you modify the alias of a snapshot schedule, the alias is assigned to the next snapshotgenerated based on the schedule. However, if you do this, the old alias is not removedfrom the last snapshot that it was assigned to. Unless you manually remove the old alias,the alias will remain attached to the last snapshot that it was assigned to.

Procedure


Snapshots


2. In the Snapshot Schedules table, in the row of the snapshot schedule you want tomodify, click View details.

3. In the Snapshot Schedule Details area, modify snapshot schedule attributes.

4. Next to each snapshot schedule attribute that you modified, click Save.

Delete a snapshot scheduleYou can delete a snapshot schedule. Deleting a snapshot schedule will not deletesnapshots that were previously generated according to the schedule.

Procedure


2. In the Snapshot Schedules table, in the row of the snapshot schedule you want todelete, click Delete.

3. In the Confirm Delete dialog box, click Delete.

View snapshot schedulesYou can view snapshot schedules.

Procedure


2. In the Snapshot Schedules table, view snapshot schedules.

3. Optional: To view detailed information about a snapshot schedule, in the SnapshotSchedules table, in the row of the snapshot schedule that you want to view, click Viewdetails.

Snapshot schedule settings are displayed in the Snapshot Schedule Details area.Snapshots that are scheduled to be generated according to the schedule aredisplayed in the Snapshot Calendar area.

Managing with snapshot locksYou can delete, create, and modify the expiration date of snapshot locks.

CAUTION

It is recommended that you do not create, delete, or modify snapshots locks unless youare instructed to do so by Isilon Technical Support.

Deleting a snapshot lock that was created by OneFS might result in data loss. If youdelete a snapshot lock that was created by OneFS, it is possible that the correspondingsnapshot might be deleted while it is still in use by OneFS. If OneFS cannot access asnapshot that is necessary for an operation, the operation will malfunction and data lossmight result. Modifying the expiration date of a snapshot lock created by OneFS can alsoresult in data loss because the corresponding snapshot can be deleted prematurely.

Create a snapshot lockYou can create snapshot locks that prevent snapshots from being deleted. Thisprocedure is available only through the command-line interface (CLI).

Although you can prevent a snapshot from being automatically deleted by creating asnapshot lock, it is recommended that you do not create snapshot locks. To prevent a

Snapshots

Delete a snapshot schedule 187

snapshot from being automatically deleted, it is recommended that you extend theduration period of the snapshot by modifying the snapshot.

Procedure


2. Create a snapshot lock by running the isi snapshot locks create command.

For example, the following command applies a snapshot lock to"SnapshotApril2012", sets the lock to expire in one month, and adds a description of"Maintenance Lock":

isi snapshot locks create SnapshotApril2012 --expires 1M \--comment "Maintenance Lock"

Modify a snapshot lock expiration dateYou can modify the expiration date of a snapshot lock. This procedure is available onlythrough the command-line interface (CLI).

CAUTION

It is recommended that you do not modify the expiration dates of snapshot locks.

Procedure


2. Modify a snapshot lock by running the isi snapshot locks modify command.

For example, the following command sets a snapshot lock that is applied to"SnapshotApril2012" and has an ID of 1 to expire in two days:isi snapshot locks modify Snapshot2012Apr16 1 --expires 2D

Delete a snapshot lockYou can delete a snapshot lock. This procedure is available only through the command-line interface (CLI).

CAUTION

It is recommended that you do not delete snapshot locks.

Procedure


2. Delete a snapshot lock by running the isi snapshot locks delete command.

For example, the following command deletes a snapshot lock that is applied toSnapshotApril2012 and has an ID of 1:

isi snapshot locks delete Snapshot2012Apr16 1

The system prompts you to confirm that you want to delete the snapshot lock.

3. Type yes and then press ENTER.

Snapshots


Snapshot lock informationYou can view snapshot lock information through the isi snapshot locks view andisi snapshot locks list commands.

IDNumerical identification number of the snapshot lock.

CommentDescription of the snapshot lock. This can be any string specified by a user.

ExpiresThe date that the snapshot lock will be automatically deleted by OneFS.

CountThe number of times the snapshot lock is held.

The file clone operation can hold a single snapshot lock multiple times. If multiplefile clones are created simultaneously, the file clone operation holds the same lockmultiple times, rather than creating multiple locks. If you delete a snapshot lock thatis held more than once, you will delete only one of the instances that the lock isheld. In order to delete a snapshot lock that is held multiple times, you must deletethe snapshot lock the same number of times as displayed in the count field.

Configure SnapshotIQ settingsYou can configure SnapshotIQ settings that determine how snapshots can be created andthe methods that users can access snapshot data.

Procedure

1. Click Data Protection > SnapshotIQ > Settings.

2. Modify SnapshotIQ settings, and then click Save.

SnapshotIQ settingsSnapshotIQ settings determine how snapshots behave and can be accessed.

The following SnapshotIQ settings can be configured:

Snapshot SchedulingDetermines whether snapshots can be generated.

Note

Disabling snapshot generation might cause some OneFS operations to fail. It isrecommended that you do not disable this setting.

Auto-create SnapshotsDetermines whether snapshots are automatically generated according tosnapshot schedules.

Auto-delete SnapshotsDetermines whether snapshots are automatically deleted according to theirexpiration dates.

Snapshots

Snapshot lock information 189

NFS Visibility & Accessibility

Root Directory AccessibleDetermines whether snapshot directories are accessible through NFS.

Root Directory VisibleDetermines whether snapshot directories are visible through NFS.

Sub-directories AccessibleDetermines whether snapshot subdirectories are accessible through NFS.

SMB Visibility & Accessible

Root Directory AccessibleDetermines whether snapshot directories are accessible through SMB.

Root Directory VisibleDetermines whether snapshot directories are visible through SMB.

Sub-directories AccessibleDetermines whether snapshot subdirectories are accessible through SMB.

Local Visibility & Accessibility

Root Directory AccessibleDetermines whether snapshot directories are accessible through the local filesystem. You can access the local file system through an SSH connection or thelocal console.

Root Directory VisibleDetermines whether snapshot directories are visible through the local filesystem. You can access the local file system through an SSH connection or thelocal console.

Sub-directories AccessibleDetermines whether snapshot subdirectories are accessible through the localfile system. You can access the local file system through an SSH connection orthe local console.

Set the snapshot reserveYou can specify a minimum percentage of cluster-storage capacity that you want toreserve for snapshots. This procedure is available only through the command-lineinterface (CLI).

The snapshot reserve does not limit the amount of space that snapshots are allowed toconsume on the cluster. Snapshots can consume more than the percentage of capacityspecified by the snapshot reserve. It is recommended that you do not specify a snapshotreserve.

Procedure


2. Set the snapshot reserve by running the isi snapshot settings modifycommand with the --reserve option.

Snapshots


For example, the following command sets the snapshot reserve to 20%:isi snapshot settings modify --reserve 20

Snapshots

Set the snapshot reserve 191

CHAPTER 10

Deduplication with SmartDedupe


u Deduplication overview.......................................................................................194u Deduplication jobs..............................................................................................194u Data replication and backup with deduplication..................................................195u Snapshots with deduplication.............................................................................195u Deduplication considerations..............................................................................195u Shadow store considerations.............................................................................. 196u SmartDedupe license functionality......................................................................196u Managing deduplication..................................................................................... 197

Deduplication with SmartDedupe 193

Deduplication overviewThe SmartDedupe software module enables you to save storage space on your cluster byreducing redundant data. Deduplication maximizes the efficiency of your cluster bydecreasing the amount of storage required to store multiple files with similar blocks.

SmartDedupe deduplicates data by scanning an Isilon cluster for identical data blocks.Each block is 8 KB. If SmartDedupe finds duplicate blocks, SmartDedupe moves a singlecopy of the blocks to a hidden file called a shadow store. SmartDedupe then deletes theduplicate blocks from the original files and replaces the blocks with pointers to theshadow store.

Deduplication is applied at the directory level, targeting all files and directoriesunderneath one or more root directories. You can first assess a directory fordeduplication and determine the estimated amount of space you can expect to save. Youcan then decide whether to deduplicate the directory. After you begin deduplicating adirectory, you can monitor how much space is saved by deduplication in real time.

SmartDedupe does not deduplicate files that are 32 KB and smaller, because doing sowould consume more cluster resources than the storage savings are worth. Each shadowstore can contain up to 255 blocks. Each block in a shadow store can be referenced32000 times.

Deduplication jobsDeduplication is performed by maintenance jobs referred to as deduplication jobs. Youcan monitor and control deduplication jobs as you would any other maintenance job onthe cluster. Although the overall performance impact of deduplication is minimal, thededuplication job consumes 256 MB of memory per node.

When a deduplication job is first run on a cluster, SmartDedupe samples blocks fromeach file and creates index entries for those blocks. If the index entries of two blocksmatch, SmartDedupe scans the blocks adjacent to the matching pair and thendeduplicates all duplicate blocks. After a deduplication job samples a file once, newdeduplication jobs will not sample the file again until the file is modified.

The first deduplication job you run might take significantly longer to complete thansubsequent deduplication jobs. The first deduplication job must scan all files under thespecified directories to generate the initial index. If subsequent deduplication jobs take along time to complete, this most likely indicates that a large amount of data is beingdeduplicated. However, it can also indicate that clients are creating a large amount ofnew data on the cluster. If a deduplication job is interrupted during the deduplicationprocess, the job will automatically restart the scanning process from where the job wasinterrupted.

It is recommended that you run deduplication jobs when clients are not modifying dataon the cluster. If clients are continually modifying files on the cluster, the amount ofspace saved by deduplication is minimal because the deduplicated blocks are constantlyremoved from the shadow store. For most clusters, it is recommended that you start adeduplication job every ten days.

The permissions required to modify deduplication settings are not the same as thoseneeded to run a deduplication job. Although a user must have the maintenance jobpermission to run a deduplication job, the user must have the deduplication permissionto modify deduplication settings. By default, the deduplication job is configured to run ata low priority.



Data replication and backup with deduplicationWhen deduplicated files are replicated to another Isilon cluster or backed up to a tapedevice, the deduplicated files no longer share blocks on the target Isilon cluster orbackup device. However, although you can deduplicate data on a target Isilon cluster,you cannot deduplicate data on an NDMP backup device.

Shadows stores are not transferred to target clusters or backup devices. Because of this,deduplicated files do not consume less space than non-deduplicated files when they arereplicated or backed up. To avoid running out of space, you must ensure that targetclusters and tape devices have enough free space to store deduplicated data as if thedata had not been deduplicated. To reduce the amount of storage space consumed on atarget Isilon cluster, you can configure deduplication for the target directories of yourreplication policies. Although this will deduplicate data on the target directory, it will notallow SyncIQ to transfer shadow stores. Deduplication is still performed by deduplicationjobs running on the target cluster.

The amount of cluster resources required to backup and replicate deduplicated data isthe same as for non-deduplicated data. You can deduplicate data while the data is beingreplicated or backed up.

Snapshots with deduplicationYou cannot deduplicate the data stored in a snapshot. However, you can createsnapshots of deduplicated data.

If you create a snapshot for a deduplicated directory, and then modify the contents ofthat directory, the references to shadow stores will be transferred to the snapshot overtime. Therefore, if you enable deduplication before you create snapshots, you will savemore space on your cluster. If you implement deduplication on a cluster that already hasa significant amount of data stored in snapshots, it will take time before the snapshotdata is affected by deduplication. Newly created snapshots can contain deduplicateddata, but snapshots created before deduplication was implemented cannot.

If you plan on reverting a snapshot, it is best to revert the snapshot before running adeduplication job. Restoring a snapshot can overwrite many of the files on the cluster.Any deduplicated files are reverted back to normal files if they are overwritten by asnapshot revert. However, after the snapshot revert is complete, you can deduplicate thedirectory and the space savings persist on the cluster.

Deduplication considerationsDeduplication can significantly increase the efficiency at which you store data. However,the effect of deduplication varies depending on the cluster.

You can reduce redundancy on a cluster by running SmartDedupe. Deduplication createslinks that can impact the speed at which you can read from and write to files. Inparticular, sequentially reading chunks smaller than 512 KB of a deduplicated file can besignificantly slower than reading the same small, sequential chunks of a non-deduplicated file. This performance degradation applies only if you are reading non-cached data. For cached data, the performance for deduplicated files is potentially betterthan non-deduplicated files. If you stream chunks larger than 512 KB, deduplication doesnot significantly impact the read performance of the file. If you intend on streaming 8 KBor less of each file at a time, and you do not plan on concurrently streaming the files, it isrecommended that you do not deduplicate the files.


Data replication and backup with deduplication 195

Deduplication is most effective when applied to static or archived files and directories.The less files are modified, the less negative impact deduplication has on the cluster. Forexample, virtual machines often contain several copies of identical files that are rarelymodified. Deduplicating a large number of virtual machines can greatly decrease theamount of storage space consumed.

SmartDedupe will not deduplicate redundant information within a file. If a file containsmultiple identical data blocks, SmartDedupe will not deduplicate that data unlessanother file contains the identical block.

SmartDedupe deduplicates directories that contain iSCSI LUNs the same as otherdirectories.

Shadow store considerationsShadow stores are hidden files that are referenced by cloned and deduplicated files. Filesthat reference shadow stores behave differently than other files.

u Reading shadow-store references might be slower than reading data directly.Specifically, reading non-cached shadow-store references is slower than reading non-cached data. Reading cached shadow-store references takes no more time thanreading cached data.

u When files that reference shadow stores are replicated to another Isilon cluster orbacked up to a Network Data Management Protocol (NDMP) backup device, theshadow stores are not transferred to the target Isilon cluster or backup device. Thefiles are transferred as if they contained the data that they reference from shadowstores. On the target Isilon cluster or backup device, the files consume the sameamount of space as if they had not referenced shadow stores.

u When OneFS creates a shadow store, OneFS assigns the shadow store to a storagepool of a file that references the shadow store. If you delete the storage pool that ashadow store resides on, the shadow store is moved to a pool occupied by anotherfile that references the shadow store.

u OneFS does not delete a shadow store block immediately after the last reference tothe block is deleted. Instead, OneFS waits until the ShadowStoreDelete job is run todelete the unreferenced block. If a large number of unreferenced blocks exist on thecluster, OneFS might report a negative deduplication savings until theShadowStoreDelete job is run.

u Shadow stores are protected at least as much as the most protected file thatreferences it. For example, if one file that references a shadow store resides in astorage pool with +2 protection and another file that references the shadow storeresides in a storage pool with +3 protection, the shadow store is protected at +3.

u Quotas account for files that reference shadow stores as if the files contained thedata referenced from shadow stores; from the perspective of a quota, shadow storereferences do not exist. However, if a quota includes data protection overhead, thequota does not account for the data protection overhead of shadow stores.

SmartDedupe license functionalityYou can deduplicate data only if you activate a SmartDedupe license on a cluster.However, you can assess deduplication savings without activating a SmartDedupelicense.

If you activate a SmartDedupe license, and then deduplicate data, the space savings arenot lost if the license becomes inactive. You can also still view deduplication savings



while the license is inactive. However, you will not be able to deduplicate additional datauntil you re-activate the SmartDedupe license.

Managing deduplicationYou can manage deduplication on a cluster by first assessing how much space you cansave by deduplicating individual directories. After you determine which directories areworth deduplicating, you can configure SmartDedupe to deduplicate those directoriesspecifically. You can then monitor the actual amount of disk space you are saving.

Assess deduplication space savingsYou can assess the amount of disk space you will save by deduplicating a directory.

Procedure

1. Click File System Management > Deduplication > Settings.

2. In the Assess Deduplication area, click Browse and select a directory that you want todeduplicate.

If you assess multiple directories, disk savings are not differentiated by directory inthe deduplication report.


4. In the Job Types table, in the row of the DedupeAssessment job, from the Actionscolumn, select Start Job.

5. Click Cluster Management > Job Operations > Job Summary.

6. Wait for the assessment job to complete.

When the DedupeAssessment job is complete, the job is removed from the ActiveJobs table.

7. Click File System Management > Deduplication > Summary.

In the Deduplication Assessment Reports table, in the row of the most recentassessment job, click View Details.

8. View the amount of disk space that will be saved if you deduplicate the directory.

The number of blocks that will be deduplicated is displayed in the Deduped blocksfield.

Specify deduplication settingsYou can specify which directories you want to deduplicate.

Procedure

1. Click File System Management > Deduplication > Settings.

2. In the Deduplication Settings area, click Browse and select a directory that you wantto deduplicate.

3. Optional: Specify additional directories.

a. Click Add another directory path.

b. Click Browse and select a directory that you want to deduplicate.

4. Click Cluster Management > Job Operations > Jobs Types.

5. In the Jobs table, in the row of the Dedupe job, click View/Edit.


Managing deduplication 197

6. Click Edit Job Type.

7. Modify the settings of the deduplication job, and then click Save Changes.

View deduplication space savingsYou can view the amount of disk space that you are currently saving with deduplication.

Procedure


2. In the Deduplication Savings area, view the amount of disk space saved.

View a deduplication reportAfter a deduplication job completes, you can view information about the job in adeduplication report.

Procedure


2. Select a deduplication report.

l To view a report about a deduplication job, in the Deduplication Reports table,click View Report.

l To view a report about a deduplication assessment job, in the DeduplicationAssessment Reports table, click View Report.

Deduplication job report informationYou can view the following deduplication specific information in deduplication jobreports:

Start timeThe time the deduplication job started.

End timeThe time the deduplication job ended.

Iteration CountThe number of times that SmartDedupe interrupted the sampling process. IfSmartDedupe is sampling a large amount of data, SmartDedupe might interruptsampling in order to start deduplicating the data. After SmartDedupe finishesdeduplicating the sampled data, SmartDedupe will continue sampling the remainingdata.

Scanned blocksThe total number of blocks located underneath the specified deduplicateddirectories.

Sampled blocksThe number of blocks that SmartDedupe created index entries for.

Deduped blocksThe number of blocks that were deduplicated.

Dedupe percentThe percentage of scanned blocks that were deduplicated.



Created dedupe requestsThe total number of deduplication requests created. A deduplication request iscreated for each matching pair of data blocks. For example, if you have 3 data blocksthat all match, SmartDedupe creates 2 requests. One of the requests could pair file1and file2 together and the other request could pair file2 and file3 together.

Successful dedupe requestsThe number of deduplication requests that completed successfully.

Failed dedupe requestsThe number of deduplication requests that failed. If a deduplication request fails, itdoesn't mean that the job failed too. A deduplication request can fail for any numberof reasons. For example, the file might have been modified since it was sampled.

Skipped filesThe number of files that were not scanned by the deduplication job. SmartDedupeskips files for a number of reasons. For example, SmartDedupe skips files that havealready been scanned and haven't been modified since. SmartDedupe also skips allfiles that are smaller than 4 KB.

Index entriesThe number of entries that currently exist in the index.

Index lookup attemptsThe total number of lookups that have been done by earlier deduplication jobs plusthe number of lookups done by this deduplication job. A lookup is when thededuplication job attempts to match a block that was indexed with a block thathasn't been indexed.

Index lookup hitsThe number of blocks that matched index entries.

Deduplication informationYou can view the amount of disk space saved by deduplication in the DeduplicationSavings area:

Space SavingsThe total amount of physical disk space saved by deduplication, includingprotection overhead and metadata. For example, if you have three identical files thatare all 5 GB, the estimated physical saving would be greater than 10 GB, becausededuplication saved space that would have been occupied by file metadata andprotection overhead.

Deduplicated dataThe amount of space on the cluster occupied by directories that were deduplicated.

Other dataThe amount of space on the cluster occupied by directories that were notdeduplicated.


Deduplication information 199

CHAPTER 11

Data replication with SyncIQ


u SyncIQ backup and recovery overview.................................................................202u Replication policies and jobs.............................................................................. 202u Replication snapshots.........................................................................................205u Data failover and failback with SyncIQ.................................................................206u Recovery times and objectives for SyncIQ............................................................207u SyncIQ license functionality................................................................................ 208u Creating replication policies................................................................................208u Managing replication to remote clusters..............................................................218u Initiating data failover and failback with SyncIQ.................................................. 220u Performing disaster recovery for SmartLock directories........................................222u Managing replication policies............................................................................. 224u Managing replication to the local cluster............................................................. 228u Managing replication performance rules............................................................. 230u Managing replication reports...............................................................................231u Managing failed replication jobs......................................................................... 233

Data replication with SyncIQ 201

SyncIQ backup and recovery overviewOneFS enables you to replicate data from one Isilon cluster to another through the SyncIQsoftware module. You must activate a SyncIQ license on both Isilon clusters before youcan replicate data between them.

You can replicate data at the directory level while optionally excluding specific files andsub-directories from being replicated. SyncIQ creates and references snapshots toreplicate a consistent point-in-time image of a root directory. Metadata such as accesscontrol lists (ACLs) and alternate data streams (ADS) are replicated along with data.

SyncIQ enables you to maintain a consistent backup copy of your data on another Isiloncluster. SyncIQ offers automated failover and failback capabilities that enable you tocontinue operations on another Isilon cluster if a primary cluster becomes unavailable.

Replication policies and jobsData replication is coordinated according to replication policies and jobs. Replicationpolicies specify what data is replicated, where the data is replicated to, and how oftenthe data is replicated. Replication jobs are the operations that replicate data from oneIsilon cluster to another. SyncIQ generates replication jobs according to replicationpolicies.

A replication policy specifies two clusters: the source and the target. The cluster on whichthe replication policy exists is the source cluster. The cluster that data is being replicatedto is the target cluster. When a replication policy starts, SyncIQ generates a replicationjob for the policy. When a replication job runs, files from a directory on the source clusterare replicated to a directory on the target cluster; these directories are known as sourceand target directories.

After the first replication job created by a replication policy finishes, the target directoryand all files contained in the target directory are set to a read-only state, and can bemodified only by other replication jobs belonging to the same replication policy. There isno limit to the number of replication policies that can exist on a cluster.

Note

To prevent permissions errors, make sure that ACL policy settings are the same acrosssource and target clusters.

You can create two types of replication policies: synchronization policies and copypolicies. A synchronization policy maintains an exact replica of the source directory onthe target cluster. If a file or sub-directory is deleted from the source directory, the file ordirectory is deleted from the target cluster when the policy is run again.

You can use synchronization policies to fail over and fail back data between source andtarget clusters. When a source cluster becomes unavailable, you can fail over data on atarget cluster and make the data available to clients. When the source cluster becomesavailable again, you can fail back the data to the source cluster.

A copy policy maintains recent versions of the files that are stored on the source cluster.However, files that are deleted on the source cluster are not deleted from the targetcluster. Failback is not supported for copy policies. Copy policies are most commonlyused for archival purposes.

Copy policies enable you to remove files from the source cluster without losing those fileson the target cluster. Deleting files on the source cluster improves performance on thesource cluster while maintaining the deleted files on the target cluster. This can be useful



if, for example, your source cluster is being used for production purposes and your targetcluster is being used only for archiving.

After creating a job for a replication policy, SyncIQ must wait until the job completesbefore it can create another job for the policy. Any number of replication jobs can exist ona cluster at a given time; however, only five replication jobs can run on a source cluster atthe same time. If more than five replication jobs exist on a cluster, the first five jobs runwhile the others are queued to run. The number of replication jobs that a single targetcluster can support concurrently is dependent on the number of workers available on thetarget cluster.

You can replicate any number of files and directories with a single replication job. Youcan prevent a large replication job from overwhelming the system by limiting the amountof cluster resources and network bandwidth that data synchronization is allowed toconsume. Because each node in a cluster is able to send and receive data, the speed atwhich data is replicated increases for larger clusters.

Automated replication policiesYou can manually start a replication policy at any time, but you can also configurereplication policies to start automatically based on source directory modifications or aschedule.

You can configure a replication policy to run according to a schedule, so that you cancontrol when replication is performed. You can also configure a replication policy to startwhen SyncIQ detects a modification to the source directory, so that SyncIQ maintains amore current version of your data on the target cluster.

Scheduling a policy can be useful under the following conditions:

u You want to replicate data when user activity is minimal

u You can accurately predict when modifications will be made to the data

Configuring a policy to start when changes are made to the source directory can be usefulunder the following conditions:

u You want retain a consistent copy of your data at all times

u You are expecting a large number of changes at unpredictable intervals

For policies that are configured to start whenever changes are made to the sourcedirectory, SyncIQ checks the source directories every ten seconds. SyncIQ does notaccount for excluded files or directories when detecting changes, so policies that excludefiles or directories from replication might be run unnecessarily. For example, assume thatnewPolicy replicates /ifs/data/media but excludes /ifs/data/media/temp. If amodification is made to /ifs/data/media/temp/file.txt, SyncIQ will runnewPolicy, but will not replicate /ifs/data/media/temp/file.txt.

If a policy is configured to start whenever changes are made to its source directory, and areplication job fails, SyncIQ will wait one minute before attempting to run the policyagain. SyncIQ will increase this delay exponentially for each failure up to a maximumdelay of eight hours. You can override the delay by running the policy manually at anytime. After a job for the policy completes successfully, SyncIQ will resume checking thesource directory every ten seconds.

Source and target cluster associationSyncIQ associates a replication policy with a target cluster by marking the target clusterwhen the job runs for the first time. Even if you modify the name or IP address of the


Automated replication policies 203

target cluster, the mark persists on the target cluster. When a replication policy is run,SyncIQ checks the mark to ensure that data is being replicated to the correct location.

On the target cluster, you can manually break an association between a replication policyand target directory. Breaking the association between a source and target cluster causesthe mark on the target cluster to be deleted. You might want to manually break a targetassociation if an association is obsolete. If you break the association of a policy, thepolicy is disabled on the source cluster and you cannot run the policy. If you want to runthe disabled policy again, you must reset the replication policy.

Note

Breaking a policy association causes either a full or differential replication to occur thenext time you run the replication policy. During a full or differential replication, SyncIQcreates a new association between the source and target clusters. Depending on theamount of data being replicated, a full or differential replication can take a very long timeto complete.

Full and differential replicationIf a replication policy encounters an issue that cannot be fixed (for example, if theassociation was broken on the target cluster), you might need to reset the replicationpolicy. If you reset a replication policy, SyncIQ performs either a full or differentialreplication the next time the policy is run. You can specify the type of replication thatSyncIQ performs.

During a full replication, SyncIQ transfers all data from the source cluster regardless ofwhat data exists on the target cluster. A full replication consumes large amounts ofnetwork bandwidth and can take a very long time to complete. However, a full replicationis less strenuous on CPU usage than a differential replication.

During a differential replication, SyncIQ first checks whether a file already exists on thetarget cluster and then transfers only data that does not already exist on the targetcluster. A differential replication consumes less network bandwidth than a fullreplication; however, differential replications consume more CPU. Differential replicationcan be much faster than a full replication if there is an adequate amount of available CPUfor the differential replication job to consume.

Controlling replication job resource consumptionYou can create rules that limit the network traffic created and the rate at which files aresent by replication jobs. You can also specify the number of workers that are spawned bya replication policy to limit the amount of cluster resources that are consumed. Also, youcan restrict a replication policy to connect only to a specific storage pool.

You can create network-traffic rules that control the amount of network traffic generatedby replication jobs during specified time periods. These rules can be useful if, forexample, you want to limit the amount of network traffic created during other resource-intensive operations.

You can create multiple network traffic rules to enforce different limitations at differenttimes. For example, you might allocate a small amount of network bandwidth during peakbusiness hours, but allow unlimited network bandwidth during non-peak hours.

When a replication job runs, OneFS generates workers on the source and target cluster.Workers on the source cluster send data while workers on the target cluster write data.OneFS generates no more than 40 workers for a replication job. You can modify themaximum number of workers generated per node to control the amount of resources thata replication job is allowed to consume. For example, you can increase the maximum



number of workers per node to increase the speed at which data is replicated to thetarget cluster.

You can also reduce resource consumption through file-operation rules that limit the rateat which replication policies are allowed to send files. However, it is recommended thatyou only create file-operation rules if the files you intend to replicate are predictablysimilar in size and not especially large.

Replication reportsAfter a replication job completes, SyncIQ generates a report that contains detailedinformation about the job, including how long the job ran, how much data wastransferred, and what errors occurred.

If a replication report is interrupted, SyncIQ might create a subreport about the progressof the job so far. If the job is then restarted, SyncIQ creates another subreport about theprogress of the job until the job either completes or is interrupted again. SyncIQ creates asubreport each time the job is interrupted until the job completes successfully. If multiplesubreports are created for a job, SyncIQ combines the information from the subreportsinto a single report.

SyncIQ routinely deletes replication reports. You can specify the maximum number ofreplication reports that SyncIQ retains and the length of time that SyncIQ retainsreplication reports. If the maximum number of replication reports is exceeded on acluster, SyncIQ deletes the oldest report each time a new report is created.

You cannot customize the content of a replication report.

Note

If you delete a replication policy, SyncIQ automatically deletes any reports that weregenerated for that policy.

Replication snapshotsSyncIQ generates snapshots to facilitate replication, failover, and failback between Isilonclusters. Snapshots generated by SyncIQ can also be used for archival purposes on thetarget cluster.

Source cluster snapshotsSyncIQ generates snapshots on the source cluster to ensure that a consistent point-in-time image is replicated and that unaltered data is not sent to the target cluster.

Before running a replication job, SyncIQ creates a snapshot of the source directory.SyncIQ then replicates data according to the snapshot rather than the current state of thecluster, allowing users to modify source-directory files while ensuring that an exact point-in-time image of the source directory is replicated.

For example, if a replication job of /ifs/data/dir/ starts at 1:00 PM and finishes at1:20 PM, and /ifs/data/dir/file is modified at 1:10 PM, the modifications are notreflected on the target cluster, even if /ifs/data/dir/file is not replicated until1:15 PM.

You can replicate data according to a snapshot generated with the SnapshotIQ tool. Ifyou replicate data according to a SnapshotIQ snapshot, SyncIQ does not generateanother snapshot of the source directory. This method can be useful if you want toreplicate identical copies of data to multiple Isilon clusters.


Replication reports 205

SyncIQ generates source snapshots to ensure that replication jobs do not transferunmodified data. When a job is created for a replication policy, SyncIQ checks whether itis the first job created for the policy. If it is not the first job created for the policy, SyncIQcompares the snapshot generated for the earlier job with the snapshot generated for thenew job.

SyncIQ replicates only data that has changed since the last time a snapshot wasgenerated for the replication policy. When a replication job is completed, SyncIQ deletesthe previous source-cluster snapshot and retains the most recent snapshot until the nextjob is run.

Target cluster snapshotsWhen a replication job is run, SyncIQ generates a snapshot on the target cluster tofacilitate failover operations. When the next replication job is created for the replicationpolicy, the job creates a new snapshot and deletes the old one.

If a SnapshotIQ license has been activated on the target cluster, you can configure areplication policy to generate additional snapshots that remain on the target cluster evenas subsequent replication jobs run.

SyncIQ generates target snapshots to facilitate failover on the target cluster regardless ofwhether a SnapshotIQ license has been configured on the target cluster. Failoversnapshots are generated when a replication job completes. SyncIQ retains only onefailover snapshot per replication policy, and deletes the old snapshot after the newsnapshot is created.

If a SnapshotIQ license has been activated on the target cluster, you can configureSyncIQ to generate archival snapshots on the target cluster that are not automaticallydeleted when subsequent replication jobs run. Archival snapshots contain the same dataas the snapshots that are generated for failover purposes. However, you can configurehow long archival snapshots are retained on the target cluster. You can access archivalsnapshots the same way that you access other snapshots generated on a cluster.

Data failover and failback with SyncIQSyncIQ enables you to perform automated data failover and failback operations betweenIsilon clusters. If a cluster is rendered unusable, you can fail over to another Isiloncluster, enabling clients to access to access their data on the other cluster. If theunusable cluster becomes accessible again, you can fail back to the original Isiloncluster.

For the purposes of explaining failover and failback procedures, the cluster originallyaccessed by clients is referred to as the primary cluster, and the cluster that client data isoriginally replicated to is referred to as the secondary cluster. Failover is the process thatallows clients to modify data on a secondary cluster. Failback is the process that allowsclients to access data on the primary cluster again and begins to replicate data back tothe secondary cluster.

Failover and failback can be useful in disaster recovery procedures. For example, if aprimary cluster is damaged by a natural disaster, you can migrate clients to a secondarycluster until the primary cluster is repaired and then migrate the clients back to theprimary cluster.

You can fail over and fail back to facilitate scheduled cluster maintenance. For example, ifyou are upgrading the primary cluster, you might want to migrate clients to a secondarycluster until the upgrade is complete and then migrate clients back to the primary cluster.



Note

Data failover and failback is not supported for SmartLock directories.

Data failoverData failover is the process of preparing data on a secondary cluster to be modified byclients. After you fail over to a secondary cluster, you can redirect clients to modify theirdata on the secondary cluster.

Before failover is performed, you must create and run a replication policy on the primarycluster. You initiate the failover process on the secondary cluster. Failover is performedper replication policy; to migrate data that is spread across multiple replication policies,you must initiate failover for each replication policy.

You can use any replication policy to fail over. However, if the action of the replicationpolicy is set to copy, any file that was deleted on the primary cluster will be present onthe secondary cluster. When the client connects to the secondary cluster, all files thatwere deleted on the primary cluster will be available to the client.

If you initiate failover for a replication policy while an associated replication job isrunning, the failover operation completes but the replication job fails. Because datamight be in an inconsistent state, SyncIQ uses the snapshot generated by the lastsuccessful replication job to revert data on the secondary cluster to the last recoverypoint.

If a disaster occurs on the primary cluster, any modifications to data that were made afterthe last successful replication job started are not reflected on the secondary cluster.When a client connects to the secondary cluster, their data appears as it was when thelast successful replication job was started.

Data failbackData failback is the process of restoring clusters to the roles they occupied before afailover operation. After data failback is complete, the primary cluster hosts clients andreplicates data to the secondary cluster for backup.

The first step in the failback process is updating the primary cluster with all of themodifications that were made to the data on the secondary cluster. The next step in thefailback process is preparing the primary cluster to be accessed by clients. The final stepin the failback process is resuming data replication from the primary to the secondarycluster. At the end of the failback process, you can redirect users to resume accessingtheir data on the primary cluster.

You can fail back data with any replication policy that meets all of the following criteria:

u The source directory is not a SmartLock directory.u The policy has been failed over.u The policy is a synchronization policy.u The policy does not exclude any files or directories from replication.

Recovery times and objectives for SyncIQThe Recovery Point Objective (RPO) and the Recovery Time Objective (RTO) aremeasurements of the impacts that a disaster can have on business operations. You cancalculate your RPO and RTO for a disaster recovery with replication policies.

RPO is the maximum amount of time for which data is lost if a cluster suddenly becomesunavailable. For an Isilon cluster, the RPO is the amount of time that has passed since


Data failover 207

the last completed replication job started. The RPO is never greater than the time it takesfor two consecutive replication jobs to run and complete.

If a disaster occurs while a replication job is running, the data on the secondary cluster isreverted to the state it was in when the last replication job completed. For example,consider an environment in which a replication policy is scheduled to run every threehours, and replication jobs take two hours to complete. If a disaster occurs an hour aftera replication job begins, the RPO is four hours, because it has been four hours since acompleted job began replicating data.

RTO is the maximum amount of time required to make backup data available to clientsafter a disaster. The RTO is always less than or approximately equal to the RPO,depending on the rate at which replication jobs are created for a given policy.

If replication jobs run continuously, meaning that another replication job is created forthe policy before the previous replication job completes, the RTO is approximately equalto the RPO. When the secondary cluster is failed over, the data on the cluster is reset tothe state it was in when the last job completed; resetting the data takes an amount oftime proportional to the time it took users to modify the data.

If replication jobs run on an interval, meaning that there is a period of time after areplication job completes before the next replication job for the policy starts, therelationship between RTO and RPO depends on whether a replication job is running whenthe disaster occurs. If a job is in progress when a disaster occurs, the RTO is roughlyequal to the RPO. However, if a job is not running when a disaster occurs, the RTO isnegligible because the secondary cluster was not modified since the last replication jobran, and the failover process is almost instantaneous.

SyncIQ license functionalityYou can replicate data to another Isilon cluster only if you activate a SyncIQ license onboth the local cluster and the target cluster.

If a SyncIQ license becomes inactive, you cannot create, run, or manage replicationpolicies. Also, all previously created replication policies are disabled. Replication policiesthat target the local cluster are also disabled. However, data that was previouslyreplicated to the local cluster is still available.

Creating replication policiesYou can create replication policies that determine when data is replicated with SyncIQ.

Excluding directories in replicationYou can exclude directories from being replicated by replication policies even if thedirectories exist under the specified source directory.

Note

You cannot fail back replication policies that exclude directories.

By default, all files and directories under the source directory of a replication policy arereplicated to the target cluster. However, you can prevent directories under the sourcedirectory from being replicated.

If you specify a directory to exclude, files and directories under the excluded directory arenot replicated to the target cluster. If you specify a directory to include, only the files and



directories under the included directory are replicated to the target cluster; anydirectories that are not contained in an included directory are excluded.

If you both include and exclude directories, any excluded directories must be containedin one of the included directories; otherwise, the excluded-directory setting has no effect.For example, consider a policy with the following settings:

u The root directory is /ifs/datau The included directories are /ifs/data/media/music and /ifs/data/

media/moviesu The excluded directories are /ifs/data/archive and /ifs/data/media/

music/workingIn this example, the setting that excludes the /ifs/data/archive directory has noeffect because the /ifs/data/archive directory is not under either of the includeddirectories. The /ifs/data/archive directory is not replicated regardless of whetherthe directory is explicitly excluded. However, the setting that excludes the /ifs/data/media/music/working directory does have an effect, because the directory would bereplicated if the setting was not specified.

In addition, if you exclude a directory that contains the source directory, the exclude-directory setting has no effect. For example, if the root directory of a policy is /ifs/data, explicitly excluding the /ifs directory does not prevent /ifs/data from beingreplicated.

Any directories that you explicitly include or exclude must be contained in or under thespecified root directory. For example, consider a policy in which the specified rootdirectory is /ifs/data. In this example, you could include both the /ifs/data/media and the /ifs/data/users/ directories because they are under /ifs/data.

Excluding directories from a synchronization policy does not cause the directories to bedeleted on the target cluster. For example, consider a replication policy thatsynchronizes /ifs/data on the source cluster to /ifs/data on the target cluster. Ifthe policy excludes /ifs/data/media from replication, and /ifs/data/media/file exists on the target cluster, running the policy does not cause /ifs/data/media/file to be deleted from the target cluster.

Excluding files in replicationIf you do not want specific files to be replicated by a replication policy, you can excludethem from the replication process through file-matching criteria statements. You canconfigure file-matching criteria statements during the replication-policy creation process.

Note

You cannot fail back replication policies that exclude files.

A file-criteria statement can include one or more elements. Each file-criteria elementcontains a file attribute, a comparison operator, and a comparison value. You cancombine multiple criteria elements in a criteria statement with Boolean "AND" and "OR"operators. You can configure any number of file-criteria definitions.

Configuring file-criteria statements can cause the associated jobs to run slowly. It isrecommended that you specify file-criteria statements in a replication policy only ifnecessary.

Modifying a file-criteria statement will cause a full replication to occur the next time that areplication policy is started. Depending on the amount of data being replicated, a fullreplication can take a very long time to complete.


Excluding files in replication 209

For synchronization policies, if you modify the comparison operators or comparisonvalues of a file attribute, and a file no longer matches the specified file-matching criteria,the file is deleted from the target the next time the job is run. This rule does not apply tocopy policies.

File criteria optionsYou can configure a replication policy to exclude files that meet or do not meet specificcriteria.

You can specify file criteria based on the following file attributes:

Date createdIncludes or excludes files based on when the file was created. This option isavailable for copy policies only.You can specify a relative date and time, such as "two weeks ago", or specific dateand time, such as "January 1, 2012." Time settings are based on a 24-hour clock.

Date accessedIncludes or excludes files based on when the file was last accessed. This option isavailable for copy policies only, and only if the global access-time-tracking option ofthe cluster is enabled.You can specify a relative date and time, such as "two weeks ago", or specific dateand time, such as "January 1, 2012." Time settings are based on a 24-hour clock.

Date modifiedIncludes or excludes files based on when the file was last modified. This option isavailable for copy policies only.You can specify a relative date and time, such as "two weeks ago", or specific dateand time, such as "January 1, 2012." Time settings are based on a 24-hour clock.



File nameIncludes or excludes files based on the file name. You can specify to include orexclude full or partial names that contain specific text.The following wildcard characters are accepted:

Note

Alternatively, you can filter file names by using POSIX regular-expression (regex) text.Isilon clusters support IEEE Std 1003.2 (POSIX.2) regular expressions. For moreinformation about POSIX regular expressions, see the BSD man pages.

Table 14 Replication file matching wildcards

Wildcard Description- -* Matches any string in place of the asterisk.

For example, m* matches movies and m123.

[ ] Matches any characters contained in the brackets, or a range of charactersseparated by a dash.For example, b[aei]t matches bat, bet, and bit.

For example, 1[4-7]2 matches 142, 152, 162, and 172.

You can exclude characters within brackets by following the first bracket with anexclamation mark.

For example, b[!ie] matches bat but not bit or bet.

You can match a bracket within a bracket if it is either the first or last character.

For example, [[c]at matches cat and [at.

You can match a dash within a bracket if it is either the first or last character.

For example, car[-s] matches cars and car-.

? Matches any character in place of the question mark.For example, t?p matches tap, tip, and top.

PathIncludes or excludes files based on the file path. This option is available for copypolicies only.You can specify to include or exclude full or partial paths that contain specified text.You can also include the wildcard characters *, ?, and [ ].

SizeIncludes or excludes files based on their size.

Note

File sizes are represented in multiples of 1024, not 1000.


File criteria options 211

TypeIncludes or excludes files based on one of the following file-system object types:

u Soft link

u Regular file

u Directory

Configure default replication policy settingsYou can configure default settings for replication policies. If you do not modify thesesettings when creating a replication policy, the specified default settings are applied.

Procedure

1. Click Data Protection > SyncIQ > Settings.

2. In the Default Policy Settings section, specify how you want replication policies toconnect to target clusters by selecting one of the following options:

l Click Connect to any nodes in the cluster.

l Click Connect to only the nodes in the subnet and pool if the target cluster namespecifies a SmartConnect zone.

3. Specify which nodes you want replication policies to connect to when a policy is run.

Options Description

Connect policies to all nodes ona source cluster.

Click Run the policy on all nodes in this cluster.

Connect policies only to nodescontained in a specified subnetand pool.

a. Click Run the policy only on nodes in thespecified subnet and pool.

b. From the Subnet and pool list, select thesubnet and pool .

Note

SyncIQ does not support dynamically allocated IP address pools. If a replication jobconnects to a dynamically allocated IP address, SmartConnect might reassign theaddress while a replication job is running, which would disconnect the job and causeit to fail.

4. Click Submit.

Create a replication policyYou can create a replication policy with SyncIQ that defines how and when data isreplicated to another Isilon cluster. Configuring a replication policy is a five-step process.

Configure replication policies carefully. If you modify any of the following policy settingsafter the policy is run, OneFS performs either a full or differential replication the next timethe policy is run:

u Source directory

u Included or excluded directories

u File-criteria statement



u Target cluster name or address

This applies only if you target a different cluster. If you modify the IP or domain nameof a target cluster, and then modify the replication policy on the source cluster tomatch the new IP or domain name, a full replication is not performed.

u Target directory

Configure basic policy settingsYou must configure basic settings for a replication policy.

Procedure

1. Click Data Protection > SyncIQ > Policies.

2. Click Create a SyncIQ policy.

3. In the Settings area, in the Policy name field, type a name for the replication policy.

4. Optional: In the Description field, type a description for the replication policy.

5. In the Action area, specify the type of replication policy.

l To copy all files from the source directory to the target directory, click Copy.

Note

Failback is not supported for copy policies.

l To copy all files from the source directory to the target directory and delete anyfiles on the target directory that are not in the source directory, click Synchronize.

6. In the Run job area, specify whether replication jobs will be run.

Options Description

Run jobs only when manuallyinitiated by a user.

Click Only manually.

Run jobs automaticallyaccording to a schedule.

a. Click On a schedule.

b. Specify a schedule.

If you configure a replication policy to run morethan once a day, you cannot configure the intervalto span across two calendar days. For example,you cannot configure a replication policy to runevery hour starting at 7:00 PM and ending at 1:00AM.

Run jobs automatically everytime a change is made to thesource directory.

Click Whenever the source is modified.

After you finish

The next step in the process of creating a replication policy is specifying sourcedirectories and files.


Create a replication policy 213

Specify source directories and filesYou must specify the directories and files you want to replicate.

Procedure

1. In the Source Cluster area, in the Source Root Directory field, type the full path of thesource directory that you want to replicate to the target cluster.

You must specify a directory contained in /ifs. You cannot specify the /ifs/.snapshot directory or subdirectory of it.

2. Optional: Prevent specific subdirectories of the root directory from being replicated.

l To include a directory, in the Included Directories area, click Add a directory path.

l To exclude a directory, in the Excluded Directories area, click Add a directory path.

3. Optional: Prevent specific files from being replicated by specifying file matchingcriteria.

a. In the File Matching Criteria area, select a filter type.

b. Select an operator.

c. Type a value.

Files that do not meet the specified criteria will not be replicated to the target cluster.For example, if you specify File Type doesn't match .txt, SyncIQ will notreplicate any files with the .txt file extension. If you specify Created after08/14/2013, SyncIQ will not replicate any files created before August 14th, 2013.If you want to specify more than one file matching criterion, you can control how thecriteria relate to each other by clicking either Add an "Or" condition or Add an "And"condition.

4. Specify which nodes you want the replication policy to connect to when the policy isrun.

Options Description

Connect the policy to all nodes inthe source cluster.

Click Run the policy on all nodes in this cluster.

Connect the policy only to nodescontained in a specified subnetand pool.

a. Click Run the policy only on nodes in thespecified subnet and pool.

b. From the Subnet and pool list, select thesubnet and pool .

Note


After you finish

The next step in the process of creating a replication policy is specifying the targetdirectory.



Specify the policy target directoryYou must specify a target cluster and directory to replicate data to.

Procedure

1. In the Target Cluster area, in the Target Host field, type one of the following:

l The fully qualified domain name of any node in the target cluster.

l The host name of any node in the target cluster.

l The name of a SmartConnect zone in the target cluster.

l The IPv4 or IPv6 address of any node in the target cluster.

l localhostThis will replicate data to another directory on the local cluster.

Note


2. In the Target Directory field, type the absolute path of the directory on the targetcluster that you want to replicate data to.

CAUTION

If you specify an existing directory on the target cluster, ensure that the directory isnot the target of another replication policy. If this is a synchronization policy, ensurethat the directory is empty. All files are deleted from the target of a synchronizationpolicy the first time the policy is run.

If the specified target directory does not already exist on the target cluster, thedirectory is created the first time the job is run. It is recommended that you do notspecify the /ifs directory. If you specify the /ifs directory, the entire target clusteris set to a read-only state, preventing you from storing any other data on the cluster.

If this is a copy policy, and files in the target directory share the same name as files inthe source directory, the target directory files are overwritten when the job is run.

3. If you want replication jobs to connect only to the nodes included in the SmartConnectzone specified by the target cluster, click Connect only to the nodes within the targetcluster SmartConnect Zone.

After you finish

The next step in the process of creating a replication policy is specifying policy targetsnapshot settings.

Configure policy target snapshot settingsYou can optionally specify how archival snapshots are generated on the target cluster.You can access archival snapshots the same way that you access SnapshotIQ snapshots.

SyncIQ always retains one snapshot on the target cluster to facilitate failover, regardlessof these settings.


Create a replication policy 215

Procedure

1. To create archival snapshots on the target cluster, in the Target Snapshots area, clickCapture snapshots on the target cluster.

2. Optional: To modify the default alias of the last snapshot created according to thereplication policy, in the Snapshot Alias Name field, type a new alias.

You can specify the alias name as a snapshot naming pattern. For example, thefollowing naming pattern is valid:

%{PolicyName}-on-%{SrcCluster}-latest

The previous example produces names similar to the following:

newPolicy-on-Cluster1-latest3. Optional: To modify the snapshot naming pattern for snapshots created according to

the replication policy, in the Snapshot Naming Pattern field, type a naming pattern.Each snapshot generated for this replication policy is assigned a name based on thispattern.

For example, the following naming pattern is valid:

%{PolicyName}-from-%{SrcCluster}-at-%H:%M-on-%m-%d-%Y

The example produces names similar to the following:

newPolicy-from-Cluster1-at-10:30-on-7-12-20124. Select one of the following options:

l Click Snapshots do not expire.

l Click Snapshots expire after... and specify an expiration period.

After you finish

The next step in the process of creating a replication policy is configuring advancedpolicy settings.

Configure advanced policy settingsYou can optionally configure advanced settings for a replication policy.

Procedure

1. Optional: In the Worker Threads Per Node field, specify the maximum number ofconcurrent processes per node that will perform replication operations.

Note

Do not modify the default setting without consulting Isilon Technical Support.

2. Optional: From the Log Level list, select the level of logging you want SyncIQ toperform for replication jobs.

The following log levels are valid, listed from least to most verbose:

l Click Error.

l Click Notice.

l Click Network Activity.

l Click File Activity.

3. Optional: If you want SyncIQ to perform a checksum on each file data packet that isaffected by the replication policy, select the Validate File Integrity check box.

If you enable this option, and the checksum values for a file data packet do notmatch, SyncIQ retransmits the affected packet.



4. Optional: To modify the length of time SyncIQ retains replication reports for the policy,in the Keep Reports For area, specify a length of time.

After the specified expiration period has passed for a report, SyncIQ automaticallydeletes the report.

Some units of time are displayed differently when you view a report than how theywere originally entered. Entering a number of days that is equal to a correspondingvalue in weeks, months, or years results in the larger unit of time being displayed. Forexample, if you enter a value of 7 days, 1 week appears for that report after it iscreated. This change occurs because SyncIQ internally records report retention timesin seconds and then converts them into days, weeks, months, or years.

5. Optional: Specify whether to record information about files that are deleted byreplication jobs by selecting one of the following options:

l Click Record when a synchronization deletes files or directories.

l Click Do not record when a synchronization deletes files or directories.

This option is applicable for synchronization policies only.

After you finish

The next step in the process of creating a replication policy is saving the replicationpolicy settings.

Save replication policy settingsSyncIQ does not create replication jobs for a replication policy until you save the policy.

Before you begin

Review the current settings of the replication policy. If necessary, modify the policysettings.

Procedure

1. Click Create Policy.

After you finish

You can increase the speed at which you can failback a replication policy by creating aSyncIQ domain for the source directory of the policy.

Create a SyncIQ domainYou can create a SyncIQ domain to increase the speed at which failback is performed fora replication policy. Because you can fail back only synchronization policies, it is notnecessary to create SyncIQ domains for copy policies.

Failing back a replication policy requires that a SyncIQ domain be created for the sourcedirectory. OneFS automatically creates a SyncIQ domain during the failback process.However, if you intend on failing back a replication policy, it is recommended that youcreate a SyncIQ domain for the source directory of the replication policy while thedirectory is empty. Creating a domain for a directory that contains less data takes lesstime.

Procedure



3. In the Domain Root Path field, type the path of a source directory of a replicationpolicy.


Create a SyncIQ domain 217

4. From the Type of domain list, select SyncIQ.

5. Ensure that the Delete domain check box is cleared.

6. Click Start Job.

Assess a replication policyBefore running a replication policy for the first time, you can view statistics on the filesthat would be affected by the replication without transferring any files. This can be usefulif you want to preview the size of the data set that will be transferred if you run the policy.

Note

You can assess only replication policies that have never been run before.

Procedure


2. In the SyncIQ Policies table, in the row of a replication policy, from the Actionscolumn, select Assess Sync.

3. Click Data Protection > SyncIQ > Summary.

4. After the job completes, in the SyncIQ Recent Reports table, in the row of thereplication job, click View Details.

The report displays the total amount of data that would have been transferred in theTotal Data field.

Managing replication to remote clustersYou can manually run, view, assess, pause, resume, cancel, resolve, and reset replicationjobs that target other clusters.

After a policy job starts, you can pause the job to suspend replication activities.Afterwards, you can resume the job, continuing replication from the point where the jobwas interrupted. You can also cancel a running or paused replication job if you want tofree the cluster resources allocated for the job. A paused job reserves cluster resourceswhether or not the resources are in use. A cancelled job releases its cluster resources andallows another replication job to consume those resources. No more than five runningand paused replication jobs can exist on a cluster at a time. However, an unlimitednumber of canceled replication jobs can exist on a cluster. If a replication job remainspaused for more than a week, SyncIQ automatically cancels the job.

Start a replication jobYou can manually start a replication job for a replication policy at any time.

If you want to replicate data according to an existing snapshot, at the OneFS commandprompt, run the isi sync jobs start command with the --source-snapshotoption. You cannot replicate data according to snapshots generated by SyncIQ.

Procedure


2. In the SyncIQ Policies table, in the Actions column for a job, select Start Job.



Pause a replication jobYou can pause a running replication job and then resume the job later. Pausing areplication job temporarily stops data from being replicated, but does not free the clusterresources replicating the data.

Procedure


2. In the Active Jobs table, in the Actions column for a job, click Pause Running Job.

Resume a replication jobYou can resume a paused replication job.

Procedure


2. In the Currently Running table, in the Actions column for a job, click Resume RunningJob.

Cancel a replication jobYou can cancel a running or paused replication job. Cancelling a replication job stopsdata from being replicated and frees the cluster resources that were replicating data. Youcannot resume a cancelled replication job. To restart replication, you must start thereplication policy again.

Procedure


2. In the Active Jobs table, in the Actions column for a job, click Cancel Running Job.

View active replication jobsYou can view information about replication jobs that are currently running or paused.

Procedure


2. In the Active Jobs table, review information about active replication jobs.

Replication job informationYou can view information about replication jobs through the Active Jobs table.

StatusThe status of the job. The following job statuses are possible:

RunningThe job is currently running without error.

PausedThe job has been temporarily paused.

Policy NameThe name of the associated replication policy.

StartedThe time the job started.


Pause a replication job 219

ElapsedHow much time has elapsed since the job started.

TransferredThe number of files that have been transferred, and the total size of all transferredfiles.

Source DirectoryThe path of the source directory on the source cluster.

Target HostThe target directory on the target cluster.

ActionsDisplays any job-related actions that you can perform.

Initiating data failover and failback with SyncIQYou can fail over from one Isilon cluster to another if, for example, a cluster becomesunavailable. You can then fail back to a primary cluster if the primary cluster becomesavailable again. You can revert failover if you decide that the failover was unnecessary, orif you failed over for testing purposes.

If you fail over a scheduled replication policy on the secondary cluster, and thecorresponding policy on the primary cluster runs a replication job, the job might fail andthe policy might be set to an unrunnable state. To resolve this, modify the replicationpolicy so that it is set to run only manually, resolve the policy, and complete the failbackprocess. After you complete the failback process, you can modify the policy to runaccording to a schedule again.

Note

Although you cannot fail over or fail back SmartLock directories, you can recoverSmartLock directories on a target cluster. After you recover SmartLock directories, you canmigrate them back to the source cluster.

Fail over data to a secondary clusterYou can fail over to a secondary Isilon cluster if, for example, a cluster becomesunavailable.

Before you begin

Create and successfully run a replication policy.

Complete the following procedure for each replication policy that you want to fail over.

Procedure

1. On the secondary Isilon cluster, click Data Protection > SyncIQ > Local Targets.

2. In the SyncIQ Local Targets table, in the row for a replication policy, from the Actionscolumn, select Allow Writes.

After you finish

Direct clients to begin accessing the secondary cluster.

Revert a failover operationFailover reversion undoes a failover operation on a secondary cluster, enabling you toreplicate data from the primary cluster to the secondary cluster again. Failover reversion



is useful if the primary cluster becomes available before data is modified on thesecondary cluster or if you failed over to a secondary cluster for testing purposes.

Before you begin

Fail over a replication policy.

Reverting a failover operation does not migrate modified data back to the primary cluster.To migrate data that clients have modified on the secondary cluster, you must fail back tothe primary cluster.

Complete the following procedure for each replication policy that you want to fail over:

Procedure

1. Click Data Protection > SyncIQ > Local Targets.

2. In the SyncIQ Local Targets table, in the row for a replication policy, from the Actionscolumn, select Disallow Writes.

Fail back data to a primary clusterAfter you fail over to a secondary cluster, you can fail back to the primary cluster.

Before you begin

Fail over a replication policy.

Procedure

1. On the primary cluster, click Data Protection > SyncIQ > Policies.

2. In the SyncIQ Policies table, in the row for a replication policy, from the Actionscolumn, select Resync-prep.

SyncIQ creates a mirror policy for each replication policy on the secondary cluster.

SyncIQ names mirror policies according to the following pattern:

<replication-policy-name>_mirror3. On the secondary cluster, replicate data to the primary cluster by using the mirror

policies.

You can replicate data either by manually starting the mirror policies or by modifyingthe mirror policies and specifying a schedule.

4. Prevent clients from accessing the secondary cluster and then run each mirror policyagain.

To minimize impact to clients, it is recommended that you wait until client access islow before preventing client access to the cluster.

5. On the primary cluster, click Data Protection > SyncIQ > Local Targets.

6. In the SyncIQ Local Targets table, from the Actions column, select Allow Writes foreach mirror policy.

7. On the secondary cluster, click Data Protection > SyncIQ > Policies.

8. In the SyncIQ Policies table, from the Actions column, select Resync-prep for eachmirror policy.

After you finish

Redirect clients to begin accessing the primary cluster.


Fail back data to a primary cluster 221

Performing disaster recovery for SmartLock directoriesAlthough you cannot fail over or fail back SmartLock directories, you can recoverSmartLock directories on a target cluster. After you recover SmartLock directories, you canmigrate them back to the source cluster.

Recover SmartLock directories on a target clusterYou can recover SmartLock directories that you have replicated to a target cluster.

Before you begin

Create and successfully run a replication policy.

Complete the following procedure for each SmartLock directory that you want to recover.

Procedure

1. On the target cluster, click Data Protection > SyncIQ > Local Targets.

2. In the SyncIQ Local Targets table, in the row of the replication policy, enable writes tothe target directory of the policy.

l If the last replication job completed successfully and a replication job is notcurrently running, select Allow Writes.

l If a replication job is currently running, wait until the replication job completes,and then select Allow Writes.

l If the primary cluster became unavailable while a replication job was running,select Break Association.

3. If you clicked Break Association, restore any files that are left in an inconsistent state.

a. Delete all files that are not committed to a WORM state from the target directory.

b. Copy all files from the failover snapshot to the target directory.

Failover snapshots are named according to the following naming pattern:

SIQ-Failover-<policy-name>-<year>-<month>-<day>_<hour>-<minute>-<second>

Snapshots are stored in the /ifs/.snapshot directory.

4. If any SmartLock directory configuration settings, such as an autocommit time period,were specified for the source directory of the replication policy, apply those settings tothe target directory.

Because autocommit information is not transferred to the target cluster, files thatwere scheduled to be committed to a WORM state on the source cluster will not bescheduled to be committed at the same time on the target cluster. To ensure that allfiles are retained for the appropriate time period, you can commit all files in targetSmartLock directories to a WORM state. For example, the following commandautomatically commits all files in /ifs/data/smartlock to a WORM state afterone minute.isi smartlock modify --path /ifs/data/smartlock --autocommit 1n

After you finish

Redirect clients to begin accessing the target cluster.



Migrate SmartLock directoriesYou might want to migrate SmartLock directories if you restored the directories on atarget cluster, and want to transfer those directories either back to the source cluster orto a new cluster.

Procedure

1. On a cluster, create a replication policy for each policy that you want to migrate.

The policies must meet the following requirements:

l The source directory is the SmartLock directory that you are migrating.

l The target directory is an empty SmartLock directory. The source and targetdirectories must be of the same SmartLock type. For example, if the targetdirectory is a compliance directory, the source must also be a compliancedirectory.

2. Replicate data to the target cluster by running the policies you created.

You can replicate data either by manually starting the policies or by specifying a policyschedule.

3. Optional: To ensure that SmartLock protection is enforced for all files, commit all filesin the SmartLock source directory to a WORM state.

Because autocommit information is not transferred to the target cluster, files thatwere scheduled to be committed to a WORM state on the source cluster will not bescheduled to be committed at the same time on the target cluster. To ensure that allfiles are retained for the appropriate time period, you can commit all files in targetSmartLock directories to a WORM state.

For example, the following command automatically commits all files in /ifs/data/smartlock to a WORM state after one minute.isi smartlock modify --path /ifs/data/smartlock --autocommit 1n

This step is unnecessary if you have not configured an autocommit time period for theSmartLock directory being replicated.

4. Prevent clients from accessing the source cluster and run the policy that you created.

To minimize impact to clients, it is recommended that you wait until client access islow before preventing client access to the cluster.

5. On the target cluster, click Data Protection > SyncIQ > Local Targets.

6. In the SyncIQ Local Targets table, in the row of each replication policy, from theActions column, select Allow Writes.

7. Optional: If any SmartLock directory configuration settings, such as an autocommittime period, were specified for the source directories of the replication policies, applythose settings to the target directories.

8. Optional: Delete the copy of your SmartLock data on the source cluster.

If the SmartLock directories are compliance directories or enterprise directories withthe privileged delete functionality permanently disabled, you cannot recover thespace consumed by the source SmartLock directories until all files are released from aWORM state. If you want to free the space before files are released from a WORMstate, contact Isilon Technical Support for information about reformatting your cluster.


Migrate SmartLock directories 223

Managing replication policiesYou can modify, view, enable and disable replication policies.

Modify a replication policyYou can modify the settings of a replication policy.

If you modify any of the following policy settings after a policy runs, OneFS performseither a full or differential replication the next time the policy runs:

u Source directory

u Included or excluded directories

u File-criteria statement

u Target clusterThis applies only if you target a different cluster. If you modify the IP or domain nameof a target cluster, and then modify the replication policy on the source cluster tomatch the new IP or domain name, a full replication is not performed.

u Target directory

Procedure


2. In the SyncIQ Policies table, in the row for a policy, click View/Edit.

3. In the View SyncIQ Policy Details dialog box, click Edit Policy.

4. Modify the settings of the replication policy, and then click Save Changes.

Delete a replication policyYou can delete a replication policy. Once a policy is deleted, SyncIQ no longer createsreplication jobs for the policy. Deleting a replication policy breaks the target associationon the target cluster, and allows writes to the target directory.

If you want to temporarily suspend a replication policy from creating replication jobs, youcan disable the policy, and then enable the policy again later.

Procedure


2. In the SyncIQ Policies table, in the row for a policy, select Delete Policy.


Enable or disable a replication policyYou can temporarily suspend a replication policy from creating replication jobs, and thenenable it again later.

Note

If you disable a replication policy while an associated replication job is running, therunning job is not interrupted. However, the policy will not create another job until thepolicy is enabled.



Procedure


2. In the SyncIQ Policies table, in the row for a replication policy, select either EnablePolicy or Disable Policy.

If neither Enable Policy nor Disable Policy appears, verify that a replication job is notrunning for the policy. If an associated replication job is not running, ensure that theSyncIQ license is active on the cluster.

View replication policiesYou can view information about replication policies.

Procedure


2. In the SyncIQ Policies table, review information about replication policies.

Replication policy informationYou can view information about replication policies through the SyncIQ Policies table.

Policy NameThe name of the policy.

StateWhether the policy is enabled or disabled.

Last Known GoodWhen the last successful job ran.

ScheduleWhen the next job is scheduled to run. A value of Manual indicates that the job canbe run only manually. A value of When source is modified indicates that the job willbe run whenever changes are made to the source directory.

Source DirectoryThe path of the source directory on the source cluster.

Target Host : DirectoryThe IP address or fully qualified domain name of the target cluster and the full pathof the target directory.

ActionsAny policy-related actions that you can perform.

Replication policy settingsYou configure replication policies to run according to replication policy settings.

Policy nameThe name of the policy.

DescriptionDescribes the policy. For example, the description might explain the purpose orfunction of the policy.

EnabledDetermines whether the policy is enabled.


View replication policies 225

ActionDetermines the how the policy replicates data. All policies copy files from the sourcedirectory to the target directory and update files in the target directory to match fileson the source directory. The action determines how deleting a file on the sourcedirectory affects the target. The following values are valid:

CopyIf a file is deleted in the source directory, the file is not deleted in the targetdirectory.

SynchronizeDeletes files in the target directory if they are no longer present on the source.This ensures that an exact replica of the source directory is maintained on thetarget cluster.

Run jobDetermines whether jobs are run automatically according to a schedule or only whenmanually specified by a user.

Last Successful RunDisplays the last time that a replication job for the policy completed successfully.

Last StartedDisplays the last time that the policy was run.

Source Root DirectoryThe full path of the source directory. Data is replicated from the source directory tothe target directory.

Included DirectoriesDetermines which directories are included in replication. If one or more directoriesare specified by this setting, any directories that are not specified are not replicated.

Excluded DirectoriesDetermines which directories are excluded from replication. Any directories specifiedby this setting are not replicated.

File Matching CriteriaDetermines which files are excluded from replication. Any files that do not meet thespecified criteria are not replicated.

Restrict Source NodesDetermines whether the policy can run on all nodes on the source cluster or run onlyon specific nodes.

Target HostThe IP address or fully qualified domain name of the target cluster.

Target DirectoryThe full path of the target directory. Data is replicated to the target directory from thesource directory.

Restrict Target NodesDetermines whether the policy can connect to all nodes on the target cluster or canconnect only to specific nodes.

Capture SnapshotsDetermines whether archival snapshots are generated on the target cluster.

Snapshot Alias NameSpecifies an alias for the latest archival snapshot taken on the target cluster.

Snapshot Naming PatternSpecifies how archival snapshots are named on the target cluster.



Snapshot ExpirationSpecifies how long archival snapshots are retained on the target cluster before theyare automatically deleted by the system.

Workers Threads Per NodeSpecifies the number of workers per node that are generated by OneFS to performeach replication job for the policy.

Log LevelSpecifies the amount of information that is recorded for replication jobs.More verbose options include all information from less verbose options. Thefollowing list describes the log levels from least to most verbose:

NoticeIncludes job and process-level activity, including job starts, stops, and workercoordination information. This is the recommended log level.

ErrorIncludes events related to specific types of failures.

Network ActivityIncludes more job-level activity and work-item information, including specificpaths and snapshot names.

File ActivityIncludes a separate event for each action taken on a file. Do not select thisoption without first consulting Isilon Technical Support.

Replication logs are typically used for debugging purposes. If necessary, you can login to a node through the command-line interface and view the contents ofthe /var/log/isi_migrate.log file on the node.

Validate File IntegrityDetermines whether OneFS performs a checksum on each file data packet that isaffected by a replication job. If a checksum value does not match, OneFS retransmitsthe affected file data packet.

Keep Reports ForSpecifies how long replication reports are kept before they are automatically deletedby OneFS.

Log Deletions on SynchronizationDetermines whether OneFS records when a synchronization job deletes files ordirectories on the target cluster.

The following replication policy fields are available only through the OneFS command-lineinterface.

Source SubnetSpecifies whether replication jobs connect to any nodes in the cluster or if jobs canconnect only to nodes in a specified subnet.

Source PoolSpecifies whether replication jobs connect to any nodes in the cluster or if jobs canconnect only to nodes in a specified pool.

Password SetSpecifies a password to access the target cluster.

Report Max CountSpecifies the maximum number of replication reports that are retained for thispolicy.


Replication policy settings 227

Target Compare Initial SyncDetermines whether full or differential replications are performed for this policy. Fullor differential replications are performed the first time a policy is run and after apolicy is reset.

Source Snapshot ArchiveDetermines whether snapshots generated for the replication policy on the sourcecluster are deleted when the next replication policy is run. Enabling archival sourcesnapshots does not require you to activate the SnapshotIQ license on the cluster.

Source Snapshot PatternIf snapshots generated for the replication policy on the source cluster are retained,renames snapshots according to the specified rename pattern.

Source Snapshot ExpirationIf snapshots generated for the replication policy on the source cluster are retained,specifies an expiration period for the snapshots.

Restrict Target NetworkDetermines whether replication jobs connect only to nodes in a given SmartConnectzone. This setting applies only if the Target Host is specified as a SmartConnectzone.

Target Detect ModificationsDetermines whether SyncIQ checks the target directory for modifications beforereplicating files. By default, SyncIQ always checks for modifications.

Note

Disabling this option could result in data loss. It is recommended that you consultIsilon Technical Support before disabling this option.

ResolveDetermines whether you can manually resolve the policy if a replication jobencounters an error.

Managing replication to the local clusterYou can interrupt replication jobs that target the local cluster.

You can cancel a currently running job that targets the local cluster, or you can break theassociation between a policy and its specified target. Breaking a source and targetcluster association causes SyncIQ to perform a full replication the next time the policy isrun.

Cancel replication to the local clusterYou can cancel a replication job that is targeting the local cluster.

Procedure


2. In the SyncIQ Local Targets table, specify whether to cancel a specific replication jobor all replication jobs targeting the local cluster.

l To cancel a specific job, in the row for a replication job, select Cancel Running Job.

l To cancel all jobs targeting the local cluster, select the check box to the left ofPolicy Name and then select Cancel Selection from the Select a bulk action list.



Break local target associationYou can break the association between a replication policy and the local cluster.Breaking the target association will allow writes to the target directory but will alsorequire you to reset the replication policy before you can run the policy again.

CAUTION

After a replication policy is reset, SyncIQ performs a full or differential replication thenext time the policy is run. Depending on the amount of data being replicated, a full ordifferential replication can take a very long time to complete.

Procedure


2. In the SyncIQ Local Targets table, in the row for a replication policy, select BreakAssociation.

3. In the Confirm dialog box, click Yes.

View replication policies targeting the local clusterYou can view information about replication policies that are currently replicating data tothe local cluster.

Procedure


2. In the SyncIQ Local Targets table, view information about replication policies.

Remote replication policy informationYou can view information about replication policies that are currently targeting the localcluster.

The following information is displayed in the SyncIQ Local Targets table:

IDThe ID of the replication policy.

Policy NameThe name of the replication policy.

Source HostThe name of the source cluster.

Source Cluster GUIDThe GUID of the source cluster.

Coordinator IPThe IP address of the node on the source cluster that is acting as the job coordinator.

UpdatedThe time when data about the policy or job was last collected from the sourcecluster.

Target PathThe path of the target directory on the target cluster.

StatusThe current status of the replication job.


Break local target association 229

ActionsDisplays any job-related actions that you can perform.

Managing replication performance rulesYou can manage the impact of replication on cluster performance by creating rules thatlimit the network traffic created and the rate at which files are sent by replication jobs.

Create a network traffic ruleYou can create a network traffic rule that limits the amount of network traffic thatreplication policies are allowed to generate during a specified time period.

Procedure

1. Click Data Protection > SyncIQ > Performance Rules.

2. Click Create a SyncIQ Performance Rule.

3. From the Rule Type list, select Bandwidth.

4. In the Limit field, specify the maximum number of bytes per second that replicationpolicies are allowed to send.

5. In the Schedule area, specify the time and days of the week that you want to apply therule.

6. Click Create Performance Rule.

Create a file operations ruleYou can create a file-operations rule that limits the number of files that replication jobscan send per second.

Procedure


2. Click Create a SyncIQ Performance Rule.

3. From the Rule Type list, select Bandwidth.

4. In the Limit field, specify the maximum number of files per second that replicationpolicies are allowed to send.

5. In the Schedule area, specify the time and days of the week that you want to apply therule.

6. Click Create Performance Rule.

Modify a performance ruleYou can modify a performance rule.

Procedure


2. In the SyncIQ Performance Rules, in the row for the rule you want to modify, clickView/Edit.

3. Click Edit Performance Rule.

4. Modify rule settings, and then click Save Changes.



Delete a performance ruleYou can delete a performance rule.

Procedure


2. In the SyncIQ Performance Rules table, in the row for the rule you want to delete,select Delete Rule.

3. In the Confirm Delete dialog box, click Delete.

Enable or disable a performance ruleYou can disable a performance rule to temporarily prevent the rule from being enforced.You can also enable a performance rule after it has been disabled.

Procedure


2. In the SyncIQ Performance Rules table, in the row for a rule you want to enable ordisable, select either Enable Rule or Disable Rule.

View performance rulesYou can view information about replication performance rules.

Procedure


2. In the SyncIQ Performance Rules table, view information about performance rules.

Managing replication reportsIn addition to viewing replication reports, you can configure how long reports are retainedon the cluster. You can also delete any reports that have passed their expiration period.

Configure default replication report settingsYou can configure the default amount of time that SyncIQ retains replication reports for.You can also configure the maximum number of reports that SyncIQ retains for eachreplication policy.

Procedure

1. Click Data Protection > SyncIQ > Settings.

2. In the Report Settings area, in the Keep Reports For area, specify how long you wantto retain replication reports for.

After the specified expiration period has passed for a report, SyncIQ automaticallydeletes the report.

Some units of time are displayed differently when you view a report than how youoriginally enter them. Entering a number of days that is equal to a correspondingvalue in weeks, months, or years results in the larger unit of time being displayed. Forexample, if you enter a value of 7 days, 1 week appears for that report after it iscreated. This change occurs because SyncIQ internally records report retention timesin seconds and then converts them into days, weeks, months, or years for display.


Delete a performance rule 231

3. In the Number of Reports to Keep Per Policy field, type the maximum number ofreports you want to retain at a time for a replication policy.

4. Click Submit.

Delete replication reportsReplication reports are routinely deleted by SyncIQ after the expiration date for thereports has passed. SyncIQ also deletes reports after the number of reports exceeds thespecified limit. Excess reports are periodically deleted by SyncIQ; however, you canmanually delete all excess replication reports at any time. This procedure is availableonly through the command-line interface (CLI).

Procedure


2. Delete excess replication reports by running the following command:

isi sync reports rotate

View replication reportsYou can view replication reports and subreports.

Procedure

1. Click Data Protection > SyncIQ > Reports.

2. In the SyncIQ Reports table, in the row for a report, click View Details.

If a report is composed of subreports, the report is displayed as a folder. Subreportsare displayed as files within report folders.

Replication report informationYou can view information about replication jobs through the Reports table.

Policy NameThe name of the associated policy for the job. You can view or edit settings for thepolicy by clicking the policy name.

StartedIndicates when the job started.

EndedIndicates when the job ended.

DurationIndicates how long the job took to complete.

TransferredThe total number of files that were transferred during the job run, and the total sizeof all transferred files. For assessed policies, Assessment appears.



Sync TypeThe action that was performed by the replication job.

Initial SyncIndicates that either a differential or a full replication was performed.

Incremental SyncIndicates that only modified files were transferred to the target cluster.

Failover / Failback Allow WritesIndicates that writes were enabled on a target directory of a replication policy.

Failover / Failback Disallow WritesIndicates that an allow writes operation was undone.

Failover / Failback Resync PrepIndicates that an association between files on the source cluster and files onthe target cluster was created. This is the first step in the failback preparationprocess.

Failover / Failback Resync Prep Domain MarkIndicates that a SyncIQ domain was created for the source directory. This is thesecond step in the failback preparation process.

Failover / Failback Resync Prep RestoreIndicates that a source directory was restored to the last recovery point. This isthe third step in the failback preparation process.

Failover / Failback Resync Prep FinalizeIndicates that a mirror policy was created on the target cluster. This is the laststep in the failback preparation process.

UpgradeIndicates that a policy-conversion replication occurred after upgrading theOneFS operating system or merging policies.

SourceThe path of the source directory on the source cluster.

TargetThe IP address or fully qualified domain name of the target cluster.

ActionsDisplays any report-related actions that you can perform.

Managing failed replication jobsIf a replication job fails due to an error, SyncIQ might disable the correspondingreplication policy. For example SyncIQ might disable a replication policy if the IP orhostname of the target cluster is modified. If a replication policy is disabled, the policycannot be run.

To resume replication for a disabled policy, you must either fix the error that caused thepolicy to be disabled, or reset the replication policy. It is recommended that you attemptto fix the issue rather than reset the policy. If you believe you have fixed the error, youcan return the replication policy to an enabled state by resolving the policy. You can thenrun the policy again to test whether the issue was fixed. If you are unable to fix the issue,


Managing failed replication jobs 233

you can reset the replication policy. However, resetting the policy causes a full ordifferential replication to be performed the next time the policy is run.

Note

Depending on the amount of data being synchronized or copied, a full and differentialreplications can take a very long time to complete.

Resolve a replication policyIf SyncIQ disables a replication policy due to a replication error, and you fix the issue thatcaused the error, you can resolve the replication policy. Resolving a replication policyenables you to run the policy again. If you cannot resolve the issue that caused the error,you can reset the replication policy.

Procedure


2. In the Policies table, in the row for a policy, select Resolve.

Reset a replication policyIf a replication job encounters an error that you cannot resolve, you can reset thecorresponding replication policy. Resetting a policy causes OneFS to perform a full ordifferential replication the next time the policy is run. Resetting a replication policydeletes the latest snapshot generated for the policy on the source cluster.

CAUTION

Depending on the amount of data being replicated, a full or differential replication cantake a very long time to complete. Reset a replication policy only if you cannot fix theissue that caused the replication error. If you fix the issue that caused the error, resolvethe policy instead of resetting the policy.

Procedure


2. In the SyncIQ Policies table, in the row for a policy, select Reset Sync State.

Perform a full or differential replicationAfter you reset a replication policy, you must perform either a full or differentialreplication.

Before you begin

Reset a replication policy.

Procedure

1. Open a secure shell (SSH) connection to any node in the cluster and log in through theroot or compliance administrator account.

2. Specify the type of replication you want to perform by running the isi syncpolicies modify command.

l To perform a full replication, disable the --target-compare-initial-syncoption.



For example, the following command disables differential synchronization fornewPolicy:

isi sync policies modify newPolicy \--target-compare-initial-sync false

l To perform a differential replication, enable the --target-compare-initial-sync option.

For example, the following command enables differential synchronization fornewPolicy:

isi sync policies modify newPolicy \--target-compare-initial-sync true

3. Run the policy by running the isi sync jobs start command.

For example, the following command runs newPolicy:isi sync jobs start newPolicy


Perform a full or differential replication 235

CHAPTER 12

Data layout with FlexProtect


u FlexProtect overview............................................................................................238u File striping......................................................................................................... 238u Requested data protection.................................................................................. 238u FlexProtect data recovery.....................................................................................239u Requesting data protection................................................................................. 240u Requested protection settings.............................................................................240u Requested protection disk space usage.............................................................. 241

Data layout with FlexProtect 237

FlexProtect overviewAn Isilon cluster is designed to continuously serve data, even when one or morecomponents simultaneously fail. OneFS ensures data availability by striping or mirroringdata across the cluster. If a cluster component fails, data stored on the failed componentis available on another component. After a component failure, lost data is restored onhealthy components by the FlexProtect proprietary system.

Data protection is specified at the file level, not the block level, enabling the system torecover data quickly. Because all data, metadata, and parity information is distributedacross all nodes, the cluster does not require a dedicated parity node or drive. Thisensures that no single node limits the speed of the rebuild process.

File stripingOneFS uses the internal network to automatically allocate and stripe data across nodesand disks in the cluster. OneFS protects data as the data is being written. No separateaction is necessary to stripe data.

OneFS breaks files into smaller logical chunks called stripes before writing the files todisk; the size of each file chunk is referred to as the stripe unit size. Each OneFS block is8 KB, and a stripe unit consists of 16 blocks, for a total of 128 KB per stripe unit. During awrite, OneFS breaks data into stripes and then logically places the data in a stripe unit.As OneFS stripes data across the cluster, OneFS fills the stripe unit according to thenumber of nodes and protection level.

OneFS can continuously reallocate data and make storage space more usable andefficient. As the cluster size increases, OneFS stores large files more efficiently.

Requested data protectionThe requested protection of data determines the amount of redundant data created onthe cluster to ensure that data is protected against component failures. OneFS enablesyou to modify the requested protection in real time while clients are reading and writingdata on the cluster.

OneFS provides several data protection settings. You can modify these protectionsettings at any time without rebooting or taking the cluster or file system offline. Whenplanning your storage solution, keep in mind that increasing the requested protectionreduces write performance and requires additional storage space for the increasednumber of nodes.

OneFS uses the Reed Solomon algorithm for N+M protection. In the N+M data protectionmodel, N represents the number of data-stripe units, and M represents the number ofsimultaneous node or drive failures—or a combination of node and drive failures—thatthe cluster can withstand without incurring data loss. N must be larger than M.

In addition to N+M data protection, OneFS also supports data mirroring from 2x to 8x,allowing from two to eight mirrors of data. In terms of overall cluster performance andresource consumption, N+M protection is often more efficient than mirrored protection.However, because read and write performance is reduced for N+M protection, datamirroring might be faster for data that is updated often and is small in size. Datamirroring requires significant overhead and might not always be the best data-protectionmethod. For example, if you enable 3x mirroring, the specified content is duplicated threetimes on the cluster; depending on the amount of content mirrored, this can consume asignificant amount of storage space.



FlexProtect data recoveryOneFS uses the FlexProtect proprietary system to detect and repair files and directoriesthat are in a degraded state due to node or drive failures.

OneFS protects data in the cluster based on the configured protection policy. OneFSrebuilds failed disks, uses free storage space across the entire cluster to further preventdata loss, monitors data, and migrates data off of at-risk components.

OneFS distributes all data and error-correction information across the cluster and ensuresthat all data remains intact and accessible even in the event of simultaneous componentfailures. Under normal operating conditions, all data on the cluster is protected againstone or more failures of a node or drive. However, if a node or drive fails, the clusterprotection status is considered to be in a degraded state until the data is protected byOneFS again. OneFS reprotects data by rebuilding data in the free space of the cluster.While the protection status is in a degraded state, data is more vulnerable to data loss.

Because data is rebuilt in the free space of the cluster, the cluster does not require adedicated hot-spare node or drive in order to recover from a component failure. Becausea certain amount of free space is required to rebuild data, it is recommended that youreserve adequate free space through the virtual hot spare feature.

As you add more nodes, the cluster gains more CPU, memory, and disks to use duringrecovery operations. As a cluster grows larger, data restriping operations become faster.

SmartfailOneFS protects data stored on failing nodes or drives through a process calledsmartfailing.

During the smartfail process, OneFS places a device into quarantine. Data stored onquarantined devices is read only. While a device is quarantined, OneFS reprotects thedata on the device by distributing the data to other devices. After all data migration iscomplete, OneFS logically removes the device from the cluster, the cluster logicallychanges its width to the new configuration, and the node or drive can be physicallyreplaced.

OneFS smartfails devices only as a last resort. Although you can manually smartfailnodes or drives, it is recommended that you first consult Isilon Technical Support.

Occasionally a device might fail before OneFS detects a problem. If a drive fails withoutbeing smartfailed, OneFS automatically starts rebuilding the data to available free spaceon the cluster. However, because a node might recover from a failure, if a node fails,OneFS does not start rebuilding data unless the node is logically removed from thecluster.

Node failuresBecause node loss is often a temporary issue, OneFS does not automatically startreprotecting data when a node fails or goes offline. If a node reboots, the file system doesnot need to be rebuilt because it remains intact during the temporary failure.

If you configure N+1 data protection on a cluster, and one node fails, all of the data is stillaccessible from every other node in the cluster. If the node comes back online, the noderejoins the cluster automatically without requiring a full rebuild.

To ensure that data remains protected, if you physically remove a node from the cluster,you must also logically remove the node from the cluster. After you logically remove anode, the node automatically reformats its own drives, and resets itself to the factory


FlexProtect data recovery 239

default settings. The reset occurs only after OneFS has confirmed that all data has beenreprotected. You can logically remove a node using the smartfail process. It is importantthat you smartfail nodes only when you want to permanently remove a node from thecluster.

If you remove a failed node before adding a new node, data stored on the failed nodemust be rebuilt in the free space in the cluster. After the new node is added, OneFSdistributes the data to the new node. It is more efficient to add a replacement node to thecluster before failing the old node because OneFS can immediately use the replacementnode to rebuild the data stored on the failed node.

Requesting data protectionYou can request the protection of a file or directory by setting its requested protection.This flexibility enables you to protect distinct sets of data at different levels.

The default requested protection of node pools is N+2:1, which means that two drives orone node can fail without causing any data loss. For clusters or node pools containingless than two petabytes or fewer than 16 nodes, N+2:1 is the recommended requestedprotection. However, if the cluster or node pool is larger, you might consider higherrequested protection.

OneFS allows you to request protection that the cluster is currently incapable ofmatching. If you request an unmatchable protection, the cluster will continue trying tomatch the requested protection until a match is possible. For example, in a four-nodecluster, you might request a protection of 5x. In this example, OneFS would protect thedata at 4x until you added a fifth node to the cluster, at which point OneFS wouldreprotect the data at the 5x.

Note

For 4U Isilon IQ X-Series and NL-Series nodes, and IQ 12000X/EX 12000 combinationplatforms, the minimum cluster size of three nodes requires a minimum of N+2:1.

Requested protection settingsRequested protection settings determine the level of hardware failure that a cluster canrecover from without suffering data loss.

Requested protectionsetting

Minimum number ofnodes required

Definition

- - -N+1 3 The cluster can recover from one

drive or node failure withoutsustaining any data loss.

N+2:1 3 The cluster can recover from twosimultaneous drive failures or onenode failure without sustaining anydata loss.

N+2 5 The cluster can recover from twosimultaneous drive or node failureswithout sustaining any data loss.



Requested protectionsetting

Minimum number ofnodes required

Definition

- - -N+3:1 3 The cluster can recover from three

simultaneous drive failures or onenode failure without sustaining anydata loss.

N+3 7 The cluster can recover from threesimultaneous drive or node failureswithout sustaining any data loss.

N+4 9 The cluster can recover from foursimultaneous drive or node failureswithout sustaining any data loss.

Nx (Data mirroring) N

For example, 5x requiresa minimum of five nodes.

The cluster can recover from N - 1node failures without sustaining dataloss. For example, 5x protectionmeans that the cluster can recoverfrom four node failures

Requested protection disk space usageIncreasing the requested protection of data also increases the amount of spaceconsumed by the data on the cluster.

The parity overhead for N + M protection depends on the file size and the number ofnodes in the cluster. The percentage of parity overhead declines as the cluster getslarger.

The following table describes the estimated amount of overhead depending on therequested protection and the size of the cluster or node pool. The does not reflectrecommended protection levels based on cluster size.

Numberof nodes

+1 +2:1 +2 +3:1 +3 +4

- - - - - - -3 2 +1 (33%) 4 + 2 (33%) 3x — — —

4 3 +1 (25%) 6 + 2 (25%) 2 + 2 (50%) 9 + 3 (25%) 4x —

5 4 +1 (20%) 8 + 2 (20%) 3 + 2 (40%) 12 + 3(20%)

4x 5x

6 5 +1 (17%) 10 + 2 (17%) 4 + 2 (33%) 15 + 3(17%)

3 + 3 (50%) 5x

7 6 +1 (14%) 12 + 2 (14%) 5 + 2 (29%) 15 + 3(17%)

4 + 3 (43%) 5x

8 7 +1 (13%) 14 + 2 (12.5%) 6 + 2 (25%) 15 + 3(17%)

5 + 3 (38%) 4 + 4(50%)

9 8 +1 (11%) 16 + 2 (11%) 7 + 2 (22%) 15 + 3(17%)

6 + 3 (33%) 5 + 4(44%)


Requested protection disk space usage 241

Numberof nodes

+1 +2:1 +2 +3:1 +3 +4

- - - - - - -10 9 +1 (10%) 16 + 2 (11%) 8 + 2 (20%) 15 + 3

(17%)7 + 3 (30%) 6 + 4

(40%)

12 11 +1 (8%) 16 + 2 (11%) 10 + 2(17%)

15 + 3(17%)

9 + 3 (25%) 8 + 4(33%)

14 13 + 1 (7%) 16 + 2 (11%) 12 + 2(14%)

15 + 3(17%)

11 + 3(21%)

10 + 4(29%)

16 15 + 1 (6%) 16 + 2 (11%) 14 + 2(13%)

15 + 3(17%)

13 + 3(19%)

12 + 4(25%)

18 16 + 1 (6%) 16 + 2 (11%) 16 + 2(11%)

15 + 3(17%)

15 + 3(17%)

14 + 4(22%)

20 16 + 1 (6%) 16 + 2 (11%) 16 + 2(11%)

16 + 3(16%)

16 + 3(16%)

16 + 4(20%)

30 16 + 1 (6%) 16 + 2 (11%) 16 + 2(11%)

16 + 3(16%)

16 + 3(16%)

16 + 4(20%)

The parity overhead for mirrored data protection is not affected by the number of nodes inthe cluster. The following table describes the parity overhead for requested mirroredprotection.

2x 3x 4x 5x 6x 7x 8x- - - - - - -50% 67% 75% 80% 83% 86% 88%



CHAPTER 13

NDMP backup


u NDMP backup and recovery overview.................................................................. 244u NDMP two way backup........................................................................................ 244u Snapshot-based incremental backups................................................................ 245u NDMP protocol support....................................................................................... 246u Supported DMAs................................................................................................. 246u NDMP hardware support..................................................................................... 247u NDMP backup limitations.................................................................................... 247u NDMP performance recommendations................................................................ 247u Excluding files and directories from NDMP backups............................................ 249u Configuring basic NDMP backup settings............................................................ 250u Managing NDMP user accounts........................................................................... 251u Managing NDMP backup devices.........................................................................252u Managing NDMP backup ports............................................................................ 254u Managing NDMP backup sessions.......................................................................255u Managing restartable backups............................................................................ 257u Sharing tape drives between clusters.................................................................. 259u Managing default NDMP settings.........................................................................259u Managing snapshot based incremental backups.................................................263u View NDMP backup logs......................................................................................264

NDMP backup 243

NDMP backup and recovery overviewIn OneFS, you can back up and restore file-system data through the Network DataManagement Protocol (NDMP). From a backup server, you can direct backup and recoveryprocesses between an Isilon cluster and backup devices such as tape devices, mediaservers, and virtual tape libraries (VTLs).

OneFS supports both NDMP three-way backup and NDMP two-way backup. During anNDMP three-way backup operation, a data management application (DMA) on a backupserver instructs the cluster to start backing up data to a tape media server that is eitherattached to the LAN or directly attached to the DMA.

During a two-way NDMP backup, a DMA on a backup server instructs a BackupAccelerator node on the cluster to start backing up data to a tape media server that isattached to the Backup Accelerator node.

NDMP two-way backup is the most efficient method in terms of cluster resourceconsumption. However, NDMP two-way backup requires that you attach one or moreBackup Accelerator nodes to the cluster.

In both the NDMP two-way and three-way backup models, file history data is transferredfrom the cluster to the backup server. Before a backup begins, OneFS creates a snapshotof the targeted directory, then backs up the snapshot, which ensures that the backupimage represents a specific point in time.

You do not need to activate a SnapshotIQ license on the cluster to perform NDMPbackups. If you have activated a SnapshotIQ license on the cluster, you can generate asnapshot through the SnapshotIQ tool, and then back up the same snapshot to multipletape devices. If you back up a SnapshotIQ snapshot, OneFS does not create anothersnapshot for the backup.

Note

If you are backing up SmartLock directories for compliance purposes, it is recommendedthat you do not specify autocommit time periods for the SmartLock directories. This isbecause, depending on the autocommit period, files in the SmartLock directories maystill be subject to change.

NDMP two way backupTo perform NDMP two-way backups, you must attach a Backup Accelerator node to yourIsilon cluster and attach a tape device to the Backup Accelerator node. You must thenuse OneFS to detect the tape device before you can back up to that device.

You can connect supported tape devices directly to the Fibre Channel ports of a BackupAccelerator node. Alternatively, you can connect Fibre Channel switches to the FibreChannel ports on the Backup Accelerator node, and connect tape and media changerdevices to the Fibre Channel switches. For more information, see your Fibre Channelswitch documentation about zoning the switch to allow communication between theBackup Accelerator node and the connected tape and media changer devices.

If you attach tape devices to a Backup Accelerator node, the cluster detects the deviceswhen you start or restart the node or when you re-scan the Fibre Channel ports todiscover devices. If a cluster detects tape devices, the cluster creates an entry for thepath to each detected device.

If you connect a device through a Fibre Channel switch, multiple paths can exist for asingle device. For example, if you connect a tape device to a Fibre Channel switch, and

NDMP backup


then connect the Fibre Channel switch to two Fibre Channel ports, OneFS creates twoentries for the device, one for each path.

Note

If you perform an NDMP two-way backup operation, you must assign static IP addressesto the Backup Accelerator node. If you connect to the cluster through a data managementapplication (DMA), you must connect to the IP address of a Backup Accelerator node. Ifyou perform an NDMP three-way backup, you can connect to any node in the cluster.

Snapshot-based incremental backupsYou can implement snapshot-based incremental backups to increase the speed at whichthese backups are performed.

During a snapshot-based incremental backup, OneFS checks the snapshot taken for theprevious NDMP backup operation and compares it to a new snapshot. OneFS then backsup all data that was modified since the last snapshot was made.

If the incremental backup does not involve snapshots, OneFS must scan the directory todiscover which files were modified. OneFS can perform incremental backups significantlyfaster if snapshots are referenced.

You can perform incremental backups without activating a SnapshotIQ license on thecluster. Although SnapshotIQ offers a number of useful features, it does not enhancesnapshot capabilities in NDMP backup and recovery.

If you implement snapshot-based incremental backups, OneFS retains each snapshottaken for NDMP backups until a new backup of the same or lower level is performed.However, if you do not implement snapshot-based incremental backups, OneFSautomatically deletes each snapshot generated after the corresponding backup iscompleted or canceled.

The following table lists whether supported data management applications (DMAs) canperform snapshot-based incremental backups:

Table 15 DMA support for snapshot-based incremental backups

DMA Supported- -Symantec NetBackup Yes

EMC Networker Yes

EMC Avamar No

Commvault Simpana No

IBM Tivoli Storage Manager No

Symantec Backup Exec Yes

Dell NetVault Yes

ASG-Time Navigator Yes

NDMP backup

Snapshot-based incremental backups 245

NDMP protocol supportYou can back up cluster data through version 3 or 4 of the NDMP protocol.

OneFS supports the following features of NDMP versions 3 and 4:

u Full (level 0) NDMP backups

u Incremental (levels 1-10) NDMP backups

Note

In a level 10 NDMP backup, only data changed since the most recent incremental(level 1-9) backup or the last level 10 backup is copied. By repeating level 10backups, you can be assured that the latest versions of files in your data set arebacked up without having to run a full backup.

u Token-based NDMP backups

u NDMP TAR backup type

u Path-based and dir/node file history format

u Direct Access Restore (DAR)

u Directory DAR (DDAR)

u Including and excluding specific files and directories from backup

u Backup of file attributes

u Backup of Access Control Lists (ACLs)

u Backup of Alternate Data Streams (ADSs)

u Backup Restartable Extension (BRE)

OneFS supports connecting to clusters through IPv4 or IPv6.

Supported DMAsNDMP backups are coordinated by a data management application (DMA) that runs on abackup server.

OneFS supports the following DMAs:

u Symantec NetBackup

u EMC NetWorker

u EMC Avamar

u Symantec Backup Exec

u IBM Tivoli Storage Manager

u Dell NetVault

u CommVault Simpana (IPv6 protocol only)

u ASG-Time Navigator

NDMP backup


NDMP hardware supportOneFS can backup data to and restore data from tape devices and virtual tape libraries(VTL).

OneFS supports the following types of emulated and physical tape devices:

u LTO-3

u LTO-4

u LTO-5

OneFS supports the following virtual tape libraries (VTLs):

u FalconStor VTL 5.20

u Data Domain VTL 5.1.04 or later

NDMP backup limitationsOneFS NDMP backups have the following limitations:

u OneFS does not back up file system configuration data, such as file protection levelpolicies and quotas.

u OneFS does not support multiple concurrent backups onto the same tape.

u OneFS does not support restoring data from a file system other than OneFS. However,you can migrate data via the NDMP protocol from a NetApp or EMC VNX storagesystem to OneFS.

u Backup Accelerator nodes cannot interact with more than 1024 device paths,including the paths of tape and media changer devices. For example, if each devicehas four paths, you can connect 256 devices to a Backup Accelerator node. If eachdevice has two paths, you can connect 512 devices.

u OneFS does not support more than 64 concurrent NDMP sessions per BackupAccelerator node.

NDMP performance recommendationsConsider the following recommendations to optimize OneFS NDMP backups.

General performance recommendations

u Install the latest patches for OneFS and your data management application (DMA).

u If you are backing up multiple directories that contain small files, set up a separateschedule for each directory.

u If you are performing three-way NDMP backups, run multiple NDMP sessions onmultiple nodes in your Isilon cluster.

u Restore files through Direct Access Restore (DAR) and Directory DAR (DDAR). This isespecially recommended if you restore files frequently. However, it is recommendedthat you do not use DAR to restore a full backup or a large number of files, as DAR isbetter suited to restoring smaller numbers of files.

u Use the largest tape record size available for your version of OneFS. The largest taperecord size for OneFS versions 6.5.5 and later is 256 KB. The largest tape record sizefor versions of OneFS earlier than 6.5.5 is 128 KB.

NDMP backup

NDMP hardware support 247

u If possible, do not include or exclude files from backup. Including or excluding filescan affect backup performance, due to filtering overhead.

u Limit the depth of nested subdirectories in your file system.

u Limit the number of files in a directory. Distribute files across multiple directoriesinstead of including a large number of files in a single directory.

Networking recommendations

u Assign static IP addresses to Backup Accelerator nodes.

u Configure SmartConnect zones to specify pools of IP address ranges that areexclusive to NDMP backup operations.

u Connect NDMP sessions only through SmartConnect zones that are exclusively usedfor NDMP backup.

u Configure multiple policies when scheduling backup operations, with each policycapturing a portion of the file system. Do not attempt to back up the entire file systemthrough a single policy.

Backup Accelerator recommendations

u Run a maximum of four concurrent streams per Backup Accelerator node.

Note

This is recommended only if you are backing up a significant amount of data. Runningfour concurrent streams might not be necessary for smaller backups.

u Attach more Backup Accelerator nodes to larger clusters. The recommended numberof Backup Accelerator nodes is listed in the following table.

Table 16 Nodes per Backup Accelerator node

Node type Recommended number of nodes per Backup Accelerator node- -X-Series 3

NL-Series 3

S-Series 3

u Attach more Backup Accelerator nodes if you are backing up to more tape devices.The following table lists the recommended number of tape devices per backupaccelerator node:

Table 17 Tape devices per Backup Accelerator node

Tape device type Recommended number of tape devices per Backup Accelerator node- -LTO-5 3

LTO-4 4

LTO-3 8

DMA-specific recommendations

u Apply path-based file history instead of directory/inode (dir/node) file history.

NDMP backup


u Turn on multi-streaming, which enables OneFS to back up data to multiple tapedevices at the same time.

Excluding files and directories from NDMP backupsYou can exclude files and directories from NDMP backup operations by specifying NDMPenvironment variables through a data management application (DMA). If you include afile or directory, all other files and directories are automatically excluded from backupoperations. If you exclude a file or directory, all files and directories except the excludedone are backed up.

You can include or exclude files and directories by specifying the following characterpatterns:

Table 18 NDMP file and directory matching wildcards

Character Description Example Includes or excludes the followingdirectories

- - - -* Takes the place of

any character orcharacters

archive* /ifs/data/archive1/ifs/data/archive42_a/media

[] Takes the place ofa range of letters ornumbers

data_store_[a-f]data_store_[0-9]

/ifs/data/data_store_a/ifs/data/data_store_c/ifs/data/data_store_8

? Takes the place ofany singlecharacter

user_? /ifs/data/user_1/ifs/data/user_2

\ Includes a blankspace

user\ 1 /ifs/data/user 1

Unanchored patterns such as home or user1 target a string of text that might belong tomany files or directories. Anchored patterns target specific file pathnames, such as ifs/data/home. You can include or exclude either type of pattern.

For example, suppose you want to back up the /ifs/data/home directory, whichcontains the following files and directories:

u /ifs/data/home/user1/file.txtu /ifs/data/home/user2/user1/file.txtu /ifs/data/home/user3/other/file.txtu /ifs/data/home/user4/emptydirectoryIf you simply include the /ifs/data/home directory, all files and directories, includingemptydirectory would be backed up.

If you specify both include and exclude patterns, any excluded files or directories underthe included directories would not be backed up. If the excluded directories are not foundin any of the included directories, the exclude specification would have no effect.

Note

Specifying unanchored patterns can degrade the performance of backups. It isrecommended that you avoid unanchored patterns whenever possible.

NDMP backup

Excluding files and directories from NDMP backups 249

Configuring basic NDMP backup settingsYou can configure NDMP backup settings to control how these backups are performed forthe cluster. You can also configure OneFS to interact with a specific data managementapplication (DMA) for NDMP backups.

Configure and enable NDMP backupOneFS prevents NDMP backups by default. Before you can perform NDMP backups, youmust enable NDMP backups and configure NDMP settings.

Procedure

1. Click Data Protection > Backup > NDMP Settings.

2. In the Service area, click Enable.

3. Optional: To specify a port through which data management applications (DMAs)access the cluster, or the DMA vendor that OneFS is to interact with, in the Settingsarea, click Edit settings.

l In the Port number field, type a port number.

l From the DMA vendor list, select the name of the DMA vendor to manage backupoperations.

If your DMA vendor is not included in the list, select generic. However, note thatany vendors not included on the list are not officially supported and might notfunction as expected.

4. Click Add administrator to add an NDMP user account through which your DMA canaccess the cluster.

a. In the Add Administrator dialog box, in the Name field, type a name for theaccount.

b. In the Password and Confirm password fields, type a password for the account.

c. Click Submit.

Disable NDMP backupYou can disable NDMP backup if you no longer want to use this backup method.

Procedure


2. In the Service area, click Disable.

View NDMP backup settingsYou can view current NDMP backup settings. These settings define whether NDMPbackup is enabled, the port through which your data management application (DMA)connects to the cluster, and the DMA vendor that OneFS is configured to interact with.

Procedure

1. Click Data Protection > Backup > NDMP Settings and view NDMP backup settings.

2. In the Settings area, review NDMP backup settings.

NDMP backup


NDMP backup settingsYou can configure the following settings to control how NDMP backups are performed onthe cluster.

Port numberThe number of the port through which the data management application (DMA) canconnect to the cluster.

DMA vendorThe DMA vendor that the cluster is configured to interact with.

Managing NDMP user accountsYou can create, delete, and modify the passwords of NDMP user accounts.

Create an NDMP user accountBefore you can perform NDMP backups, you must create an NDMP user account throughwhich your data management application (DMA) can access the Isilon cluster.

Procedure


2. In the NDMP Administrators area, click Add administrator.

3. In the Add Administrator dialog box, in the Name field, type a name for the account.

4. In the Password and Confirm password fields, type a password for the account.

5. Click Submit.

Modify the password of an NDMP user accountYou can modify the password for an NDMP user account.

Procedure


2. In the NDMP Administrator table, in the row for an NDMP user account, click Changepassword.

3. In the Password and Confirm password fields, type a new password for the account.

4. Click Submit.

Delete an NDMP user accountYou can delete an NDMP user account.

Procedure


2. In the NDMP Administrators table, in the row for an NDMP user account, click Delete.


NDMP backup

NDMP backup settings 251

View NDMP user accountsYou can view information about NDMP user accounts.

Procedure


2. In the NDMP administrators area, review information about NDMP user accounts.

Managing NDMP backup devicesAfter you attach a tape or media changer device to a Backup Accelerator node, you mustconfigure OneFS to detect and establish a connection to the device. After the connectionbetween the cluster and the backup device is established, you can modify the name thatthe cluster has assigned to the device, or disconnect the device from the cluster.

Detect NDMP backup devicesIf you connect a tape device or media changer to a Backup Accelerator node, you mustconfigure OneFS to detect the device. Only then can OneFS back up data to and restoredata from the device. In OneFS, you can scan a specific node, a specific port, or all portson all nodes.

Procedure

1. Click Data Protection > Backup > Devices.

2. Click Discover devices.

3. Optional: To scan only a specific node for NDMP devices, from the Nodes list, select anode.

4. Optional: To scan only a specific port for NDMP devices, from the Ports list, select aport.

If you specify a port and a node, only the specified port on the node is scanned.However, if you specify only a port, the specified port will be scanned on all nodes.

5. Optional: To remove entries for devices or paths that have become inaccessible,select the Delete inaccessible paths or devices check box.

6. Click Submit.

Results

For each device that is detected, an entry is added to either the Tape Devices or MediaChangers tables.

Modify the name of an NDMP backup deviceYou can modify the name of an NDMP backup device in OneFS.

Procedure


2. In the Tape Devices table, click the name of a backup device entry.

3. In the Rename Device dialog box, in the Device Name field, type a new name for thebackup device.

4. Click Submit.

NDMP backup


Delete an entry for an NDMP backup deviceIf you physically remove an NDMP device from a cluster, OneFS retains the entry for thedevice. You can delete a device entry for a removed device. You can also remove thedevice entry for a device that is still physically attached to the cluster; this causes OneFSto disconnect from the device.

If you remove a device entry for a device that is connected to the cluster, and you do notphysically disconnect the device, OneFS will detect the device the next time it scans theports. You cannot remove a device entry for a device that is currently in use.

Procedure


2. In the Tape Devices table, in the row for the target device, click Delete device.


View NDMP backup devicesYou can view information about tape and media changer devices that are currentlyattached to your Isilon cluster.

Procedure


2. In the Tape Devices and Media Changers tables, review information about NDMPbackup devices.

NDMP backup device settingsOneFS creates a device entry for each device you attach to the cluster through a BackupAccelerator node.

The following table describes the settings available in the Tape Devices and MediaChangers tables:

Table 19 NDMP backup device settings

Setting Description- -Name A device name assigned by OneFS.

State Indicates whether the device is in use. If data is currently being backed up toor restored from the device, Read/Write appears. If the device is not in

use, Closed appears.

WWN The world wide node name (WWNN) of the device.

Product The name of the device vendor and the model name or number of the device.

Serial Number The serial number of the device.

Paths The name of the Backup Accelerator node that the device is attached to andthe port number or numbers to which the device is connected.

LUN The logical unit number (LUN) of the device.

Port ID The port ID of the device that binds the logical device to the physical device.

NDMP backup

Delete an entry for an NDMP backup device 253

Table 19 NDMP backup device settings (continued)

Setting Description- -WWPN The world wide port name (WWPN) of the port on the tape or media changer

device.

Managing NDMP backup portsYou can manage the Fibre Channel ports that connect tape and media changer devices toa Backup Accelerator node. You can also enable, disable, or modify the settings of anNDMP backup port.

Modify NDMP backup port settingsYou can modify the settings of an NDMP backup port.

Procedure

1. Click Data Protection > Backup > Ports.

2. In the Sessions table, click the name of a port.

3. In the Edit Port dialog box, modify port settings as needed, and then click Submit.

Enable or disable an NDMP backup portYou can enable or disable an NDMP backup port.

Procedure


2. In the Ports table, in the row of a port, click Enable or Disable.

View NDMP backup portsYou can view information about Fibre Channel ports of Backup Accelerator nodesattached to a cluster.

Procedure


2. In the Ports table, review information about NDMP backup ports.

NDMP backup port settingsOneFS assigns default settings to each port on each Backup Accelerator node attached tothe cluster. These settings identify each port and specify how the port interacts withNDMP backup devices.

The settings that appear in the Ports table are as follows:

Table 20 NDMP backup port settings

Setting Description- -Port The name of the Backup Accelerator node, and the number of the port.

NDMP backup


Table 20 NDMP backup port settings (continued)

Setting Description- -Topology The type of Fibre Channel topology that the port is configured to support..

Options are:

Point to PointA single backup device or Fibre Channel switch directly connected tothe port.

LoopMultiple backup devices connected to a single port in a circularformation.

AutoAutomatically detects the topology of the connected device. This is therecommended setting, and is required if you are using a switched-fabric topology.

WWNN The world wide node name (WWNN) of the port. This name is the same foreach port on a given node.

WWPN The world wide port name (WWPN) of the port. This name is unique to theport.

Rate The rate at which data is sent through the port. The rate can be set to 1Gb/s, 2 Gb/s, 4 Gb/s, 8 Gb/s, and Auto. 8 Gb/s is available for

A100 nodes only. If set to Auto, OneFS automatically negotiates with the

DMA to determine the rate. Auto is the recommended setting.

Managing NDMP backup sessionsYou can view the status of NDMP backup sessions or terminate a session that is inprogress.

End an NDMP sessionYou can end an NDMP backup or restore session at any time.

Procedure

1. Click Data Protection > Backup > Sessions.

2. In the Sessions table, in the row of the NDMP session that you want to end, click Kill.


View NDMP sessionsYou can view information about active NDMP sessions.

Procedure

1. Click Data Protection > Backup > Sessions.

2. In the Sessions table, review information about NDMP sessions.

NDMP backup

Managing NDMP backup sessions 255

NDMP session informationYou can view information about active NDMP sessions.

The following information is included in the Sessions table, as follows:

Table 21 NDMP session information

Item Description- -Session The unique identification number that OneFS assigned to the

session.

Elapsed How much time has elapsed since the session started.

Transferred The amount of data that was transferred during the session.

Throughput The average throughput of the session over the past five minutes.

Client/Remote The IP address of the backup server that the data managementapplication (DMA) is running on. If a three-way NDMP backup orrestore operation is currently running, the IP address of the remotetape media server also appears.

Mover/Data The current state of the data mover and the data server. The firstword describes the activity of the data mover. The second worddescribes the activity of the data server.The data mover and data server send data to and receive data fromeach other during backup and restore operations. The data mover isa component of the backup server that receives data duringbackups and sends data during restore operations. The data serveris a component of OneFS that sends data during backups andreceives information during restore operations.

The following states might appear:

ActiveThe data mover or data server is currently sending or receivingdata.

PausedThe data mover is temporarily unable to receive data. While thedata mover is paused, the data server cannot send data to thedata mover. The data server cannot be paused.

IdleThe data mover or data server is not sending or receiving data.

ListenThe data mover or data server is waiting to connect to the dataserver or data mover.

Operation The type of operation (backup or restore) that is currently inprogress. If no operation is in progress, this field is blank.

Backup (0-10)Indicates that data is currently being backed up to a mediaserver. The number indicates the level of NDMP backup.

NDMP backup


Table 21 NDMP session information (continued)

Item Description- -

RestoreIndicates that data is currently being restored from a mediaserver.

Source/Destination If an operation is currently in progress, specifies the /ifsdirectories that are affected by the operation. If a backup is inprogress, displays the path of the source directory that is beingbacked up. If a restore operation is in progress, displays the path ofthe directory that is being restored along with the destinationdirectory to which the tape media server is restoring data. If you arerestoring data to the same location that you backed up your datafrom, the same path appears twice.

Device The name of the tape or media changer device that iscommunicating with the cluster.

Mode How OneFS is interacting with data on the backup media server, asfollows:

Read/WriteOneFS is reading and writing data during a backup operation.

ReadOneFS is reading data during a restore operation.

RawThe DMA has access to tape drives, but the drives do notcontain writable tape media.

Managing restartable backupsA restartable backup is a type of NDMP backup that you can enable in your datamanagement application (DMA). If a restartable backup fails, for example, because of apower outage, you can restart the backup from a checkpoint close to the point of failure.In contrast, when a non-restartable backup fails, you must back up all data from thebeginning, regardless of what was transferred during the initial backup process.

After you enable restartable backups from your DMA, you can manage restartable backupcontexts from OneFS. These contexts are the data that OneFS stores to facilitaterestartable backups. Each context represents a checkpoint that the restartable backupprocess can return to if a backup fails.

Restartable backups are supported only for EMC NetWorker 8.1 and later.

Configure restartable backupsYou must configure EMC NetWorker to enable restartable backups and, optionally, definethe checkpoint interval.

If you do not specify a checkpoint interval, NetWorker uses the default interval of 5 GB.

NDMP backup

Managing restartable backups 257

Procedure

1. Configure the client and the directory path that you want to back up as you wouldnormally.

2. In the Client Properties dialog box, enable restartable backups.

a. On the General screen, click the Checkpoint enabled checkbox.

b. Specify File in the Checkpoint granularity drop-down list.

3. In the Application information field, type any NDMP variables that you want to specify.

The following specifies a checkpoint interval of 1 GB.CHECKPOINT_INTERVAL_IN_BYTES=1GB

4. Finish configuration and click OK in the Client Properties dialog box.

5. Start the backup.

6. If the backup is interrupted, for example, because of a power failure, restart it.

a. Browse to the Monitoring screen, and locate the backup process in the Groups list.

b. Right-click on the backup process, and in the context menu, click Restart.

NetWorker automatically restarts the backup from the last checkpoint.

Delete a restartable backup contextAfter a restartable backup context is no longer needed, your data managementapplication (DMA) automatically requests that OneFS delete the context. You canmanually delete a restartable backup context before the DMA requests it.

Note

It is recommended that you do not manually delete restartable backup contexts.Manually deleting a restartable backup context requires you to restart the correspondingNDMP backup from the beginning.

Procedure

1. Run the isi ndmp extensions contexts delete command.

The following command deletes a restartable backup context with an ID of792eeb8a-8784-11e2-aa70-0025904e91a4:isi ndmp extensions contexts delete 792eeb8a-8784-11e2-aa70-0025904e91a4

View restartable backup contextsYou can view restartable backup contexts that have been configured.

Procedure

1. View all backup contexts by running the following command:

isi ndmp extensions contexts list2. To view detailed information about a specific backup context, run the isi ndmp

extensions contexts view command.

The following command displays detailed information about a backup context with anID of 792eeb8a-8784-11e2-aa70-0025904e91a4:isi ndmp extensions contexts view 792eeb8a-8784-11e2-aa70-0025904e91a4

NDMP backup


Configure restartable backup settingsYou can specify the number of restartable backup contexts that OneFS retains at a time,up to a maximum of 1024 contexts.

Procedure

1. Run the isi ndmp extensions settings modify command.

The following command sets the maximum number of restartable backup contexts to128:isi ndmp extensions settings modify --bre_max_contexts 128

View restartable backup settingsYou can view the current limit of restartable backup contexts that OneFS retains at onetime.

Procedure

1. Run the following command:

isi ndmp extensions settings view

Sharing tape drives between clustersMultiple Isilon clusters, or an Isilon cluster and a third-party NAS system, can beconfigured to share a single tape drive. This helps to maximize the use of the tapeinfrastructure in your data center.

In your data management application (DMA), you must configure NDMP to control thetape drive and ensure that it is shared properly. The following configurations aresupported.

OneFS Versions* Supported DMAs Tested configurations- - -l 7.1.0.1

l 7.0.2.5

l 6.6.5.26

l EMC NetWorker 8.0 and later

l Symantec NetBackup 7.5and later

l Isilon Backup Accelerator with asecond Backup Accelerator

l Isilon Backup Accelerator with aNetApp storage system

* The tape drive sharing function is not supported in the OneFS 7.0.1 release.

EMC NetWorker refers to the tape drive sharing capability as DDS (dynamic drive sharing).Symantec NetBackup uses the term SSO (shared storage option). Consult your DMAvendor documentation for configuration instructions.

Managing default NDMP settingsIn OneFS, you can manage NDMP backup and restore operations by specifying defaultNDMP environment variables. You can also override default NDMP environment variables

NDMP backup

Configure restartable backup settings 259

through your data management application (DMA). For more information about specifyingNDMP environment variables through your DMA, see your DMA documentation.

Set default NDMP settings for a directoryYou can set default NDMP settings for a directory.

Procedure


2. Set default NDMP settings by running the isi ndmp settings variablescreate command.

For example, the following command sets the default file history format to path-basedformat for /ifs/data/media:

isi ndmp settings variables create /ifs/data/media HIST F

Modify default NDMP settings for a directoryYou can modify the default NDMP settings for a directory.

Procedure


2. Modify default NDMP settings by running the isi ndmp settings variablesmodify command.

For example, the following command sets the default file history format to path-basedformat for /ifs/data/media:

isi ndmp settings variables modify /ifs/data/media HIST F3. Optional: To remove a default NDMP setting for a directory, run the isi ndmp

settings variables delete command:

For example, the following command removes the default file history formatfor /ifs/data/media:

isi ndmp settings variables delete /ifs/data/media --name HIST

View default NDMP settings for directoriesYou can view the default NDMP settings for directories.

Procedure


2. View default NDMP settings by running the following command:

isi ndmp settings variables list

NDMP backup


NDMP environment variablesYou can specify default settings of NDMP backup and restore operations through NDMPenvironment variables. You can also specify NDMP environment variables through yourdata management application (DMA).

Table 22 NDMP environment variables

Environment variable Valid values Default Description- - - -BACKUP_MODE= TIMESTAMP

SNAPSHOTTIMESTAMP Enables or disables snapshot-

based incremental backups. Toenable snapshot-basedincremental backups, specifySNAPSHOT. To disable snapshot-

based incremental backups,specify TIMESTAMP.

FILESYSTEM= <file-path> None Specifies the full path of thedirectory you want to back up.Must be specified by the DMAbefore starting the backup, or anerror is generated.

LEVEL= <integer> 0 Specifies the level of NDMPbackup to perform. The followingvalues are valid:

0Performs a full NDMPbackup.

1 - 9Performs an incrementalbackup at the specified level.

10Performs unlimitedincremental backups.

UPDATE= YN

Y Determines whether OneFSupdates the dump dates file.

YOneFS updates the dumpdates file.

NOneFS does not update thedump dates file.

HIST= <file-history-format>

Y Specifies the file history format.The following values are valid:

DSpecifies dir/node filehistory.

NDMP backup

NDMP environment variables 261

Table 22 NDMP environment variables (continued)

Environment variable Valid values Default Description- - - -

FSpecifies path-based filehistory.

YSpecifies the default filehistory format determined byyour NDMP backup settings.

NDisables file history.

DIRECT= YN

N Enables or disables Direct AccessRestore (DAR) and Directory DAR(DDAR). The following values arevalid:

YEnables DAR and DDAR.

NDisables DAR and DDAR.

FILES= <file-matching-pattern>

None If you specify this option, OneFSbacks up only files and directoriesthat meet the specified pattern.Separate multiple patterns with aspace.

EXCLUDE= <file-matching-pattern>

None If you specify this option, OneFSdoes not back up files anddirectories that meet the specifiedpattern. Separate multiplepatterns with a space.

RESTORE_HARDLINK_BY_TABLE=

YN

N Determines whether OneFSrecovers hard links by building ahard-link table during restoreoperations. Specify this option ifhard links were incorrectly backedup, and restore operations arefailing.If a restore operation fails becausehard links were incorrectly backedup, the following messageappears in the NDMP backup logs:

Bad hardlink path for <path>

CHECKPOINT_INTERVAL_IN_BYTES=

<size> 5 GB Specifies the checkpoint intervalfor a restartable backup. If arestartable backup fails during thebackup process, you can restartthe backup from where the

NDMP backup


Table 22 NDMP environment variables (continued)

Environment variable Valid values Default Description- - - -

process failed. The <size>parameter is the space betweeneach checkpoint.Note that this variable can only beset from the DMA. For example, ifyou specify 2 GB, your DMA wouldcreate a checkpoint each time 2GB of data were backed up.

Restartable backups aresupported only for EMC NetWorker8.1 and later.

Managing snapshot based incremental backupsAfter you enable snapshot-based incremental backups, you can view and delete thesnapshots created for these backups.

Enable snapshot-based incremental backups for a directoryYou can configure OneFS to perform snapshot-based incremental backups for a directoryby default. You can also override the default setting in your data management application(DMA).

Procedure

1. Run the isi ndmp settings variable create command.

The following command enables snapshot-based incremental backups for /ifs/data/media:

isi ndmp settings variables create /ifs/data/media BACKUP_MODE SNAPSHOT

Delete snapshots for snapshot-based incremental backupsYou can delete snapshots created for snapshot-based incremental backups.

Note

It is recommended that you do not delete snapshots created for snapshot-basedincremental backups. If all snapshots are deleted for a path, the next backup performedfor the path is a full backup.

Procedure

1. Run the isi ndmp dumpdates delete command.

The following command deletes all snapshots created for backing up /ifs/data/media:isi ndmp dumpdates delete /ifs/data/media

NDMP backup

Managing snapshot based incremental backups 263

View snapshots for snapshot-based incremental backupsYou can view snapshots generated for snapshot-based incremental backups.

Procedure

1. Run the following command:

isi ndmp dumpdates list

View NDMP backup logsYou can view information about NDMP backup and restore operations through NDMPbackup logs.

Procedure

1. Click Data Protection > Backup > Logs.

2. In the Log Location area, from the Node list, select a node.

3. In the Log Contents area, review information about NDMP backup and restoreoperations.

NDMP backup


CHAPTER 14

File retention with SmartLock


u SmartLock overview............................................................................................ 266u Compliance mode............................................................................................... 266u SmartLock directories..........................................................................................266u Replication and backup with SmartLock.............................................................. 267u SmartLock license functionality...........................................................................268u SmartLock considerations................................................................................... 268u Set the compliance clock.................................................................................... 268u View the compliance clock.................................................................................. 269u Creating a SmartLock directory............................................................................ 269u Managing SmartLock directories......................................................................... 271u Managing files in SmartLock directories.............................................................. 274

File retention with SmartLock 265

SmartLock overviewYou can prevent users from modifying and deleting files on an EMC Isilon cluster with theSmartLock software module. You must activate a SmartLock license on a cluster toprotect data with SmartLock.

With the SmartLock software module, you can create SmartLock directories and commitfiles within those directories to a write once read many (WORM) state. You cannot eraseor re-write a file committed to a WORM state. After a file is removed from a WORM state,you can delete the file. However, you can never modify a file that has been committed toa WORM state, even after it is removed from a WORM state.

Compliance modeSmartLock compliance mode enables you to protect your data in compliance with theregulations defined by U.S. Securities and Exchange Commission rule 17a-4. You canupgrade a cluster to compliance mode during the initial cluster configuration process,before you activate the SmartLock license. To upgrade a cluster to SmartLock compliancemode after the initial cluster configuration process, contact Isilon Technical Support.

If you upgrade a cluster to compliance mode, you will not be able to log in to that clusterthrough the root user account. Instead, you can log in to the cluster through thecompliance administrator account that is configured either during initial clusterconfiguration or when the cluster is upgraded to compliance mode. If you are logged inthrough the compliance administrator account, you can perform administrative tasksthrough the sudo command.

SmartLock directoriesIn a SmartLock directory, you can commit a file to a WORM state manually or you canconfigure SmartLock to automatically commit the file. You can create two types ofSmartLock directories: enterprise and compliance. However, you can create compliancedirectories only if the cluster has been upgraded to SmartLock compliance mode. Beforeyou can create SmartLock directories, you must activate a SmartLock license on thecluster.

If you commit a file to a WORM state in an enterprise directory, the file can never bemodified and cannot be deleted until the retention period passes. However, if you arelogged in through the root user account, you can delete the file before the retentionperiod passes through the privileged delete feature. The privileged delete feature is notavailable for compliance directories. Enterprise directories reference the system clock tofacilitate time-dependent operations, including file retention.

Compliance directories enable you to protect your data in compliance with theregulations defined by U.S. Securities and Exchange Commission rule 17a-4. If youcommit a file to a WORM state in a compliance directory, the file cannot be modified ordeleted before the specified retention period has expired. You cannot delete committedfiles, even if you are logged in to the compliance administrator account. Compliancedirectories reference the compliance clock to facilitate time-dependent operations,including file retention.

You must set the compliance clock before you can create compliance directories. You canset the compliance clock only once. After you set the compliance clock, you cannotmodify the compliance clock time. The compliance clock is controlled by the complianceclock daemon. Because a user can disable the compliance clock daemon, it is possible



for a user to increase the retention period of WORM committed files in compliance mode.However, it is not possible to decrease the retention period of a WORM committed file.

Replication and backup with SmartLockYou must ensure that SmartLock directories remain protected during replication andbackup operations.

If you are replicating SmartLock directories with SyncIQ, it is recommended that youconfigure all nodes on the source and target clusters into Network Time Protocol (NTP)peer mode to ensure that the node clocks are synchronized. For compliance clusters, it isrecommended that you configure all nodes on the source and target clusters into NTPpeer mode before you set the compliance clock to ensure that the compliance clocks areinitially set to the same time.

Note

Do not configure SmartLock settings for a target SmartLock directory unless you are nolonger replicating data to the directory. Configuring an autocommit time period for atarget SmartLock directory can cause replication jobs to fail. If the target SmartLockdirectory commits a file to a WORM state, and the file is modified on the source cluster,the next replication job will fail because it cannot update the file.

SmartLock replication and backup limitationsBe aware of the limitations of replicating and backing up SmartLock directories withSyncIQ and NDMP.

If the source or target directory of a SyncIQ policy is a SmartLock directory, replicationmight not be allowed. For more information, see the following table:

Source directory type Target directory type Allowed- - -Non-SmartLock Non-SmartLock Yes

Non-SmartLock SmartLock enterprise Yes

Non-SmartLock SmartLock compliance No

SmartLock enterprise Non-SmartLock Yes; however, retentiondates and commit statusof files will be lost.

SmartLock enterprise SmartLock enterprise Yes

SmartLock enterprise SmartLock compliance No

SmartLock compliance Non-SmartLock No

SmartLock compliance SmartLock enterprise No

SmartLock compliance SmartLock compliance Yes

If you replicate SmartLock directories to another cluster with SyncIQ, the WORM state offiles is replicated. However, SmartLock directory configuration settings are not transferredto the target directory.

For example, if you replicate a directory that contains a committed file that is set to expireon March 4th, the file is still set to expire on March 4th on the target cluster. However, if


Replication and backup with SmartLock 267

the directory on the source cluster is set to prevent files from being committed for morethan a year, the target directory is not automatically set to the same restriction.

If you back up data to an NDMP device, all SmartLock metadata relating to the retentiondate and commit status is transferred to the NDMP device. If you restore data to aSmartLock directory on the cluster, the metadata persists on the cluster. However, if thedirectory that you restore to is not a SmartLock directory, the metadata is lost. You canrestore to a SmartLock directory only if the directory is empty.

SmartLock license functionalityYou must activate a SmartLock license on a cluster before you can create SmartLockdirectories and commit files to a WORM state.

If a SmartLock license becomes inactive, you will not be able to create new SmartLockdirectories on the cluster, modify SmartLock directory configuration settings, or deletefiles committed to a WORM state in enterprise directories before their expiration dates.However, you can still commit files within existing SmartLock directories to a WORMstate.

If a SmartLock license becomes inactive on a cluster that is running in SmartLockcompliance mode, root access to the cluster is not restored.

SmartLock considerationsu It is recommended that you create files outside of SmartLock directories and then

transfer them into a SmartLock directory after you are finished working with the files.If you are uploading files to a cluster, it is recommended that you upload the files to anon-SmartLock directory, and then later transfer the files to a SmartLock directory. If afile is committed to a WORM state while the file is being uploaded, the file willbecome trapped in an inconsistent state.

Files can be committed to a WORM state while they are still open. If you specify anautocommit time period for a directory, the autocommit time period is calculatedaccording to the length of time since the file was last modified, not when the file wasclosed. If you delay writing to an open file for more than the autocommit time period,the file will be committed to a WORM state the next time you attempt to write to it.

u In a Microsoft Windows environment, if you commit a file to a WORM state, you canno longer modify the hidden or archive attributes of the file. Any attempt to modifythe hidden or archive attributes of a WORM committed file will generate an error. Thiscan prevent third-party applications from modifying the hidden or archive attributes.

Set the compliance clockBefore you can create SmartLock compliance directories, you must set the complianceclock. This procedure is available only through the command-line interface (CLI).

Setting the compliance clock configures the clock to the same time as the cluster systemclock. Before you set the compliance clock, ensure that the cluster system clock is set tothe correct time. After the compliance clock is set, if the compliance clock becomesunsynchronized with the system clock, the compliance clock slowly corrects itself tomatch the system clock. The compliance clock corrects itself at a rate of approximatelyone week per year.

Procedure

1. Open a secure shell (SSH) connection to any node in the cluster and log in through thecompliance administrator account.



2. Set the compliance clock by running the following command.

isi worm cdate set

View the compliance clockYou can view the current time of the compliance clock. This procedure is available onlythrough the command-line interface (CLI).

Procedure

1. Open a secure shell (SSH) connection to any node in the cluster and log in through thecompliance administrator account.

2. View the compliance clock by running the following command:

isi worm cdate

The system displays output similar to the following:Current Compliance Clock Date/Time: 2012-07-07 01:00:00

Creating a SmartLock directoryYou can create a SmartLock directory and configure settings that control how long filesare retained in a WORM state and when files are automatically committed to a WORMstate.

It is recommended that you set SmartLock configuration settings only once and do notmodify the settings after files have been added to the SmartLock directory. If anautocommit time period is specified for the directory, modifying SmartLock configurationsettings can affect the retention period of files, even if the autocommit time period of thefiles has already expired. You cannot move or rename a directory that contains aSmartLock directory.

Retention periodsA retention period is the length of time that a file remains in a WORM state before beingreleased from a WORM state. You can configure SmartLock directory settings that enforcedefault, maximum, and minimum retention periods for the directory.

If you manually commit a file, you can optionally specify the date that the file is releasedfrom a WORM state. You can configure a minimum and a maximum retention period for aSmartLock directory to prevent files from being retained for too long or too short a timeperiod. It is recommended that you specify a minimum retention period for all SmartLockdirectories.

For example, assume that you have a SmartLock directory with a minimum retentionperiod of two days. At 1:00 PM on Monday, you commit a file to a WORM state, andspecify the file to be released from a WORM state on Tuesday at 3:00 PM. The file will bereleased from a WORM state two days later on Wednesday at 1:00 PM, because releasingthe file earlier would violate the minimum retention period.

You can also configure a default retention period that is assigned when you commit a filewithout specifying a date to release the file from a WORM state.

Autocommit time periodsYou can configure an autocommit time period for SmartLock directories. After a file hasbeen in a SmartLock directory without being modified for the specified autocommit time


View the compliance clock 269

period, the file is automatically committed to a WORM state the next time that file isaccessed by a user.

After the autocommit time period for a file passes, the file continues to reference thecurrent autocommit time period until the file is accessed by a user. Therefore, increasingthe autocommit time period of a directory might cause files to be committed to a WORMstate later than expected. For example, assume that you have a SmartLock directory withan autocommit time period of one day, and an expiration period of one day. You thencopy a file into the SmartLock directory on Monday, at 3:00 PM. At 5:00 PM on Tuesday,you increase the autocommit time period to two days. If the file was not accessed, userscan modify or delete the file until 3:00 PM on Wednesday.

Decreasing the autocommit time period of a directory can cause a file to be released froma WORM state earlier than expected. For example, assume that you have a SmartLockdirectory with an autocommit time period of one day, and a default expiration period ofone day. You then copy a file into the SmartLock directory on Monday, at 3:00 PM. If, at4:00 PM on Tuesday, the file was not accessed by a user, and you decrease theautocommit time period to two hours, the file is set to be removed from a WORM state at5:00 PM on Tuesday, instead of 3:00 PM on Wednesday.

Modifying the minimum, maximum, or default retention period of a SmartLock directorycan modify the retention period of files, even after the autocommit time period of a fileexpires. For example, assume that you have a SmartLock directory with an autocommittime period of two days, and a default expiration period of one day. You then copy a fileinto the SmartLock directory on Monday, at 3:00 PM. If, by 4:00 PM on Wednesday, thefile was not accessed by a user, and you decrease the default retention period to twohours, the file is removed from a WORM state at 5:00 PM on Wednesday, instead of 3:00PM on Thursday.

If you specify an autocommit time period along with a minimum, maximum, or defaultretention period, the retention period is calculated according to the time that theautocommit period expires. For example, assume that you have a SmartLock directorywith a minimum retention period of two days and an autocommit time period of one day.At 1:00 PM on Monday, you modify a file; then, at 5:00 PM on Tuesday, you access thefile, and the file is committed to a WORM state. The retention period expires on Thursdayat 1:00 PM, two days after the autocommit time period for the file expired.

If the autocommit time period expires for a file, and the file is accessed by a user, the fileis committed to a WORM state. However, the read-write permissions of the file are notmodified. The file is still committed to a WORM state; the file can never be modified, andit cannot be deleted until the specified retention period expires. However, the WORMstate is not indicated by the read-write permissions.

Create a SmartLock directoryYou can create a SmartLock directory and commit files in that directory to a WORM state.This procedure is available only through the command-line interface (CLI).

Before creating a SmartLock directory, be aware of the following conditions andrequirements:

u You cannot create a SmartLock directory as a subdirectory of an existing SmartLockdirectory.

u Hard links cannot cross SmartLock directory boundaries.

u Creating a SmartLock directory causes a corresponding SmartLock domain to becreated for that directory.

Procedure




2. Run the isi worm domains create command to create a SmartLock directory.

The path specified in the isi worm domains create command cannot be thepath of an existing directory.

The following command creates a compliance directory with a default retention periodof four years, a minimum retention period of three years, and an maximum retentionperiod of five years:sudo isi worm mkdir --path /ifs/data/dir --compliance \--default 4y --min 3y --max 5y

The following command creates an enterprise directory with an autocommit timeperiod of thirty minutes and a minimum retention period of three months:isi worm mkdir --path /ifs/data/dir --autocommit 30n --min 3m

Managing SmartLock directoriesYou can modify the default, minimum, and maximum retention period and theautocommit period for a SmartLock directory at any time.

A SmartLock directory can be renamed only if the directory is empty.

Modify a SmartLock directoryYou can modify the SmartLock configuration settings for a SmartLock directory. Thisprocedure is available only through the command-line interface (CLI).

Note

It is recommended that you set SmartLock configuration settings only once and do notmodify the settings after files are added to the SmartLock directory.

Procedure


2. Modify SmartLock configuration settings by running the isi worm modifycommand.

For example, the following command sets the default retention period to one year:

isi worm modify --path /ifs/data/protected_directory \--default 1y

View SmartLock directory settingsYou can view the SmartLock configuration settings for SmartLock directories. Thisprocedure is available only through the command-line interface (CLI).

Procedure


2. View all SmartLock domains by running the following command:

isi worm domains list

The system displays output similar to the following example:

ID Path Type-----------------------------------------------65536 /ifs/data/SmartLock/directory1 enterprise65537 /ifs/data/SmartLock/directory2 enterprise65538 /ifs/data/SmartLock/directory3 enterprise-----------------------------------------------


Managing SmartLock directories 271

3. Optional: To view detailed information about a specific SmartLock directory, run theisi worm domains view command.The following command displays detailed information about /ifs/data/SmartLock/directory2:isi worm domains view /ifs/data/SmartLock/directory2


ID: 65537 Path: /ifs/data/SmartLock/directory2 Type: enterprise LIN: 4295426060Autocommit Offset: 30m Override Date: -Privileged Delete: offDefault Retention: 1Y Min Retention: 3M Max Retention: - Total Modifies: 3/32 Max

SmartLock directory configuration settingsYou can configure SmartLock directory settings that determine when files are committedto and how long files are retained in a WORM state for a given directory.

All SmartLock directories are assigned the following settings:

IDThe numerical ID of the corresponding SmartLock domain.

Root pathThe path of the directory.

TypeThe type of directory. Enterprise directories display SmartLock. Compliancedirectories display Compliance.

Override dateThe override retention date for the directory. Files committed to a WORM state arenot released from a WORM state until after the specified date, regardless of themaximum retention period for the directory or whether a user specifies a retentionperiod expiration date.

Default retention periodThe default retention period for the directory. If a retention period expiration date isnot explicitly assigned by a user, the default retention period is assigned to the filewhen it is committed to a WORM state.

Times are expressed in the format "<integer> <time>", where <time> is one of thefollowing values:

Value Description- -y Specifies years

m Specifies months

w Specifies weeks

d Specifies days



Minimum retention periodThe minimum retention period for the directory. Files are retained in a WORM statefor at least the specified amount of time, even if a user specifies an expiration datethat results in a shorter retention period.Times are expressed in the format "<integer> <time>", where <time> is one of thefollowing values:


m Specifies months

w Specifies weeks

d Specifies days

Maximum retention periodThe maximum retention period for the directory. Files are retained in a WORM statefor longer than the specified amount of time, even if a user specifies an expirationdate that results in a longer retention period.



m Specifies months

w Specifies weeks

d Specifies days

Autocommit periodThe autocommit time period for the directory. After a file exists in this SmartLockdirectory without being modified for the specified time period, the file isautomatically committed the next time the file is accessed by a user.



m Specifies months

w Specifies weeks

d Specifies days

h Specifies hours

n Specifies minutes


SmartLock directory configuration settings 273

Privileged deleteIndicates whether files in the directory can be deleted through the privileged deletefunctionality.

OnA root user can delete files committed to a WORM state by running the isiworm filedelete command.

OffWORM committed files cannot be deleted, even through the isi wormfiledelete command.

Disabled (Permanently)WORM committed files cannot be deleted, even through the isi wormfiledelete command. After this setting is set, the setting cannot bemodified.

Managing files in SmartLock directoriesYou can commit files in SmartLock directories to a WORM state by removing the read-write privileges of the file. You can also set a specific date at which the retention periodof the file expires. Once a file is committed to a WORM state, you can increase theretention period of the file, but you cannot decrease the retention period of the file. Youcannot move a file that has been committed to a WORM state, even after the retentionperiod for the file has expired.

The retention period expiration date is set by modifying the access time of a file. In aUNIX command line, the access time can be modified through the touch command.Although there is no method of modifying the access time through Windows Explorer, youcan modify the access time through Windows Powershell. Accessing a file does not setthe retention period expiration date.

If you run the touch command on a file in a SmartLock directory without specifying adate on which to release the file from a SmartLock state, and you commit the file, theretention period is automatically set to the minimum retention period specified for theSmartLock directory. If you have not specified a minimum retention period for theSmartLock directory, the file is assigned a retention period of zero seconds. It isrecommended that you specify a minimum retention period for all SmartLock directories.

Set a retention period through a UNIX command lineYou can specify when a file will be released from a WORM state through a UNIX commandline.

Procedure

1. Open a connection to any node in the cluster through a UNIX command line and log in.

2. Set the retention period by modifying the access time of the file through the touchcommand.

The following command sets an expiration date of June 1, 2015 for /ifs/data/test.txt:touch -at 201506010000 /ifs/data/test.txt



Set a retention period through Windows PowershellYou can specify when a file will be released from a WORM state through MicrosoftWindows Powershell.

Procedure

1. Open the Windows PowerShell command prompt.

2. Optional: Establish a connection to the cluster by running the net use command.

The following command establishes a connection to the /ifs directory oncluster.ip.address.com:net use "\\cluster.ip.address.com\ifs" /user:root password

3. Specify the name of the file you want to set a retention period for by creating anobject.

The file must exist in a SmartLock directory.

The following command creates an object for /smartlock/file.txt:$file = Get-Item "\\cluster.ip.address.com\ifs\smartlock\file.txt"

4. Specify the retention period by setting the last access time for the file.

The following command sets an expiration date of July 1, 2015 at 1:00 PM:$file.LastAccessTime = Get-Date "2015/7/1 1:00 pm"

Commit a file to a WORM state through a UNIX command lineYou can commit a file to a WORM state through a UNIX command line.

To commit a file to a WORM state, you must remove all write privileges from the file. If afile is already set to a read-only state, you must first add write privileges to the file, andthen return the file to a read-only state.

Procedure

1. Open a connection to the cluster through a UNIX command line and log in.

2. Remove write privileges from a file by running the chmod command.

The following command removes write privileges of /ifs/data/smartlock/file.txt:chmod ugo-w /ifs/data/smartlock/file.txt

Commit a file to a WORM state through Windows ExplorerYou can commit a file to a WORM state through Microsoft Windows Explorer. Thisprocedure describes how to commit a file through Windows 7.

To commit a file to a WORM state, you must apply the read-only setting. If a file is alreadyset to a read-only state, you must first remove the file from a read-only state and thenreturn it to a read-only state.

Procedure

1. In Windows Explorer, navigate to the file you want to commit to a WORM state.

2. Right-click the folder and then click Properties.

3. In the Properties window, click the General tab.

4. Select the Read-only check box, and then click OK.


Set a retention period through Windows Powershell 275

Override the retention period for all files in a SmartLock directoryYou can override the retention period for files in a SmartLock directory. All filescommitted to a WORM state within the directory will remain in a WORM state until afterthe specified day. This procedure is available only through the command-line interface(CLI).

If files are committed to a WORM state after the retention period is overridden, theoverride date functions as a minimum retention date. All files committed to a WORM statedo not expire until at least the given day, regardless of user specifications.

Procedure


2. Override the retention period expiration date for all WORM committed files in aSmartLock directory by running the isi worm modify command.

For example, the following command overrides the retention period expiration dateof /ifs/data/smartlock to June 1, 2014:isi worm modify --path /ifs/data/smartlock --override 20140601

Delete a file committed to a WORM stateYou can delete a WORM committed file before the expiration date only if you are logged inas the root user. This procedure is available only through the command-line interface(CLI).

Before you begin

Privileged delete functionality must not be permanently disabled for the SmartLockdirectory that contains the file.Procedure

1. Open a connection to the cluster through a UNIX command line and log in through theroot user account.

2. If privileged delete functionality was disabled for the SmartLock directory, modify thedirectory by running the isi worm domains modify command with the --privileged-delete option.

For example, the following command enables privileged delete for /ifs/data/enterprise:

isi worm modify --path /ifs/data/enterprise --privdel on3. Delete the WORM committed file by running the isi worm filedelete command.

For example, the following command deletes /ifs/worm/enterprise/file:

isi worm filedelete /ifs/worm/enterprise/file

The system displays output similar to the following:!! Are you sure? Please enter 'yes' to confirm: (yes, [no])

4. Type yes and then press ENTER.



View WORM status of a fileYou can view the WORM status of an individual file. This procedure is available onlythrough the command-line interface (CLI).

Procedure


2. View the WORM status of a file by running the isi worm files view command.

For example, the following command displays the WORM status of /ifs/worm/enterprise/file:isi worm info --path /ifs/worm/enterprise/file --verbose


View WORM status of a file 277

CHAPTER 15

Protection domains


u Protection domains overview.............................................................................. 280u Protection domain considerations.......................................................................280u Create a protection domain................................................................................. 281u Delete a protection domain................................................................................. 281u View protection domains.....................................................................................281u Protection domain types..................................................................................... 281

Protection domains 279

Protection domains overviewProtection domains are markers that prevent modifications to files and directories. If adomain is applied to a directory, the domain is also applied to all of the files andsubdirectories under the directory. You can specify domains manually; however, OneFSusually creats domains automatically.

There are three types of domains: SyncIQ, SmartLock, and SnapRevert. SyncIQ domainscan be assigned to source and target directories of replication policies. OneFSautomatically creates a SyncIQ domain for the target directory of a replication policy thefirst time that the policy is run. OneFS also automatically creates a SyncIQ domain for thesource directory of a replication policy during the failback process. You can manuallycreate a SyncIQ domain for a source directory before you initiate the failback process, butyou cannot delete a SyncIQ domain that mark the target directory of a replication policy.

SmartLock domains are assigned to SmartLock directories to prevent committed filesfrom being modified or deleted. OneFS automatically creates a SmartLock domain when aSmartLock directory is created. You cannot delete a SmartLock domain. However, if youdelete a SmartLock directory, OneFS automatically deletes the SmartLock domainassociated with the directory.

SnapRevert domains are assigned to directories that are contained in snapshots toprevent files and directories from being modified while a snapshot is being reverted.OneFS does not automatically create SnapRevert domains. You cannot revert a snapshotuntil you create a SnapRevert domain for the directory that the snapshot contains. Youcan create SnapRevert domains for subdirectories of directories that already haveSnapRevert domains. For example, you could create SnapRevert domains for both /ifs/data and /ifs/data/archive. You can delete a SnapRevert domain if you no longerwant to revert snapshots of a directory.

Protection domain considerationsYou can manually create protection domains before they are required by OneFS toperform certain actions. However, manually creating protection domains can limit yourability to interact with the data marked by the domain.

u Copying a large number of files into a protection domain might take a very long timebecause each file must be marked individually as belonging to the protectiondomain.

u You cannot move directories in or out of protection domains. However, you can movea directory contained in a protection domain to another location within the sameprotection domain.

u Creating a protection domain for a directory that contains a large number of files willtake more time than creating a protection domain for a directory with fewer files.Because of this, it is recommended that you create protection domains for directorieswhile the directories are empty, and then add files to the directory.

u If a domain is currently preventing the modification or deletion of a file, you cannotcreate a protection domain for a directory that contains that file. For example,if /ifs/data/smartlock/file.txt is set to a WORM state by a SmartLockdomain, you cannot create a SnapRevert domain for /ifs/data/.

Protection domains


Create a protection domainYou can create replication or snapshot revert domains to facilitate snapshot revert andfailover operations. You cannot create a SmartLock domain. OneFS automatically createsa SmartLock domain when you create a SmartLock directory.

Procedure



3. In the Domain Root Path field, type the path of the directory you want to create aprotection domain for.

4. From the Type of domain list, specify the type of domain you want to create.

5. Ensure that the Delete this domain check box is cleared.

6. Click Start Job.

Delete a protection domainYou can delete a replication or snapshot revert domain if you want to move directoriesout of the domain. You cannot delete a SmartLock domain. OneFS automatically deletes aSmartLock domain when you delete a SmartLock directory.

Procedure



3. In the Domain Root Path field, type the path of the directory you want to delete aprotection domain for.

4. From the Type of domain list, specify the type of domain you want to delete.

5. Select Delete this domain.

6. Click Start Job.

View protection domainsYou can view protection domains on a cluster. This procedure is available only throughthe command-line interface (CLI).

Procedure


2. View protection domains by running the isi domain list command.

Protection domain typesThere are three general protection domain types: SmartLock, SnapRevert, and SyncIQ.Each protection domain type can be divided into additional subcategories.

The following domain types appear in the output of the isi domain list command.

Protection domains

Create a protection domain 281

SmartLockSmartLock domain of an enterprise directory.

ComplianceSmartLock domain of a compliance directory.

SyncIQSyncIQ domain that prevents users from modifying files and directories.

SyncIQ, WritableSyncIQ domain that allows users to modify files and directories.

SnapRevertSnapRevert domain that prevents users from modifying files and directories while asnapshot is being reverted.

Writable, SnapRevertSnapRevert domain that allows users to modify files and directories.

If Incomplete is appended to a domain type, OneFS is in the process of creating thedomain. An incomplete domain does not prevent files from being modified or deleted.

Protection domains


CHAPTER 16

Data-at-rest-encryption


u Data-at-rest encryption overview......................................................................... 284u Self-encrypting drives..........................................................................................284u Data security on self-encrypted drives................................................................. 284u Data migration to a self-encrypted-drives cluster................................................. 285u Chassis and drive states..................................................................................... 285u Smartfailed drive ERASE and REPLACE state examples........................................ 287

Data-at-rest-encryption 283

Data-at-rest encryption overviewYou can enhance data security with a EMC Isilon cluster that contains only self-encrypting-drive nodes, providing data-at-rest protection.

The OneFS system is available as a cluster that is composed of Isilon OneFS nodes thatcontain only self-encrypting drives (SEDs). The system requirements and management ofdata at rest on self-encrypting nodes are identical to that of nodes that do not containself-encrypting drives. Clusters of mixed node types are not supported.

Self-encrypting drivesSelf-encrypting drives store data on a EMC Isilon cluster that is specially designed fordata-at-rest encryption.

Data-at-rest- encryption on self-encrypted drives occurs when data that is stored on adevice is encrypted to prevent unauthorized data access. All data written to the storagedevice is encrypted when it is stored, and all data read from the storage device isdecrypted when it is read. The stored data is encrypted with a 256-bit data AESencryption key and decrypted in the same manner. OneFS controls data access bycombining the drive authentication key with on-disk data-encryption keys.

Note

All nodes in a cluster must be of the self-encrypting drive type. Mixed nodes are notsupported.

Data security on self-encrypted drivesSmartfailing self-encrypted drives guarantees data security after removal.

Data on self-encrypted drives is protected from unauthorized access by authenticatingencryption keys. Encryption keys never leave the drive. When a drive is locked, successfulauthentication unlocks the drive for data access.

The data on self-encrypted drives is rendered inaccessible in the following conditions:

u When a self-encrypting drive is smartfailed, drive authentication keys are deletedfrom the node. The data on the drive cannot be decrypted and is thereforeunreadable, which secures the drive.

u When a drive is smartfailed and removed from a node, the encryption key on the driveis removed. Because the encryption key for reading data from the drive must be thesame key that was used when the data was written, it is impossible to decrypt datathat was previously written to the drive. When you smartfail and then remove a drive,it is cryptographically erased.

Note

Smartfailing a drive is the preferred method for removing a self-encrypted drive.Removing a node that has been smartfailed guarantees that data is inaccessible.

u When a self-encrypting drive loses power, the drive locks to prevent unauthorizedaccess. When power is restored, data is again accessible when the appropriate driveauthentication key is provided.



Data migration to a self-encrypted-drives clusterYou can migrate data from your existing cluster to a cluster of self-encrypted-drive nodes.

The Isilon cluster does not support the coexistence of regular and self-encrypted nodes.However, if you have data on an existing Isilon cluster that you want to migrate to acluster of self-encrypted nodes, you can add self-encrypted nodes to your existing clusterone time only to migrate your data.

Note

Before you begin the data-migration process, both clusters must be upgraded to thesame OneFS version.

During data migration, an error is generated that indicates you are running in mixedmode, which is not supported and is not secure. The data migrated to the self-encrypteddrives is not secure until the smartfail process is completed for the non-encrypted drives.

CAUTION

Data migration to a cluster of self-encrypted-drive nodes must be performed by IsilonProfessional Services. For more information, contact your EMC Isilon representative.

Chassis and drive statesYou can view chassis and drive state details.

In a cluster, the combination of nodes in different degraded states determines whetherread requests, write requests, or both work. A cluster can lose write quorum but keepread quorum. OneFS provides details about the status of chassis and drives in yourcluster. The following table describes all the possible states that you may encounter inyour cluster.


- - - -HEALTHY All drives in the node are functioning correctly. CLI, web

administrationinterface

SMARTFAIL or

Smartfail orrestripe inprogress

The drive is in the process of being removedsafely from the file system, either because ofan I/O error or by user request. Nodes or drivesin a smartfail or read-only state affect onlywrite quorum.


NOT AVAILABLE A drive can be unavailable for a variety ofreasons. You can click the bay to view detailedinformation about this condition.

Note


includes the ERASE and SED_ERRORcommand-line interface states.


X


Data migration to a self-encrypted-drives cluster 285


- - - -SUSPENDED This state indicates that drive activity is

temporarily suspended and the drive is not inuse. The state is manually initiated and doesnot occur during normal cluster activity.


NOT IN USE A node in an offline state affects both read andwrite quorum.


REPLACE The drive was smartfailed successfully and isready to be replaced.

CLI only

STALLED The drive is stalled and undergoing stallevaluation. Stall evaluation is the process ofchecking drives that are slow or having otherissues. Depending on the outcome of theevaluation, the drive may return to service orbe smartfailed. This is a transient state.

CLI only

NEW The drive is new and blank. This is the statethat a drive is in when you run the isi dev -a add command.

CLI only

USED The drive was added and contained an IsilonGUID but the drive is not from this node. Thisdrive likely will be formatted into the cluster.

CLI only

PREPARING The drive is undergoing a format operation.The drive state changes to HEALTHY when theformat is successful.

CLI only

EMPTY No drive is in this bay. CLI only

WRONG_TYPE The drive type is wrong for this node. Forexample, a non-SED drive in a SED node, SASinstead of the expected SATA drive type.

CLI only

BOOT_DRIVE Unique to the A100 drive, which has bootdrives in its bays.

CLI only

SED_ERROR The drive cannot be acknowledged by theOneFS system.

Note



CLI , webadministrationinterface

X

ERASE The drive is ready for removal but needs yourattention because the data has not beenerased. You can erase the drive manually toguarantee that data is removed.

Note



CLI only




- - - -INSECURE Data on the self-encrypted drive is accessible

by unauthorized personnel. Self-encryptingdrives should never be used for non-encrypteddata purposes.

Note


is labeled Unencrypted SED.

CLI only X

UNENCRYPTEDSED

Data on the self-encrypted drive is accessibleby unauthorized personnel. Self-encryptingdrives should never be used for non-encrypteddata purposes.

Note

In the command-line interface, this state is

labeled INSECURE.

Webadministrationinterface only

X

Smartfailed drive ERASE and REPLACE state examplesYou can see different drive states during the smartfail process.

The following command-line output examples demonstrate the difference between theREPLACE and ERASE states during the smartfail process.

my-cluster# isi devNode 1, [ATTN] Bay 1 Lnum 11 [SMARTFAIL] SN:Z296M8HK 000093172YE04 /dev/da1 Bay 2 Lnum 10 [HEALTHY] SN:Z296M8N5 00009330EYE03 /dev/da2 Bay 3 Lnum 9 [HEALTHY] SN:Z296LBP4 00009330EYE03 /dev/da3 Bay 4 Lnum 8 [HEALTHY] SN:Z296LCJW 00009327BYE03 /dev/da4 Bay 5 Lnum 7 [HEALTHY] SN:Z296M8XB 00009330KYE03 /dev/da5 Bay 6 Lnum 6 [HEALTHY] SN:Z295LXT7 000093172YE03 /dev/da6 Bay 7 Lnum 5 [HEALTHY] SN:Z296M8ZF 00009330KYE03 /dev/da7 Bay 8 Lnum 4 [HEALTHY] SN:Z296M8SD 00009330EYE03 /dev/da8 Bay 9 Lnum 3 [HEALTHY] SN:Z296M8QA 00009330EYE03 /dev/da9 Bay 10 Lnum 2 [HEALTHY] SN:Z296M8Q7 00009330EYE03 /dev/da10 Bay 11 Lnum 1 [HEALTHY] SN:Z296M8SP 00009330EYE04 /dev/da11 Bay 12 Lnum 0 [HEALTHY] SN:Z296M8QZ 00009330JYE03 /dev/da12

Smartfail completes successfully and the drive state changes to REPLACE:

my-cluster# isi devNode 1, [ATTN] Bay 1 Lnum 11 [REPLACE] SN:Z296M8HK 000093172YE04 /dev/da1


Smartfailed drive ERASE and REPLACE state examples 287

Bay 2 Lnum 10 [HEALTHY] SN:Z296M8N5 00009330EYE03 /dev/da2 Bay 3 Lnum 9 [HEALTHY] SN:Z296LBP4 00009330EYE03 /dev/da3 Bay 4 Lnum 8 [HEALTHY] SN:Z296LCJW 00009327BYE03 /dev/da4 Bay 5 Lnum 7 [HEALTHY] SN:Z296M8XB 00009330KYE03 /dev/da5 Bay 6 Lnum 6 [HEALTHY] SN:Z295LXT7 000093172YE03 /dev/da6 Bay 7 Lnum 5 [HEALTHY] SN:Z296M8ZF 00009330KYE03 /dev/da7 Bay 8 Lnum 4 [HEALTHY] SN:Z296M8SD 00009330EYE03 /dev/da8 Bay 9 Lnum 3 [HEALTHY] SN:Z296M8QA 00009330EYE03 /dev/da9 Bay 10 Lnum 2 [HEALTHY] SN:Z296M8Q7 00009330EYE03 /dev/da10 Bay 11 Lnum 1 [HEALTHY] SN:Z296M8SP 00009330EYE04 /dev/da11 Bay 12 Lnum 0 [HEALTHY] SN:Z296M8QZ 00009330JYE03 /dev/da12

A drive in bay 3 is smartfailed:

my-cluster# isi devisi dev Node 1, [ATTN] Bay 1 Lnum 11 [REPLACE] SN:Z296M8HK 000093172YE04 /dev/da1 Bay 2 Lnum 10 [HEALTHY] SN:Z296M8N5 00009330EYE03 /dev/da2 Bay 3 Lnum 9 [SMARTFAIL] SN:Z296LBP4 00009330EYE03 N/A Bay 4 Lnum 8 [HEALTHY] SN:Z296LCJW 00009327BYE03 /dev/da4 Bay 5 Lnum 7 [HEALTHY] SN:Z296M8XB 00009330KYE03 /dev/da5 Bay 6 Lnum 6 [HEALTHY] SN:Z295LXT7 000093172YE03 /dev/da6 Bay 7 Lnum 5 [HEALTHY] SN:Z296M8ZF 00009330KYE03 /dev/da7 Bay 8 Lnum 4 [HEALTHY] SN:Z296M8SD 00009330EYE03 /dev/da8 Bay 9 Lnum 3 [HEALTHY] SN:Z296M8QA 00009330EYE03 /dev/da9 Bay 10 Lnum 2 [HEALTHY] SN:Z296M8Q7 00009330EYE03 /dev/da10 Bay 11 Lnum 1 [HEALTHY] SN:Z296M8SP 00009330EYE04 /dev/da11 Bay 12 Lnum 0 [HEALTHY] SN:Z296M8QZ 00009330JYE03 /dev/da12

Smartfail is unsuccessful for bay 3, changing the drive state to ERASE. Because the drivecould not be crypto-erased, OneFS attempts to delete the drive password:

my-cluster# isi devNode 1, [ATTN] Bay 1 Lnum 11 [REPLACE] SN:Z296M8HK 000093172YE04 /dev/da1 Bay 2 Lnum 10 [HEALTHY] SN:Z296M8N5 00009330EYE03 /dev/da2 Bay 3 Lnum 9 [ERASE] SN:Z296LBP4 00009330EYE03 /dev/da3 Bay 4 Lnum 8 [HEALTHY] SN:Z296LCJW 00009327BYE03 /dev/da4 Bay 5 Lnum 7 [HEALTHY] SN:Z296M8XB 00009330KYE03 /dev/da5 Bay 6 Lnum 6 [HEALTHY] SN:Z295LXT7 000093172YE03 /dev/da6 Bay 7 Lnum 5 [HEALTHY] SN:Z296M8ZF



00009330KYE03 /dev/da7 Bay 8 Lnum 4 [HEALTHY] SN:Z296M8SD 00009330EYE03 /dev/da8 Bay 9 Lnum 3 [HEALTHY] SN:Z296M8QA 00009330EYE03 /dev/da9 Bay 10 Lnum 2 [HEALTHY] SN:Z296M8Q7 00009330EYE03 /dev/da10 Bay 11 Lnum 1 [HEALTHY] SN:Z296M8SP 00009330EYE04 /dev/da11 Bay 12 Lnum 0 [HEALTHY] SN:Z296M8QZ 00009330JYE03 /dev/da12


Smartfailed drive ERASE and REPLACE state examples 289

CHAPTER 17

SmartQuotas


u SmartQuotas overview........................................................................................ 292u Quota types........................................................................................................ 292u Default quota type...............................................................................................293u Usage accounting and limits............................................................................... 294u Disk-usage calculations...................................................................................... 296u Quota notifications............................................................................................. 297u Quota notification rules.......................................................................................297u Quota reports......................................................................................................298u Creating quotas...................................................................................................298u Managing quotas................................................................................................ 300u Managing quota notifications..............................................................................303u Managing quota reports...................................................................................... 306u Basic quota settings............................................................................................307u Advisory limit quota notification rules settings.................................................... 308u Soft limit quota notification rules settings........................................................... 309u Hard limit quota notification rules settings..........................................................309u Limit notification settings....................................................................................310u Quota report settings.......................................................................................... 310u Custom email notification template variable descriptions................................... 311

SmartQuotas 291

SmartQuotas overviewThe SmartQuotas module is an optional quota-management tool that monitors andenforces administrator-defined storage limits. Using accounting and enforcement quotalimits, reporting capabilities, and automated notifications, SmartQuotas managesstorage use, monitors disk storage, and issues alerts when disk-storage limits areexceeded.

Quotas help you manage storage usage according to criteria that you define. Quotas areused as a method of tracking—and sometimes limiting—the amount of storage that auser, group, or project consumes. Quotas are a useful way of ensuring that a user ordepartment does not infringe on the storage that is allocated to other users ordepartments. In some quota implementations, writes beyond the defined space aredenied, and in other cases, a simple notification is sent.

The SmartQuotas module requires a separate license. For additional information aboutthe SmartQuotas module or to activate the module, contact your EMC Isilon salesrepresentative.

Quota typesOneFS uses the concept of quota types as the fundamental organizational unit of storagequotas. Storage quotas comprise a set of resources and an accounting of each resourcetype for that set. Storage quotas are also called storage domains.

Storage quotas creation requires three identifiers:

u The directory to monitor.

u Whether snapshots are to be tracked against the quota limit.

u The quota type (directory, user, or group).

You can choose a quota type from the following entities:

DirectoryA specific directory and its subdirectories.

UserEither a specific user or default user (every user). Specific-user quotas that youconfigure take precedence over a default user quota.

GroupAll members of a specific group or all members of a default group (every group). Anyspecific-group quotas that you configure take precedence over a default groupquota. Associating a group quota with a default group quota creates a linked quota.

You can create multiple quota types on the same directory, but they must be of a differenttype or have a different snapshot option. You can specify quota types for any directory inOneFS and nest them within each other to create a hierarchy of complex storage-usepolicies.

Nested storage quotas can overlap. For example, the following quota settings ensure thatthe finance directory never exceeds 5 TB, while limiting the users in the financedepartment to 1 TB each:

u Set a 5 TB hard quota on /ifs/data/finance.

u Set 1 TB soft quotas on each user in the finance department.

SmartQuotas


Note

You should not create quotas of any type on the OneFS root (/ifs). A root-level quota

may significantly degrade performance.

Default quota typeDefault quotas automatically create other quotas for users or groups in a specifieddirectory.

A default quota specifies a policy for new entities that match a trigger. The default-user@/ifs/cs becomes specific-user@/ifs/cs for each specific-user that is not otherwisedefined.

For example, you can create a default-user quota on the /ifs/dir-1 directory, wherethat directory is owned by the root user. The default-user type automatically creates anew domain on that directory for root and adds the usage there:my-OneFS-1# mkdir /ifs/dir-1my-OneFS-1# isi quota quotas create /ifs/dir-1 default-usermy-OneFS-1# isi quota quotas ls --path=/ifs/dir-1 Type AppliesTo Path Snap Hard Soft Adv Used---------------------------------------------------------------default-user DEFAULT /ifs/dir-1 No - - - 0buser root /ifs/dir-1 No - - - 0b---------------------------------------------------------------

Now add a file that is owned by a different user (admin).my-OneFS-1# touch /ifs/dir-1/somefilemy-OneFS-1# chown admin /ifs/dir-1/somefilemy-OneFS-1# isi quota quotas ls --path=/ifs/dir-1Type AppliesTo Path Snap Hard Soft Adv Used---------------------------------------------------------------default-user DEFAULT /ifs/dir-1 No - - - 0buser root /ifs/dir-1 No - - - 26buser admin /ifs/dir-1 No - - - 0b---------------------------------------------------------------Total: 3

In this example, the default-user type created a new specific-user type automatically(user:admin) and added the new usage to it. Default-user does not have any usagebecause it is used only to generate new quotas automatically. Default-user enforcementis copied to a specific-user (user:admin), and the inherited quota is called a linked quota.In this way, each user account gets its own usage accounting.

Defaults can overlap. For example, default-user@/ifs/dir-1 and default-user@/ifs/csboth may be defined. If the default enforcement changes, OneFS storage quotaspropagate the changes to the linked quotas asynchronously. Because the update isasynchronous, there is some delay before updates are in effect. If a default type, such asevery user or every group, is deleted, OneFS deletes all children that are marked asinherited. As an option, you can delete the default without deleting the children, but it isimportant to note that this action breaks inheritance on all inherited children.

Continuing with the example, add another file that is owned by the root user. Because theroot type exists, the new usage is added to it.

my-OneFS-1# touch /ifs/dir-1/anotherfilemy-OneFS-1# isi quota ls -v --path=/ifs/dir-1 --format=list Type: default-user AppliesTo: DEFAULT Path: /ifs/dir-1 Snap: NoThresholds Hard : -

SmartQuotas

Default quota type 293

Soft : - Adv : - Grace : - Usage Files : 0 With Overhead : 0.00b W/O Overhead : 0.00b Over: - Enforced: No Container: No Linked: ----------------------------------------------------------------------- Type: user AppliesTo: root Path: /ifs/dir-1 Snap: NoThresholds Hard : - Soft : - Adv : - Grace : - Usage Files : 2 With Overhead : 3.50K W/O Overhead : 55.00b Over: - Enforced: No Container: No Linked: Yes----------------------------------------------------------------------- Type: user AppliesTo: admin Path: /ifs/dir-1 Snap: NoThresholds Hard : - Soft : - Adv : - Grace : - Usage Files : 1 With Overhead : 1.50K W/O Overhead : 0.00b Over: - Enforced: No Container: No Linked: Yes

The enforcement on default-user is copied to the specific-user when the specific-userallocates within the type, and the new inherited quota type is also a linked quota.

Note

Configuration changes for linked quotas must be made on the parent quota that thelinked quota is inheriting from. Changes to the parent quota are propagated to allchildren. To override configuration from the parent quota, you must unlink the quota first.

Usage accounting and limitsStorage quotas support two usage types that you can create to manage storage space.The usage types are accounting and enforcement limits.

You can configure OneFS quotas by usage type to track or limit storage use. Theaccounting option, which monitors disk-storage use, is useful for auditing, planning, andbilling. Enforcement limits set storage limits for users, groups, or directories.

SmartQuotas


AccountingThe accounting option tracks but does not limit disk-storage use. Using theaccounting option for a quota, you can monitor inode count and physical and logicalspace resources. Physical space refers to all of the space used to store files anddirectories, including data and metadata in the domain. Logical space refers to thesum of all files sizes, excluding file metadata and sparse regions. User data storageis tracked using logical-space calculations, which do not include protectionoverhead. As an example, by using the accounting option, you can do the following:

u Track the amount of disk space used by various users or groups to bill each user,group, or directory for only the disk space used.

u Review and analyze reports that help you identify storage usage patterns anddefine storage policies.

u Plan for capacity and other storage needs.

Enforcement limitsEnforcement limits include all of the functionality of the accounting option, plus theability to limit disk storage and send notifications. Using enforcement limits, you canlogically partition a cluster to control or restrict how much storage that a user, group,or directory can use. For example, you can set hard- or soft-capacity limits to ensurethat adequate space is always available for key projects and critical applications andto ensure that users of the cluster do not exceed their allotted storage capacity.Optionally, you can deliver real-time email quota notifications to users, groupmanagers, or administrators when they are approaching or have exceeded a quotalimit.

Note

If a quota type uses the accounting-only option, enforcement limits cannot be used forthat quota.

The actions of an administrator logged in as root may push a domain over a quotathreshold. For example, changing the protection level or taking a snapshot has thepotential to exceed quota parameters. System actions such as repairs also may push aquota domain over the limit.

The system provides three types of administrator-defined enforcement thresholds.

Thresholdtype

Description

- -Hard Limits disk usage to a size that cannot be exceeded. If an operation, such as a file

write, causes a quota target to exceed a hard quota, the following events occur:

l the operation fails

l an alert is logged to the cluster

l a notification is issued to specified recipients.

Writes resume when the usage falls below the threshold.

Soft Allows a limit with a grace period that can be exceeded until the grace periodexpires. When a soft quota is exceeded, an alert is logged to the cluster and anotification is issued to specified recipients; however, data writes are permittedduring the grace period.

SmartQuotas

Usage accounting and limits 295

Thresholdtype

Description

- -If the soft threshold is still exceeded when the grace period expires, data writesfail, and a hard-limit notification is issued to the recipients you have specified.

Writes resume when the usage falls below the threshold.

Advisory An informational limit that can be exceeded. When an advisory quota threshold isexceeded, an alert is logged to the cluster and a notification is issued to specifiedrecipients. Advisory thresholds do not prevent data writes.

Disk-usage calculationsFor each quota that you configure, you can specify whether data-protection overhead isincluded in future disk-usage calculations.

Most quota configurations do not need to include overhead calculations. If you do notinclude data-protection overhead in usage calculations for a quota, future disk-usagecalculations for the quota include only the space that is required to store files anddirectories. Space that is required for the data-protection setting of the cluster is notincluded.

Consider the same example user, who is now restricted by a 40 GB quota that does notinclude data-protection overhead in its disk-usage calculations. If your cluster isconfigured with a 2x data-protection level and the user writes a 10 GB file to the cluster,that file consumes 20 GB of space but the 10GB for the data-protection overhead is notcounted in the quota calculation. In this example, the user has reached 25 percent of the40 GB quota by writing a 10 GB file to the cluster. This method of disk-usage calculationis recommended for most quota configurations.

If you include data-protection overhead in usage calculations for a quota, future disk-usage calculations for the quota include the total amount of space that is required tostore files and directories, in addition to any space that is required to accommodate yourdata-protection settings, such as parity or mirroring. For example, consider a user who isrestricted by a 40 GB quota that includes data-protection overhead in its disk-usagecalculations. If your cluster is configured with a 2x data-protection level (mirrored) andthe user writes a 10 GB file to the cluster, that file actually consumes 20 GB of space: 10GB for the file and 10 GB for the data-protection overhead. In this example, the user hasreached 50 percent of the 40 GB quota by writing a 10 GB file to the cluster.

Note

Cloned and deduplicated files are treated as ordinary files by quotas. If the quotaincludes data protection overhead, the data protection overhead for shared data is notincluded in the usage calculation.

You can configure quotas to include the space that is consumed by snapshots. A singlepath can have two quotas applied to it: one without snapshot usage, which is the default,and one with snapshot usage. If you include snapshots in the quota, more files areincluded in the calculation than are in the current directory. The actual disk usage is thesum of the current directory and any snapshots of that directory. You can see whichsnapshots are included in the calculation by examining the .snapshot directory for thequota path.

SmartQuotas


Note

Only snapshots created after the QuotaScan job finishes are included in the calculation.

Quota notificationsQuota notifications are generated for enforcement quotas, providing users withinformation when a quota violation occurs. Reminders are sent periodically while thecondition persists.

Each notification rule defines the condition that is to be enforced and the action that is tobe executed when the condition is true. An enforcement quota can define multiplenotification rules. When thresholds are exceeded, automatic email notifications can besent to specified users, or you can monitor notifications as system alerts or receiveemails for these events.

Notifications can be configured globally, to apply to all quota domains, or be configuredfor specific quota domains.

Enforcement quotas support the following notification settings. A given quota can useonly one of these settings.

Limit notification settings Description- -Turn Off Notifications for thisQuota

Disables all notifications for the quota.

Use Default Notification Rules Uses the global default notification for the specified type ofquota.

Use Custom Notification Rules Enables the creation of advanced, custom notifications thatapply to the specific quota. Custom notifications can beconfigured for any or all of the threshold types (hard, soft, oradvisory) for the specified quota.

Quota notification rulesYou can write quota notification rules to generate alerts that are triggered by eventthresholds.

When an event occurs, a notification is triggered according to your notification rule. Forexample, you can create a notification rule that sends an email when a disk-spaceallocation threshold is exceeded by a group.

You can configure notification rules to trigger an action according to event thresholds (anotification condition). A rule can specify a schedule, such as "every day at 1:00 AM," forexecuting an action or immediate notification of certain state transitions. When an eventoccurs, a notification trigger may execute one or more actions, such as sending an emailor sending a cluster alert to the interface. The following examples demonstrate the typesof criteria that you can use to configure notification rules.

u Notify when a threshold is exceeded; at most, once every 5 minutesu Notify when allocation is denied; at most, once an houru Notify while over threshold, daily at 2 AMu Notify while grace period expired weekly, on Sundays at 2 AM

Notifications are triggered for events grouped by the following categories:

SmartQuotas

Quota notifications 297

Instant notificationsIncludes the write-denied notification, triggered when a hard threshold denies awrite, and the threshold-exceeded notification, triggered at the moment a hard, soft,or advisory threshold is exceeded. These are one-time notifications because theyrepresent a discrete event in time.

Ongoing notificationsGenerated on a scheduled basis to indicate a persisting condition, such as a hard,soft, or advisory threshold being over a limit or a soft threshold's grace period beingexpired for a prolonged period.

Quota reportsThe OneFS SmartQuotas module provides reporting options that enable administrators tomanage cluster resources and analyze usage statistics.

Storage quota reports provide a summarized view of the past or present state of thequota domains. After raw reporting data is collected by OneFS, you can produce datasummaries by using a set of filtering parameters and sort types. Storage-quota reportsinclude information about violators, grouped by threshold types. You can generatereports from a historical data sample or from current data. In either case, the reports areviews of usage data at a given time. OneFS does not provide reports on data aggregatedover time, such as trending reports, but you can use raw data to analyze trends. There isno configuration limit on the number of reports other than the space needed to storethem.

OneFS provides three methods of data collection and reporting:

u Scheduled reports are generated and saved on a regular interval.

u Ad hoc reports are generated and saved at the request of the user.

u Live reports are generated for immediate and temporary viewing.

Scheduled reports are placed by default in the /ifs/.isilon/smartquotas/reports directory, but the location is configurable to any directory under /ifs. Eachgenerated report includes quota domain definition, state, usage, and globalconfiguration settings. By default, ten reports are kept at a time, and older reports arepurged. You can create ad hoc reports at any time to view the current state of the storagequotas system. These live reports can be saved manually. Ad hoc reports are saved to alocation that is separate from scheduled reports to avoid skewing the timed-report sets.

Creating quotasYou can create two types of storage quotas to monitor data: accounting quotas andenforcement quotas. Storage quota limits and restrictions can apply to specific users,groups, or directories.

The type of quota that you create depends on your goal.

u Enforcement quotas monitor and limit disk usage. You can create enforcementquotas that use any combination of hard limits, soft limits, and advisory limits.

Note

Enforcement quotas are not recommended for snapshot-tracking quota domains.

u Accounting quotas monitor, but do not limit, disk usage.

SmartQuotas


Note

After you create a new quota, it begins to report data almost immediately, but the data isnot valid until the QuotaScan job completes. Before using quota data for analysis or otherpurposes, verify that the QuotaScan job has finished.

Create an accounting quotaYou can create an accounting quota to monitor but not limit disk usage.

Optionally, you can include snapshot data, data-protection overhead, or both in theaccounting quota.

Procedure

1. Click File System Management > SmartQuotas > Quotas & Usage.

2. On the Storage Quotas & Usage page, click Create a storage quota.

3. From the Quota Type list, select the target for this quota: a directory, user, or group.

4. Depending on the target that you selected, select the entity that you want to apply thequota to. For example, if you selected User from the Quota Type list, you can targeteither all users or a specific user.

5. In the Directory path field, type the path and directory for the quota, or click Browse,and then select a directory.

6. Optional: In the Usage Accounting area, select the options that you want.

l To include snapshot data in the accounting quota, select the Include SnapshotData check box.

l To include the data-protection overhead in the accounting quota, select theInclude Data-Protection Overhead check box.

l To include snapshot data in the accounting quota, select the Include SnapshotData check box.

7. In the Usage Limits area, click No Usage Limit (Accounting Only).

8. Click Create Quota.

After you finish

After you create a quota, it begins to report data almost immediately, but the data is notvalid until the QuotaScan job completes. Before using quota data for analysis or otherpurposes, verify that the QuotaScan job has finished.

Create an enforcement quotaYou can create an enforcement quota to monitor and limit disk usage.

You can create enforcement quotas that set hard, soft, and advisory limits.

Procedure


2. On the Storage Quotas & Usage page, click Create a storage quota.

3. From the Quota Type list, select the target for this quota: a directory, user, or group.

4. Depending on the target that you selected, select the entity that you want to apply thequota to. For example, if you selected User from the Quota Type list, you can target allusers or a specific user.

SmartQuotas

Create an accounting quota 299

5. In the Directory path field, type the path and directory for the quota, or click Browse,and then select a directory.

6. Optional: In the Usage Accounting area, click the Include Snapshot Data check box,the Include Data-Protection Overhead check box, or both to include them in thequota.

7. In the Usage Limits area, click Specify Usage Limits.

8. Click the check box next to the option for each type of limit that you want to enforce.

9. Type numerals in the fields and select from the lists the values that you want to usefor the quota.

10.In the Limit Notations area, click the notification option that you want to apply to thequota.

11.To generate an event notification, select the Create cluster event check box.

12.Optional: If you selected the option to use custom notification rules, click the link toexpand the custom notification type that applies to the usage-limit selections.

13.Click Create Quota.

After you finish

After you create a quota, it begins to report data almost immediately but the data is notvalid until the QuotaScan job completes. Before using quota data for analysis or otherpurposes, verify that the QuotaScan job has finished.

Managing quotasYou can modify the configured values of a storage quota, and you can enable or disable aquota. You can also create quota limits and restrictions that apply to specific users,groups, or directories.

Quota management in OneFS is simplified by the quota search feature, which helps youto locate a quota or quotas by using filters. You can unlink quotas that are associatedwith a parent quota, and configure custom notifications for quotas. You can also disablea quota temporarily and then enable it when needed.

Note

Moving quota directories across quota domains is not supported.

Search for quotasYou can search for a quota using a variety of search criteria.

By default, all storage quotas and display options are listed on this page before youapply report or search filters. If the Quotas & Storage section is collapsed, click Definequota display.

Procedure


2. In the Quotas & Usage area, for Report Filters, select Search for specific quotas withinthis report.

3. In the Quota Type list, select the quota type that you want to find.

4. If you selected User Quota or Group quota for a quota type, type a full or partial useror group name in the User or Group field.

You can use the wildcard character (*) in the User or Group field.

SmartQuotas


l To search for only default users, select the Only show default users checkbox.

l To search for only default groups, select the Only show default groups check box.

5. In the Directory Path field, type a full or partial path.

You can use the wildcard character (*) in the Directory Path field.

l To search subdirectories, select the Include subdirectories check box.

l To search for only quotas that are in violations, select the Only show quotas forwhich usage limits are currently in violation check box.

6. Optional: Click Update Display.

Quotas that match the search criteria appear in the sections where quotas are listed.

Results

An accounting or enforcement quota with a threshold value of zero is indicated by a dash(–). You can click the column headings to sort the result set.

Note

To clear the result set and display all storage quotas, in the Quotas & Usage area, selectShow all quotas and usage for this report for Report Filters, and then click UpdateDisplay.

Manage quotasQuotas help you monitor and analyze the current or historical use of disk storage. Youcan search for quotas, and you can view, modify, delete, and unlink a quota.

An initial QuotaScan job must run for the default or scheduled quotas, or the datadisplayed may be incomplete.

Before you modify a quota, consider how the changes will affect the file system and endusers.

Note

u The options to edit or delete a quota appear only when the quota is not linked to adefault quota.

u The option to unlink a quota is available only when the quota is linked to a defaultquota.

Procedure


2. From the Quota Report options, select the type of quota report that you want to viewor manage.

l To monitor and analyze current disk storage use, click Show current quotas andusage (Live Report).

l To monitor and analyze historical disk storage use, click Show archived quotareport to select from the list of archived scheduled and manually generated quotareports.

3. For Report Filters, select the filters to be used for this quota report.

l To view all information in the quota report, click Show all quotas and usage forthis report.

SmartQuotas

Manage quotas 301

l To filter the quota report, click Search for specific quotas within this report, andthen select the filters that you want to apply.

4. Click Update Display.

5. Optional: Select a quota to view its settings or to perform the following managementactions.

l To review or edit this quota, click View details.

l To delete this quota, click Delete.

l To unlink a linked quota, click Unlink.

Note

Configuration changes for linked quotas must be made on the parent (default)quota that the linked quota is inheriting from. Changes to the parent quota arepropagated to all children. If you want to override configuration from the parentquota, you must first unlink the quota.

Export a quota configuration fileYou can export quota settings as a configuration file, which can then be imported forreuse to another Isilon cluster. You can also store the exported quota configurations in alocation outside of the cluster.

You can pipe the XML report to a file or directory. The file can then be imported to anothercluster.

Procedure


2. At the command prompt, run the following command:

isi_classic quota list --exportThe quota configuration file displays as raw XML.

Import a quota configuration fileYou can import quota settings in the form of a configuration file that has been exportedfrom another Isilon cluster.

Procedure


2. Navigate to the location of the exported quota configuration file.

3. At the command prompt, run the following command, where <filename> is the name ofan exported configuration file:

isi_classic quota import --from-file=<filename>The system parses the file and imports the quota settings from the configuration file.Quota settings that you configured before importing the quota configuration file areretained, and the imported quota settings are effective immediately.

SmartQuotas


Managing quota notificationsQuota notifications can be enabled or disabled, modified, and deleted.

By default, a global quota notification is already configured and applied to all quotas.You can continue to use the global quota notification settings, modify the globalnotification settings, or disable or set a custom notification for a quota.

Enforcement quotas support four types of notifications and reminders:

u Threshold exceeded

u Over-quota reminder

u Grace period expired

u Write access denied

If a directory service is used to authenticate users, you can configure notificationmappings that control how email addresses are resolved when the cluster sends a quotanotification. If necessary, you can remap the domain that is used for quota emailnotifications and you can remap Active Directory domains, local UNIX domains, or both.

Configure default quota notification settingsYou can configure default global quota notification settings that apply to all quotas of aspecified threshold type.

The custom notification settings that you configure for a quota take precedence over thedefault global notification settings.

Procedure

1. Click File System Management > SmartQuotas > Settings.

2. Optional: On the Quota Settings page, for Scheduled Reporting, select On.

3. Click Change Schedule, and then select a report frequency from the list.

4. Select the reporting schedule options that you want, and then click Select.

5. In the Scheduled Report Archiving area, you can configure the following size anddirectory options:

l To configure the number of live reports that you want to archive, type the numberof reports in the Limit archive size field.

l To specify an archive directory that is different from the default, in the ArchiveDirectory field, type the path or click Browse to select the path.

6. In the Manual Report Archiving area, you can configure the following size anddirectory options:

l To configure the number of live reports that you want to archive, type the numberof reports in the Limit archive size field.

l To specify an archive directory that is different from the default, in the ArchiveDirectory field, type the path or click Browse to select the path.

7. In the Email Mapping Rules area, choose each mapping rule that you want to use byselecting the check box in the Provider Type column.

8. In the Notification Rules area, define default notification rules for each rule type.

l To expand the list of limit notifications rules types, click Default NotificationsSettings.

SmartQuotas

Managing quota notifications 303

l To display default settings options for advisory-limit notification rules, clickAdvisory Limit Notification Rules.

l To set the advisory-limit options that you want, click Event: Advisory Limit ValueExceeded and Event: While Advisory Limit Remains Exceeded.

l To display default settings for soft-limit notifications, click Soft Limit NotificationRules.

l To set the soft-limit options that you want, click Event: Soft Limit Value Exceeded,Event: While Soft Limit Remains Exceeded, Event: Soft Limit Grace Period Expired,and Event: Soft Limit Write Access Denied.

l To display the options for a hard-limit notification rule, click Hard LimitNotification Rules .

l To set the hard-limit options that you want, click Event: Hard Limit Write AccessDenied and Event: While Hard Limit Remains Exceeded.

9. Click Save.

After you finish


Configure custom quota notification rulesYou can configure custom quota notification rules that apply only to a specified quota.

Before you begin

To configure a custom notification rule, an enforcement quota must exist or be in theprocess of being created. To configure notifications for an existing enforcement quota,follow the procedure to modify a quota and then use these steps from the Quota Detailspane of the specific quota.

Quota-specific custom notification rules must be configured for that quota. If notificationrules are not configured for a quota, the default event notification configuration is used.For more information about configuring default notification rules, see Create an eventnotification rule.

Procedure

1. In the Limit Notifications area, click Use Custom Notification Rules.

The links display for the rules options that are available for the type of notificationthat you have selected for this quota.

2. Click the View details, and then click Edit limit notifications.

3. Click the link for the limit notification type that you want to configure for this quota.From the list, select the target for this quota: a directory, user, or group.

The Limit Notification Rules options display for the selection type.

4. Select or type the values to configure the custom notification rule for this quota.

5. Click Create quota when you have completed configuring the settings for thisnotification rule.

Results

The quota appears in the Quotas & Usage list.

SmartQuotas


After you finish


Map an email notification rule for a quotaEmail notification mapping rules control how email addresses are resolved when thecluster sends a quota notification.

If necessary, you can remap the domain used for SmartQuotas email notifications. Youcan remap Active Directory Windows domains, local UNIX domains, or NIS domains.

Note

You must be logged in to the web administration interface to perform this task.

Procedure


2. Optional: In the Email Mapping area, click Create an email mapping rule.

3. From the Provider Type list, select the provider type for this notification rule.

4. From the Current Domain list, select the domain that you want to use for the mappingrule.

5. In the Map-to-Domain field, type the name of the domain that you want to map emailnotifications to.

Repeat this step if you want to map more than one domain.

6. Click Save Rule.

Configure a custom email quota notification templateIf email notifications are enabled, you can configure custom templates for emailnotifications.

If the default email notification templates do not meet your needs, you can configure yourown custom email notification templates using a combination of text and SmartQuotasvariables.

Procedure

1. Open a text editor and create a .txt file that includes any combination of text andOneFS email notification variables.

2. Save the template file as ASCII text or in ISO-8859-1 format.

3. Upload the file to an appropriate directory on the Isilon cluster.

For example, /ifs/templates.

Example 1 Example of a custom quota email notification text file

The following example illustrates a custom email template to notify recipients about anexceeded quota.Text-file contents with variables

The disk quota on directory <ISI_QUOTA_PATH> owned by<ISI_QUOTA_OWNER> was exceeded.The <ISI_QUOTA_TYPE> quota limit is <ISI_QUOTA_THRESHOLD>, and<ISI_QUOTA_USAGE> is in use. Please free some disk space

SmartQuotas

Map an email notification rule for a quota 305

Example 1 Example of a custom quota email notification text file (continued)

by deleting unnecessary files.For more information, contact Jane Anderson in IT.

Email contents with resolved variables

The disk quota on directory /ifs/data/sales_tools/collateral owned by jsmith was exceeded.The hard quota limit is 10 GB, and 11 GB is in use. Pleasefree some disk space by deleting unnecessary files.For more information, contact Jane Anderson in IT.

After you finish

To use the custom template, click Cluster Managements > General Settings > EmailSettings, and select the custom template in the Event Notification Settings area.

Managing quota reportsYou can configure and schedule reports to help you monitor, track, and analyze storageuse on an Isilon cluster.

You can view and schedule reports and customize report settings to track, monitor, andanalyze disk storage use. Quota reports are managed by configuring settings that giveyou control over when reports are scheduled, how they are generated, where and howmany are stored, and how they are viewed. The maximum number of scheduled reportsthat are available for viewing in the web-administration interface can be configured foreach report type. When the maximum number of reports are stored, the system deletesthe oldest reports to make space for new reports as they are generated.

Create a quota report scheduleYou can configure quota report settings to generate the quota report on a specifiedschedule.

These settings determine whether and when scheduled reports are generated, and whereand how the reports are stored. If you disable a scheduled report, you can still rununscheduled reports at any time.

Procedure


2. Optional: On the Quota settings page, for Scheduled Reporting, click On.

The Report Frequency option appears.

3. Click Change schedule, and select the report frequency that you want to set from thelist.

4. Select the reporting schedule options that you want.

5. Click Save.

Results

Reports are generated according to your criteria and can be viewed in the GeneratedReports Archive.

SmartQuotas


Generate a quota reportIn addition to scheduled quota reports, you can generate a report to capture usagestatistics at a point in time.

Before you begin

Quotas must exist and the initial QuotaScan job must run before you can generate aquota report.

Procedure

1. Click File System Management > SmartQuotas > Generated Reports Archive.

2. In the Generated Quota Reports Archive area, click Generate a quota report.

3. Click Generate Report.

Results

The new report appears in the Quota Reports list.

Locate a quota reportYou can locate quota reports, which are stored as XML files, and use your own tools andtransforms to view them.

Procedure


2. Navigate to the directory where quota reports are stored. The following path is thedefault quota report location:

/ifs/.isilon/smartquotas/reportsIf quota reports are not in the default directory, you can run the isi quotasettings command to find the directory where they are stored.

3. At the command prompt, run one of the following commands:

Options Description

To view a list of all quota reports in the specifieddirectory

Run the following command:

ls -a *.xml

To view a specific quota report in the specifieddirectory

Run the following command:

ls <filename>.xml

Basic quota settingsWhen you create a storage quota, the following attributes must be defined, at aminimum. When you specify usage limits, additional options are available for definingyour quota.

Option Description- -Directory Path The directory that the quota is on.

SmartQuotas

Generate a quota report 307

Option Description- -User Quota Select to automatically create a quota for

every current or future user that stores data inthe specified directory.

Group Quota Select to automatically create a quota forevery current or future group that stores datain the specified directory.

Include Snapshot Data Select to count all snapshot data in usagelimits; cannot be changed after the quota iscreated.

Include Data-Protection Overhead Select to count protection overhead in usagelimits.

No Usage Limit Select to account for usage only.

Specify Usage Limits Select to enforce advisory, soft, or absolutelimits.

Advisory limit quota notification rules settingsYou can configure custom quota notification rules for advisory limits for a quota. Thesesettings are available when you select the option to use custom notification rules.

Option Description Exceeded Remainsexceeded

- - - -Send email Specify the type of email to use. Yes Yes

Notify owner Select to send an email notification to the owner of the entity. Yes Yes

Notify another Select to send an email notification to another recipient and type therecipient's email address.

Yes Yes

Message template Select from the following template types for use in formatting emailnotifications:

l Default (leave Message Template field blank to use default)

l Custom

Yes Yes

Create cluster event Select to generate an event notification for the quota when exceeded. Yes Yes

Delay Specify the length of time (hours, days, weeks) to delay before generating anotification.

Yes No

Frequency Specify the notification and alert frequency: daily, weekly, monthly, yearly;depending on selection, specify intervals, day to send, time of day, multipleemails per rule.

No Yes

SmartQuotas


Soft limit quota notification rules settingsYou can configure custom soft limit notification rules for a quota. These settings areavailable when you select the option to use custom notification rules.

Option Description Exceeded Remainsexceeded

Graceperiodexpired

Writeaccessdenied

- - - - - -Send email Specify the recipient of the email notification. Yes Yes Yes Yes

Notify owner Select to send an email notification to the owner ofthe entity.

Yes Yes Yes Yes

Notify another Select to send an email notification to anotherrecipient and type the recipient's email address.

Yes Yes Yes Yes

Messagetemplate

Select from the following template types for use informatting email notifications:

l Default (leave Message Template field blankto use default)

l Custom

Yes Yes Yes Yes

Create clusterevent

Select to generate an event notification for thequota.

Yes Yes Yes Yes

Delay Specify the length of time (hours, days, weeks) todelay before generating a notification.

Yes No No Yes

Frequency Specify the notification and alert frequency: daily,weekly, monthly, yearly; depending on selection,specify intervals, day to send, time of day, multipleemails per rule.

No Yes Yes No

Hard limit quota notification rules settingsYou can configure custom quota notification rules for hard limits for a quota. Thesesettings are available when you select the option to use custom notification rules.

Option Description Write accessdenied

Exceeded

- - - -Send email Specify the recipient of the email notification. Yes Yes

Notify owner Select to send an email notification to the owner of the entity. Yes Yes

Notify another Select to send an email notification to another recipient and type therecipient's email address.

Yes Yes

Message template Select from the following template types for use in formatting emailnotifications:

l Default (leave Message Template field blank to use default)

Yes Yes

SmartQuotas

Soft limit quota notification rules settings 309

Option Description Write accessdenied

Exceeded

- - - -l Custom

Create cluster event Select to generate an event notification for the quota when exceeded. Yes Yes

Delay Specify the length of time (hours, days, weeks) to delay before generating anotification.

Yes No

Frequency Specify the notification and alert frequency: daily, weekly, monthly, yearly;depending on selection, specify intervals, day to send, time of day, multipleemails per rule.

No Yes

Limit notification settingsYou have three notification options when you create an enforcement quota: use defaultnotification rules, turn off notifications, or use custom notification rules. Enforcementquotas support the following notification settings for each threshold type. A quota canuse only one of these settings.

Notification setting Description- -Use Default Notification Rules Uses the default notification rules that you configured for

the specified threshold type.

Turn Off Notifications for this Quota Disables all notifications for the quota.

Use Custom Notification Rules Provides settings to create basic custom notifications thatapply to only this quota.

Quota report settingsYou can configure quota report settings that track disk usage. These settings determinewhether and when scheduled reports are generated, and where and how reports arestored. When the maximum number of reports are stored, the system deletes the oldestreports to make space for new reports as they are generated.

Setting Description- -Scheduledreporting

Enables or disables the scheduled reporting feature.

l Off. Manually generated on-demand reports can be run at any time.

l On. Reports run automatically according to the schedule that you specify.

Reportfrequency

Specifies the interval for this report to run: daily, weekly, monthly, or yearly. You can use the following options tofurther refine the report schedule.

Generate report every. Specify the numeric value for the selected report frequency; for example, every 2months.

Generate reports on. Select the day or multiple days to generate reports.

Select report day by. Specify date or day of the week to generate the report.

SmartQuotas



Generate one report per specified by. Set the time of day to generate this report.

Generate multiple reports per specified day. Set the intervals and times of day to generate the report forthat day.

Scheduledreport archiving

Determines the maximum number of scheduled reports that are available for viewing on the SmartQuotas

Reports page.

Limit archive size for scheduled reports to a specified number of reports. Type the integer to specify themaximum number of reports to keep.

Archive Directory. Browse to the directory where you want to store quota reports for archiving.

Manual reportarchiving

Determines the maximum number of manually generated (on-demand) reports that are available for viewing on

the SmartQuotas Reports page.

Limit archive size for live reports to a specified number of reports. Type the integer to specify the maximumnumber of reports to keep.

Archive Directory. Browse to the directory where you want to store quota reports for archiving.

Custom email notification template variable descriptionsIf the default OneFS email notification templates do not meet your needs, you canconfigure and upload your own custom email templates for SmartQuotas notifications.

An email template can contain text and, optionally, variables that represent values. Youcan use any of the SmartQuotas variables in your templates. Template files must besaved as a .txt file.

Variable Description Example- - -ISI_QUOTA_PATH Path of quota domain /ifs/data

ISI_QUOTA_THRESHOLD Threshold value 20 GB

ISI_QUOTA_USAGE Disk space in use 10.5 GB

ISI_QUOTA_OWNER Name of quota domain owner jsmith

ISI_QUOTA_TYPE Threshold type Advisory

ISI_QUOTA_GRACE Grace period, in days 5 days

ISI_QUOTA_EXPIRATION Expiration date of grace period Fri Feb 23 14:23:19 PST2007

ISI_QUOTA_NODE Hostname of the node on which thequota event occurred

someHost-prod-wf-1

SmartQuotas

Custom email notification template variable descriptions 311

CHAPTER 18

Storage Pools


u Storage pools overview....................................................................................... 314u About storage pools............................................................................................ 314u Autoprovisioning.................................................................................................315u Virtual hot spare..................................................................................................316u Spillover..............................................................................................................316u Node pools......................................................................................................... 317u SSD pools........................................................................................................... 318u Tiers....................................................................................................................319u File pools............................................................................................................ 319u File pool policies................................................................................................. 319u Managing node pools..........................................................................................320u Managing tiers.................................................................................................... 321u Creating file pool policies....................................................................................322u Managing file pool policies................................................................................. 327u Monitoring storage pools.................................................................................... 332

Storage Pools 313

Storage pools overviewStorage pools allows you to organize cluster resources as logical groups (called nodepools and tiers), and to create policies that store or move files among these groupsautomatically, based on criteria you specify.

Node pools are sets of like nodes that are grouped into a single pool of storage. Nodepool membership changes automatically through the addition or removal of nodes to orfrom the cluster. Storage pools are node pools and user-defined tiers. File pools arelogical sets of files that are defined by the user, such as all JPG files, all files under thedirectory /ifs/data/ingest, or all files created more than a month ago. Data storagein storage pools is defined by file pool policies. File pool policies identify logical groupsof files, and the actions to perform on them. For example, one policy can store all files ofa specific type and age in a specific node pool. Another can move all files in a specificpath, that have been changed after a specific date, to a specific tier.

If a SmartPools license is not active, all files belong to the default file pool and aregoverned by the default file pool policy. Additional features are available when youactivate the SmartPools license, such as the ability to create multiple file pools and filepool policies that store specified files and directories in a particular storage pool.Spillover management, another feature that is available when you activate a SmartPoolslicense, lets you define how write operations are handled when a storage pool is notwritable. Virtual hot spare allocation, which reserves space for data repair if a drive fails,is available regardless of whether a SmartPools license is active.

The following table compares storage pool features based on whether a SmartPoolslicense is active.

Feature Inactive Active- - -Automatic storage pool provisioning Yes Yes

Directed spillover No Yes

Policy-based data movement No Yes

Virtual hot spare Yes Yes

About storage poolsOneFS provides storage pools to simplify the management and storage of data. Nodepools and tiers are types of storage pools.

Storage pools are the groupings of physical devices where OneFS stores your data. Filepools are logical groupings of files that you define. Node pools are sets of physical nodesthat are grouped by equivalence class to optimize reliability and requested dataprotection settings. OneFS creates node pools automatically when the system is installedand whenever nodes are added or removed. The automatic creation of node pools isreferred to as auto-provisioning. Tiers are collections of node pools that you group tooptimize storage according to need, such as a mission-critical high-speed tier that is bestsuited to data archiving.

File pool policies specify operations on files in the file pool. For example, you can create afile pool policy for a file extension that requires high availability, and then direct thosefiles to a storage pool that provides the fastest reads or read/writes. And you can create

Storage Pools


another file pool policy to evaluate the last accessed date, allowing you to target storagepools best suited for archiving for historical or regulatory purposes.

When you install OneFS, the system creates one file pool for the Isilon cluster. Thisdefault file pool contains all the files in the cluster and is governed by a default file poolpolicy. The default file pool policy operations apply to all files not governed by a higher-priority file pool policy. If you activate a SmartPools license, you can create multiple filepools.

OneFS includes the following basic features.

Default file poolA single set of files of all types that is governed by the default file pool policy.

Node poolsGroups of equivalence-class nodes that are associated in a single pool of storage.

TiersGroups of node pools used to optimize data storage according to your storageneeds.

Activating a SmartPools license adds the following features.

File poolsLogical sets of files that you define by characteristics such as file type, directory, orage.

File pool policiesRules-based filtering and operations that you configure to store data in specifiedstorage pools. By creating file-filtering rules such as file size, type, access time, andlocation to configure a file pool policy, you can configure requested protection andI/O optimization settings and automate data storage according to your needs. If youactivate a SmartPools license, you can create customizable file pool templates thatare optimized for archiving, extra protection, performance, and VMware files.

Storage pool spilloverAutomated node-capacity overflow management. Spillover defines how to handlewrite operations when a storage pool is not writable. When spillover is enabled, datais redirected to a specified storage pool. If spillover is disabled, new data writes failand an error message appears.

Note

If a SmartPools license has not been activated, files are stored on any available nodepools across the cluster.

AutoprovisioningAutoprovisioning is the process of automatically assigning storage by node type toimprove the performance and reliability of the file storage system.

When you configure a cluster, OneFS automatically assigns nodes to node pools in yourcluster. This node assignment is called autoprovisioning. Autoprovisioning reduces thetime required for the manual management tasks associated with configuring storagepools and resource planning.

Storage Pools

Autoprovisioning 315

Note

Nodes are not provisioned, meaning they are not associated with each other and notwritable, until at least three nodes of an equivalence class are added to the cluster. If youhave added only two nodes of an equivalence class to your cluster, no data is stored onthe nodes until you add a third node of the same equivalence-class.

If you remove nodes from a provisioned cluster so that fewer than three equivalence-class nodes remain, the pool is underprovisioned. In this situation, when two like nodesremain, they are still writable; if only one node remains, it is not writable but it remainsreadable.

Virtual hot spareVirtual hot spare allocation settings allow you to allocate space to use to repair data inthe event a drive fails.

Virtual hot spare reserves the free space needed to rebuild the data if a disk fails. Forexample, if you specify two virtual drives and 15 percent, each node pool reserves virtualdrive space that is equivalent to two drives or 15 percent of their total capacity (whichever is larger) to rebuild a failed drive.

You can reserve space in node pools across the cluster for this purpose, up to theequivalent of a maximum of four full drives. You define virtual hot spare allocation usingthese options:u A minimum number of virtual drives in the node pool (1-4).u A minimum percentage of total disk space (0-20 percent) .u A combination of minimum virtual drives and total disk space.

Note

The larger number of the two factors (minimum number of virtual drives or percentageof total disk space), rather than their sum, determines the space allocated for virtualhot spare.

It is important to understand the following information when configuring VHS settings:u If you configure both settings, the enforced minimum value satisfies both

requirements.u If you select the option to reduce the amount of available space, free-space

calculations do not include the space reserved for the virtual hot spare. The reservedvirtual hot spare free space is used for write operations unless you select the optionto deny new data writes. If Reduce amount of available space is enabled while Denynew data writes is disabled, it is possible for the file system to report utilization asmore than 100 percent.

Note

Virtual hot spare reservations affect spillover. If the virtual hot spare option Deny writesis enabled but Reduce amount of available space is disabled, spillover occurs before thefile system reports 100% utilization.

SpilloverIf you activate a SmartPools license, you can designate a storage pool to receive spilldata when the hardware specified by a file pool policy is not writable. If you do not want

Storage Pools


data to spill over from a different location because the specified node pool or tier is full ornot writable, you can disable this feature.

Spillover management is available after you activate a SmartPools license. You can directwrite operations to a specified storage pool in the cluster when there is not enough spaceto write a file according to the storage pool policy.

Note

Virtual hot spare reservations affect spillover. If the setting Deny writes is enabled butDeduce amount of available space is disabled, spillover occurs before the file systemreports 100% utilization.

Node poolsA node pool is a grouping of equivalence-class nodes. As you add nodes to the OneFScluster, OneFS automatically place them into pools. This is referred to asautoprovisioning.

Each node in the OneFS cluster is a peer, and any node can handle a data request. Youcan apply file pool policies to move files to specific node pools or tiers that have differentperformance and capacity characteristics. Each node added to a cluster increases thecluster's aggregate disk, cache, CPU, and network capacity. OneFS automatically addsnodes to specific node pools based on matching characteristics such as drive size, RAM,series, and SSD-node ratio. Nodes with similar characteristics are called equivalence-class nodes.

Note

If you attempt to remove nodes from either a manually managed or automaticallymanaged node pool so that the removal leaves only one or two nodes in the pool, theremoval fails. You can, however, move all nodes from an automatically managed nodepool into one that is manually managed.

When you remove a node from a manually managed node pool, OneFS autoprovisions thenode into a node pool of the same equivalence class.

Manual node pool managementYou can manually provision nodes from an existing node pool into a node pool that youdefine.

If the node pools autoprovisioned by OneFS do not meet your needs, you can create nodepools manually by moving nodes into a pool that you specify and configure. This enablesyou to use SmartPools to store data on specific nodes according to your purposes.

CAUTION

Manually managed node pools may not provide the same performance and efficiency asautomatically managed node pools, particularly if your changes result in fewer than 20nodes in the manually managed node pool. For this reason, it is best to allow OneFS toautoprovision nodes unless you have an advanced understanding of how theSmartPools feature works.

Storage Pools

Node pools 317

SSD poolsOneFS clusters can contain both HDDs and SSDs. When OneFS autoprovisions nodes, itplaces nodes with SSDs into equivalence-class node pools. The SSD strategy defined inthe default file pool determines how SSD nodes are used within the cluster.

Clusters that include both hard-disk drives (HDDs) and solid-state drives (SSDs) areoptimized by your SSD strategy options to increase performance across a wide range ofworkflows. You can configure file pool policies to apply specific SSD strategies asneeded. When you select SSD options during the creation of a file pool policy, you canidentify the directories and files in the OneFS cluster that require faster or slowerperformance. When the file pool policy runs, OneFS automatically moves that data to theappropriate storage pool and drive type.

Global namespace acceleration (GNA) allows data stored on node pools without SSDs touse SSDs elsewhere in the cluster to store extra metadata mirrors. Extra metadata mirrorsaccelerate metadata read operations. You can only enable GNA if 20% or more of thenodes in the cluster contain at least one SSD and 1.5% or more of the total clusterstorage is SSD-based. For best results, ensure that at least 2.0% of the total clusterstorage is SSD-based before enabling global namespace acceleration.

Note

GNA requires a minimum of 20% of accessible cluster nodes to have SSD drives. If theratio of SSD to non-SSD nodes falls below this threshold, GNA is not active even ifenabled. GNA is reactivated when the ratio is corrected. When GNA is inactive, existingSSD mirrors are readable but newly written metadata does not include the extra SSDmirror.

The following SSD strategy options are listed in order of slowest to fastest choices:

Avoid SSDsWrites all associated file data and metadata to HDDs only.

CAUTION

Use this option to free SSD space only after consulting with Isilon Technical Supportpersonnel. Using this strategy may negatively affect performance.

Metadata read accelerationWrites both file data and metadata to HDDs. This is the default setting. An extramirror of the file metadata is written to SSDs, if available. The SSD mirror is inaddition to the number required to satisfy the requested protection. Enabling GNAmakes read acceleration available to files in node pools that do not contain SSDs.GNA is only for metadata and extra mirrors.

Metadata read/write accelerationWrites file data to HDDs and metadata to SSDs, when available. This strategyaccelerates metadata writes in addition to reads but requires about four to five timesmore SSD storage than the Metadata read acceleration setting. Enabling GNA doesnot affect read/write acceleration.

Storage Pools


Data on SSDsUses SSD node pools for both data and metadata, regardless of whether globalnamespace acceleration is enabled. This SSD strategy does not result in the creationof additional mirrors beyond the normal requested protection but requiressignificantly increased storage requirements compared with the other SSD strategyoptions.

TiersA tier is a user-defined collection of node pools that you can specify as a storage pool forfiles. A node pool can belong to only one tier.

You can create tiers to assign your data to any of the node pools in the tier. For example,you can assign a collection of node pools to a tier specifically created to store data thatrequires high availability and fast access. In a three-tier system, this classification maybe Tier 1. You can classify data that is used less frequently or that is accessed by fewerusers as Tier-2 data. Tier 3 usually comprises data that is seldom used and can bearchived for historical or regulatory purposes.

File poolsFile pools are sets of files that you define to apply policy-based control of the storagecharacteristics of your data.

The initial installation of OneFS places all files in the cluster into a single file pool, whichis subject to the default file pool policy. SmartPools enables you to define additional filepools, and create policies that move files in these pools to specific node pools and tiers.

File pool policies match specific file characteristics (such as file size, type, date of lastaccess or a combination of these and other factors), and define specific storageoperations for files that match them. The following examples demonstrate a few ways youcan configure file pool policies:u You can create a file pool policy for a specific file extension that requires high

availability.u You can configure a file pool policy to store that type of data in a storage pool that

provides the fastest reads or read/writes.u You can create another file pool policy to evaluate last accessed date, allowing you to

store older files in storage pool best suited for archiving for historical or regulatorypurposes.

File pool policiesFile pool policies define file storage among storage pools, optimization for file accesspatterns, and requested file protection settings. SmartPools augments the basic OneFSstorage pool features to give you the ability to create multiple file pools, so you can storefiles in specific node pools or tiers based on criteria you define.

You configure file pool policies with filtering rules and operations that are both system-and user-defined. You can then set requested protection settings and I/O optimizationsettings for files types that you specify. You can include multiple criteria in a file poolpolicy, including time-based filters for the date that a file was last accessed, modified, orcreated. You can also define a relative elapsed time instead of a date, such as three daysbefore the current date.

The unlicensed OneFS SmartPools technology allows you to configure the default file poolpolicy for managing the node pools that are created when the cluster is autoprovisioned.

Storage Pools

Tiers 319

The default file pool contains all files and is stored in any node pool. Default file pooloperations are defined by settings of the default file pool policy.

You cannot reorder or remove the default file pool policy. The settings in the default filepool policy apply to all files that are not covered by another file pool policy. For example,data that is not covered by a file pool policy can be moved to a tier that you identify as adefault for this purpose.

All file pool policy operations are executed when the SmartPools job runs. When new filesare created, OneFS temporarily chooses a storage pool policy, using a mechanism basedon file pool policies used when the last SmartPools job ran. The system may apply newstorage settings and move these files again when the next SmartPools job runs, based ona matching file pool policy.

Managing node poolsYou can add nodes to a tier or change the requested protection for a node pool.

Add or move node pools in a tierYou can group node pools into tiers and move node pools among tiers.

Procedure

1. Click File System Management > SmartPools > Summary.

The SmartPools page appears and displays two groupings: the current capacity usageand a list of tiers and node pools.

2. In the Tiers & Node Pools area, select and drag a node pool to the tier name to add itto the tier.

l To add a node pool that is currently in another tier, expand that tier and drag anddrop the node pool to the target tier name.

l To remove a node pool from a tier, drag and drop the node pool to the cluster icon.

3. Continue dragging and dropping node pools until you complete the tier.

Each node that you added to the tier appears under the tier name when it is in anexpanded state.

Change the name or requested protection of a node poolYou can change the name or the requested protection of a node pool.

Procedure


2. In the Tiers & Node Pools section, in the row of the node pool that you want to modify,click Edit.

A dialog box appears.

3. Enter a name for the node pool, select the requested protection from the list, or doboth.

A node pool name can contain alphanumeric characters and underscores but cannotbegin with a number.

4. Click Submit.

Storage Pools


Managing tiersYou can group node pools into tiers and move node pools among tiers to most efficientlyuse resources or for other cluster management purposes.

Create a tierYou can group node pools into a tier that you can specify as a storage pool for files.

Procedure



2. In the Tiers & Node Pools section, click Create a Tier.

3. In the dialog box that displays, enter a name for the tier, and click Submit.

The tier appears in the list of tiers and node pools.

4. Select and drag a node pool to the tier name to add it to the tier. Continue draggingand dropping node pools until you complete the tiered group.

Each node pool that you added to the tier appears under the tier name when it is in anexpanded state.

Rename a tierYou can modify the name of a tier.

A tier name can contain alphanumeric characters and underscores but cannot begin witha number.

Procedure



2. In the Tiers & Node Pools area, in the row of the tier you want to rename, click Edit.

3. In the dialog box that displays, type a name for this tier and click Submit.

The newly named tier appears in the list of tiers and node pools.

Delete a tierWhen you delete a tier, the nodes it contains become top-level storage pools.

Procedure


The SmartPools page appears and displays two groupings: current capacity usage anda list of tiers and node pools.

2. In the Tiers & Node Pools area, in the row of the tier that you want to delete, clickDelete.

3. In the confirmation dialog box that displays, click Yes to confirm the deletion.

Storage Pools

Managing tiers 321

Results

The tier is removed from the list of tiers and node pools.

Creating file pool policiesYou configure file pool policies to identify logical groups of files, and specify storageoperations for them. You must activate a SmartPools license before you can create filepool policies.

File pool policies have two parts: the criteria that define a logical group of files, and theoperations the policy applies to them. File pool policies support detailed definition of filepools. You can define file pools based on file characteristics, such as file type, size, pathand time of last access. You can combine these criteria with boolean AND and ORoperators. Operations to apply to these groups include requested protection and I/Ooptimization. For example, you can create one file pool policy that identifies all JPG filesin a certain path, that are larger than 2MB AND were last accessed one month ago, andmoves them to a specific node pool. Another policy might take all files that match aspecific matching pattern (all files ending in .xls, for example), that exist in a specificdirectory, and that were accessed in the last week and move them to a pool with SSDnodes. File pool policies make it possible to policies that perform these and otheroperations, quickly and easily.

As many as four file pool policies can apply to a file (one per operation) if the stopprocessing option is not selected. Four file pool policy operations are available:

u Set the data/snapshot storage target (including SSD strategy).

u Set the SmartCache to enabled/disabled.

u Set the requested protection.

u Set the data access pattern.

You can configure each of these settings for a file in four separate file pool policies,include them all in one file pool policy, or use a combination of the two. When theSmartPools job runs, it applies file pool policies in the order of the display. When a filetype matches the criteria defined in the filter settings, the operations are applied. Afterthe list is traversed, if any of the four operations are not applied to a file type, theoperations defined in the default file pool policy then apply to those files. In this way, thedefault file pool policy ensures that all four operations apply to each file, even if anoperation is missing from a user-defined file pool policy.

If a file belongs to multiple file pools with operations specifying the same setting, onlythe operation of the file pool policy with the higher priority applies. Consider thefollowing example: File pool policy 1 targets the X storage pool for all JPG files, and filepool policy 2 targets the NL storage pool for all files created more than month ago.Sometimes files match the criteria in both file pool policies, so those files are stored inthe X storage pool because that file pool policy is higher in the list.

If a file type matches multiple file pool policies, subsequent file pool policies in the listare not evaluated. If one file pool policy operation sets the data storage target of JPG filesto a nearline node pool and a subsequent file pool policy operation sets the data accesspattern to random for all files smaller than 2 MB, then all JPG files smaller than 2 MB aremoved to nearline storage, but their data access pattern is set to random only if the Stopprocessing option is not selected in the first file pool policy. If that option is selected,the data access pattern for JPG files smaller than 2MB is set according to the operationdefined in the default file pool policy.

OneFS provides customizable template policies that archive older files, increase therequested protection for specified files, send files that are saved to a particular path to ahigher-performance storage pool, and change the data access pattern for VMWare files.

Storage Pools


You also can copy any file pool policy except the default file pool policy, and modify thesettings to meet your needs.

Note

You can reorder the file pool policy list at any time, but the default file pool policy isalways last in the list of file pool policies.

Add a file pool policyYou can configure a file pool policy to filter files according to criteria that you specify,including requested protection and I/O optimization settings. Zero or more operationscan be defined per policy.

CAUTION

If existing file pool policies direct data to a specific storage pool, do not configure otherfile pool policies that match this data with anywhere for the Data storage target option.Because the specified storage pool is included when you use anywhere, target specificstorage pools to avoid unintentional file storage locations.

Procedure

1. Click File System Management > SmartPools > File Pool Policies.

The SmartPools page appears and displays three groupings: a list of file pool policies,a list of template policies, and latest scan job results.

2. On the SmartPools page, click Add a file pool policy.

3. In the Basic Settings area, type a policy name and a description for this file poolpolicy.

4. In the Filter Settings areas, click Add criteria.

5. In the Configure File Matching Criteria window, configure file matching criteria andclick Add.

6. Optional: Select or clear the Stop processing more rules for files matching this filtercheck box.

Note

If the stop processing option is selected, only the first matching policy in the list willbe applied if multiple file pool policies are configured for this entity.

7. In the Protections Settings area, choose and configure the target and requestedprotection.

8. In the I/O Optimization Settings area, choose the options that you want this file poolpolicy to apply.

9. Click Submit.

The Storage Pools page appears and the file pool policy that you created is displayedin the list.

Results

The file pool policy is not applied until the next scheduled SmartPools job runs. To runthe job immediately, click Start SmartPools Job.

Storage Pools

Add a file pool policy 323

File pool file-matching optionsYou can configure a file pool policy for files that meet specific criteria.

The following options can be selected in File System Management > SmartPools > FilePool Policies > Filter Settings.

Note

OneFS supports UNIX shell-style (glob) pattern matching for file name attributes andpaths. For information about the characters that can be selected for pattern matching,see Valid wildcard characters.

The following table lists the file attributes that you can use to define a file pool policy.

File attribute Specifies- -File name Includes or excludes files based on the file name.

You can specify whether to include or exclude full or partialnames that contain specific text. Wildcard characters areallowed.

Location (Path) Includes or excludes files based on the file path.

You can specify whether to include or exclude full or partialpaths that contain specified text. You can also include the

wildcard characters *, ?, and [ ].

File type Includes or excludes files based on one of the following file-system object types:

l Regular file

l Directory

l Other

File size Includes or excludes files based on their size.

Note

File sizes are represented in multiples of 1024, not 1000.

Modified time (mtime) Includes or excludes files based on when the file was lastmodified.

You can specify a relative date and time, such as "older than 2weeks," or a specific date and time, such as "before January 1,2012." Time settings are based on a 24-hour clock.

Create time (birthtime) Includes or excludes files based on when the file was created.


Metadata change time (ctime) Includes or excludes files based on when the file metadatawas last modified. This option is available only if the globalaccess-time-tracking option of the cluster is enabled.

Storage Pools


File attribute Specifies- -


Access time (atime) Includes or excludes files based on when the file was lastaccessed based on the following units of time:


Note

Because it affects performance, access time tracking as a filepool policy criterion is disabled by default.

User attribute Includes or excludes files based on a custom user-definedattribute.

Valid wildcard charactersYou can combine wildcard characters with file-matching options to define a file poolpolicy.

OneFS supports UNIX shell-style (glob) pattern matching for file name attributes andpaths. For more information about file criteria that can be used for pattern matching, seeFile pool file-matching options.

The following table lists the valid wildcard characters that you can combine with file-matching options to define a file pool policy.


For example, m* matches movies and m123.

[a-z ] Matches any characters contained in the brackets, or a range of characters separatedby a hyphen.For example, b[aei]t matches bat, bet, and bit.

For example, 1[4-7]2 matches 142, 152, 162, and 172.


For example, b[!ie] matches bat but not bit or bet.


For example, [[c]at matches cat and [at.

You can match a hyphen within a bracket if it is either the first or last character.

For example, car[-s] matches cars and car-.

? Matches any character in place of the question mark.For example, t?p matches tap, tip, and top.

Storage Pools

Valid wildcard characters 325

Default file pool requested protection settingsDefault protection settings include specifying the data storage target, snapshot storagetarget, requested protection, and SSD strategy for files that are filtered by the default filepool policy.

Setting Description Notes- - -Datastoragetarget

Specifies the storage pool that you want totarget with this file pool policy.

CAUTION

If existing file pool policies direct data to aspecific storage pool, do not configure otherfile pool policies with anywhere for the

Data storage target option. Because thespecified storage pool is included when youuse anywhere, target specific storage pools

to avoid unintentional file storage locations.

Select one of the following options to defineyour SSD strategy:

Metadata read accelerationDefault. Write both file data andmetadata to HDDs and metadata toSSDs. Accelerates metadata reads only.

Uses less SSD space than the Metadataread/write acceleration setting.

Metadata read/write accelerationWrite metadata to SSD pools. Usessignificantly more SSD space than

Metadata read acceleration, butaccelerates metadata reads and writes.

Avoid SSDsWrite all associated file data andmetadata to HDDs only.

CAUTION

Use this to free SSD space only afterconsulting with Isilon Technical Supportpersonnel; may negatively affectperformance.

Data on SSDsUse SSDs for both data and metadata.Regardless of whether globalnamespace acceleration is enabled, anySSD blocks reside on the storage targetif there is room.

Note

If GNA is not enabled and thestorage pool that you choose totarget does not contain SSDs, youcannot define a strategy.

Metadata read accelerationwrites both file data and metadatato HDD storage pools but adds anadditional SSD mirror if possible toaccelerate read performance. UsesHDDs to provide reliability and anextra metadata mirror to SSDs, ifavailable, to improve readperformance. Recommended formost uses.

When you select Metadata read/write acceleration, the strategyuses SSDs, if available in thestorage target, for performance andreliability. The extra mirror may befrom a different storage pool usingGNA enabled or from the samenode pool.

Neither the Data on SSDs strategy

nor the Metadata read/writeacceleration strategy result in thecreation of additional mirrorsbeyond the normal requestedprotection. Both file data andmetadata are stored on SSDs ifavailable within the file pool policy.This option requires a significantamount of SSD storage.

Storage Pools


Setting Description Notes- - -Snapshotstoragetarget

Specifies the storage pool that you want totarget for snapshot storage with this file poolpolicy. The settings are the same as those forData storage target, but apply to snapshotdata.

Notes for Data storage target applyto snapshot storage target

Requestedprotection

Default of storage pool. Assign the defaultrequested protection of the storage pool tothe filtered files.

Specific level. Assign a specified requestedprotection to the filtered files.

To change the requestedprotection , select a new value fromthe list.

Default file pool I/O optimization settingsYou can manage the I/O optimization settings that are used in the default file pool policy,including files with manually managed attributes.

To allow SmartPools to overwrite optimization settings that were configured using FileSystem Explorer or the isi set command, select the Including files with manually-managed I/O optimization settings option in the Default Protection Settings group.

Setting Description Notes- - -SmartCache Enables or disables

SmartCache.SmartCache enabled is the recommended setting for optimal writeperformance. With asynchronous writes, the Isilon server buffers writes inmemory. However, if you want to disable this buffering, please configure yourapplications to use synchronous writes. If that is not possible, disableSmartCache.

Data accesspattern

Defines the optimizationsettings for accessing data:Concurrency, Streaming, orRandom.

By default, iSCSI LUNs are configured to use a random access pattern.

Other files and directories use a concurrent access pattern by default.

Managing file pool policiesFile pool policies can be modified, reordered, copied, or removed. The default file poolpolicy can be modified, and template policies can be applied.

You can perform the following file pool policy management tasks:

u Modify file pool policies

u Modify the default file pool policy

u Copy file pool policies

u Use a file pool policy template

u Reorder file pool policies

u Delete file pool policies

Storage Pools

Default file pool I/O optimization settings 327

Configure default file pool policy settingsYou configure file pool policies to filter and store files according to criteria that youdefine. File pool policy settings include protection settings and I/O optimization settings.

Procedure

1. Click File System Management > SmartPools > Settings.

2. On the SmartPools Settings page, select your default settings, and then click Submit.

Results

OneFS applies changes to the default file pool policy settings when the next scheduledSmartPools job runs.

Configure default file pool protection settingsYou can configure default file pool protection settings. The default settings are applied toany file that is not covered by another file pool policy.

CAUTION

If existing file pool policies direct data to a specific storage pool, do not add or modify afile pool policy to the anywhere option for the Data storage target option. Target aspecific file pool instead.

Procedure


2. In the SmartPools Settings section, choose the settings that you want apply as theglobal default for Data storage target, Snapshot storage target, or Protection level.

3. Click Submit.

The settings that you selected are applied to any entity that is not covered by anotherfile pool policy.

Configure default I/O optimization settingsYou can configure default I/O optimization settings.

Procedure


2. In the Default File Pool I/O Optimization Settings area, select the settings that youwant apply as the global default for SmartCache and Data access pattern.

3. Click Submit.

OneFS applies the new settings to all files and directories not covered by another filepool policy.

Storage Pools


Modify a file pool policyYou can modify the name, description, filter criteria, and the protection and I/Ooptimization settings applied by a file pool policy.

CAUTION

If existing file pool policies direct data to a specific storage pool, do not configure otherfile pool policies with anywhere for the Data storage target option. Because the

specified storage pool is included when you use anywhere, target specific storage

pools to avoid unintentional file storage locations.

Procedure


2. In the File Pool Policies area, in the Actions column of the file pool policy you want tomodify, click Copy.

3. Make your changes in the appropriate areas and click Submit.

Results

Changes to the file pool policy are applied when the next SmartPools job runs. To run thejob immediately, click Start SmartPools Job.

Copy a file pool policyYou can copy and modify the settings of any file pool policy with the exception of thedefault file pool policy.

Procedure



2. In the File Pool Policies area, click Copy in the Actions column of the file pool policythat you want to copy.

3. Make changes in the appropriate areas and click Submit.

A copy of the file pool policy is added to the list of policies in the File Pool Policiesarea.

Results

OneFS prefaces the copied policy with Copy of, so that you can differentiate it from thesource policy.

Prioritize a file pool policyFile pool policies are evaluated in descending order according to their position in the filepool policies list.

By default, new policies are inserted immediately above the default file pool policy. Youcan give a policy higher or lower priority by moving it up or down the list. The defaultpolicy is always the last in the list, and applies to all files that are not matched by anyothrer file pool policy.

Storage Pools

Modify a file pool policy 329

Procedure


The SmartPools page appears and displays three groupings: a list of file pool policies,a list of template policies, and the latest scan job results.

2. In the Order column of the File Pool Policies area, select the policy that you want tomove.

3. Click either Move up or Move down until the policy is positioned where you want it inthe order.

Use a file pool template policyYou can use a OneFS template to configure file pool policies.

Procedure


The SmartPools page appears and displays three groups: a list of file pool policies, alist of template policies, and the latest scan job results.

2. In the Action column of the Template Policies area, in the row of the template youwant to use, click Use.

The file pool policy settings options appear, with values pre-configured for the type oftemplate that you selected.

3. Optional: Rename the template or modify the template policy settings.

4. Click Submit.

Results

The policy is added to the File Pool Policies list.

Delete a file pool policyYou can delete any file pool policy except the default policy.

When you delete a file pool policy, the operations for the next matching file pool policyapply and may store them in a different storage pool. Files are not moved to anotherstorage pool until the SmartPools job runs.

Procedure



2. In the File Pool Policies area, in the Actions column of the file pool policy you want toremove, click Delete.

3. In the confirmation dialog box, click Yes to confirm the deletion.

Results

The file pool policy is removed from the list in the File Pool Policies area.

Storage Pools


SmartPools settingsSmartPools settings include directory protection, global namespace acceleration, virtualhot spare, spillover, requested protection management, and I/O optimizationmanagement.

Setting Description Notes- - -Directoryprotection

Increases the amount of protection for directories at a higherlevel than the directories and files that they contain, so thatdata that is not lost can still be accessed.

When devices failures result in data loss (for example, threedrives or two nodes in a +2:1 policy), enabling this settingensures that intact data is still accessible.

The option to Protect directories at one levelhigher should be enabled.

When this setting is disabled, the directory thatcontains a file pool is protected according to yourprotection-level settings, but the devices used tostore the directory and the file may not be thesame. There is potential to lose nodes with filedata intact but not be able to access the databecause those nodes contained the directory.

As an example, consider a cluster that has a +2default file pool protection setting and noadditional file pool policies. OneFS directories arealways mirrored, so they are stored at 3x, which isthe mirrored equivalent of the +2 default.

This configuration can sustain a failure of twonodes before data loss or inaccessibility. If thissetting is enabled, all directories are protected at4x. If the cluster experiences three node failures,although individual files may be inaccessible, thedirectory tree is available and provides access tofiles that are still accessible.

In addition, if another file pool policy protectssome files at a higher level, these too areaccessible in the event of a three-node failure.

Globalnamespaceacceleration

Specifies whether to allow per-file metadata to use SSDs inthe node pool.

l Disabled. Restrict per-file metadata to the storage poolpolicy of the file, except in the case of spillover. This isthe default setting.

l Enabled. Allow per-file metadata to use the SSDs inany node pool.

This setting is available only if 20 percent or moreof the nodes in the cluster contain SSDs and atleast 1.5 percent of the total cluster storage isSSD-based.

Note that if nodes are added to or removed from acluster, and the SSD thresholds are no longersatisfied, GNA becomes inactive. GNA remainsenabled, so that if the SSD thresholds are metagain, GNA is reactivated.

Virtual hotspare

Reserves a minimum amount of space in the node pool thatcan be used for data repair in the event of a drive failure.

To reserve disk space for use as a virtual hot spare, selectone or both of the following options:

l Reduce amount of available space. Subtracts thespace reserved for virtual hot spare when calculatingavailable free space.

l Deny new data writes. Prevents write operations fromusing reserved disk space.

If you configure both the minimum number ofvirtual drives and a minimum percentage of totaldisk space when you configure reserved VHSspace, the enforced minimum value satisfies bothrequirements.

If this setting is enabled and Deny new datawrites is disabled, it is possible for the filesystem utilization to be reported at more than100%.

Storage Pools

SmartPools settings 331

Setting Description Notes- - -

l VHS space to reserve. You can reserve a minimumnumber of virtual drives (1-4), as well as a minimumpercentage of total disk space (0-20%).

Global spillover Specifies how to handle write operations to a node pool thatis not writable.

l Enabled. Redirect write operations from a node poolthat is not writable to another node pool.

l Disabled. Return a disk space error for write operationsto a node pool that is not writable

Spillover datato

Specifies which storage pool to target when a storage poolis not writable.

When spillover is enabled but it is important that

data writes do not fail, select anywhere for the

Spillover data to setting, even if file poolpolicies send data to specific pools.

Protectionmanagement

Uses SmartPools technology to manage storage poolpolicies and requested protection settings.

l SmartPools manages protection settings. Specifythat SmartPools manages the protection settings. Youcan optionally modify the default settings under

Default

l Including files with manually-managed protectionsettings. Overwrite any protection settings that wereconfigured through File System Explorer or thecommand-line interface.

Disabling both protection management and I/Ooptimization management settings disablesSmartPools functionality.

I/Ooptimizationmanagement

Uses SmartPools technology to manage I/O optimization.

SmartPools manages I/O optimization settings.Specify that SmartPools technology is used to manage I/Ooptimization.

Including files with manually-managed protectionsettings. Overwrite any I/O optimization settings that wereconfigured through File System Explorer or the command-line interface

Disabling both protection management and I/Ooptimization management settings disablesSmartPools functionality.

You can modify the default settings in the DefaultI/O Optimization Settings group (optional).

Monitoring storage poolsYou can access information on storage pool health, performance, and status forindividual nodes. Information is available on node-specific network traffic, internal andexternal network interfaces, and drive status.

You can assess pool health and performance by viewing the following information:

u Subpool status

u Node status

u New events

u Cluster size

Storage Pools


u Cluster throughput

u CPU usage

You also can configure real-time and historical performance to be graphed in the webadministration interface.

Monitor storage poolsYou can view the status and details of storage pools.

Note

OneFS shortens storage pool names that are longer than 40 characters. To view the fullstorage pool name, rest the mouse pointer over the shortened name to display a tooltipof the long name.

Procedure


The SmartPools page appears and displays two groupings: the current capacity usageand a list of storage pools.

2. In the Current Capacity Usage area, move the pointer over the usage bar-graphmeasurements to view details.

3. In the Tiers & Node Pools area, expand any tiers to view all storage pool information.

View unhealthy subpoolsOneFS exposes unhealthy subpools in a list.

Procedure


The SmartPools page appears and displays three groupings: the current capacityusage, a list of storage pools, and any unhealthy subpools.

2. In the Unhealthy Subpools area, review details of any problematic subpools.

View file pool job resultsYou can review detailed results from the last time the SmartPools job ran.

Procedure


The SmartPools page appears and displays three groupings: a list of file pool policies,a list of template policies, and the latest scan job results.

2. In the Latest Job Results area, scroll through the job results to see the details for eachfile pool policy.

Storage Pools

Monitor storage pools 333

CHAPTER 19

System jobs


u System jobs overview..........................................................................................336u System jobs library..............................................................................................336u Job operation...................................................................................................... 339u Job performance impact...................................................................................... 340u Job priorities....................................................................................................... 340u Managing system jobs........................................................................................ 341u Managing impact policies................................................................................... 343u Viewing job reports and statistics........................................................................345

System jobs 335

System jobs overviewThe most critical function of OneFS is maintaining the integrity of data on your Isiloncluster. Other important system maintenance functions include monitoring andoptimizing performance, detecting and mitigating drive and node failures, and freeing upavailable space.

Because maintenance functions use system resources and can take hours to run, OneFSperforms them as jobs that run in the background through a service called Job Engine.The time it takes for a job to run can vary significantly depending on a number of factors.These include other system jobs that are running at the same time; other processes thatare taking up CPU and I/O cycles while the job is running; the configuration of yourcluster; the size of your data set; and how long since the last iteration of the job was run.

Up to three jobs can run simultaneously. To ensure that maintenance jobs do not hinderyour productivity or conflict with each other, Job Engine categorizes them, runs them atdifferent priority and impact levels, and can temporarily suspend them (with no loss ofprogress) to enable higher priority jobs and administrator tasks to proceed.

In the case of a power failure, Job Engine uses a checkpoint system to resume jobs asclose as possible to the point at which they were interrupted. The checkpoint systemhelps Job Engine keep track of job phases and tasks that have already been completed.When the cluster is back up and running, Job Engine restarts the job at the beginning ofthe phase or task that was in process when the power failure occurred.

As system administrator, through the Job Engine service, you can monitor, schedule, run,terminate, and apply other controls to system maintenance jobs. The Job Engine providesstatistics and reporting tools that you can use to determine how long different systemjobs take to run in your OneFS environment.

Note

To initiate any Job Engine tasks, you must have the role of SystemAdmin in the OneFSsystem.

System jobs libraryOneFS contains a library of jobs that runs in the background to maintain your Isiloncluster. Some jobs are automatically started by OneFS when particular conditions arise,and some jobs have a default schedule. However, you can run all jobs manually orschedule them according to your workflow.

Job name Description ExclusionSet

ImpactPolicy

Priority Operation

- - - - - -AutoBalance Balances free space in

a cluster, and is mostefficient in clustersthat contain only harddisk drives (HDDs).Run as part ofMultiScan, orautomatically by thesystem if MultiScan isdisabled.

Restripe Low 4 Auto

System jobs



ImpactPolicy

Priority Operation

- - - - - -AutoBalanceLin Balances free space in

a cluster, and is mostefficient in clusterswhen file systemmetadata is stored onsolid state drives(SSDs). Run as part ofMultiScan, orautomatically by thesystem if MultiScan isdisabled.

Restripe Low 4 Auto

AVScan Performs an antivirusscan on all files.

None Low 6 Manual

Collect Reclaims free spacethat previously couldnot be freed becausethe node or drive wasunavailable. Run aspart of MultiScan, orautomatically by thesystem if MultiScan isdisabled.

Mark Low 4 Auto

Dedupe* Scans a directory forredundant data blocksand deduplicates allredundant data storedin the directory.Available only if youactivate aSmartDedupe license.

None Low 4 Manual

DedupeAssessment Scans a directory forredundant data blocksand reports anestimate of theamount of space thatcould be saved bydeduplicating thedirectory.

None Low 6 Manual

DomainMark Associates a path, andthe contents of thatpath, with a domain.

None Low 5 Manual

FlexProtect Scans the file systemafter a device failure toensure that all filesremain protected.FlexProtect is mostefficient in clustersthat contain onlyHDDs.

Restripe Medium 1 Auto

System jobs

System jobs library 337


ImpactPolicy

Priority Operation

- - - - - -FlexProtectLin Scans the file system

after a node failure toensure that all filesremain protected.Most efficient whenfile system metadata isstored on SSDs.

Restripe Medium 1 Auto

FSAnalyze Gathers informationabout the file system.

None Low 1 Scheduled

IntegrityScan Verifies file systemintegrity.

Mark Medium 1 Manual

MediaScan Locates and clearsmedia-level errors fromdisks.

Restripe Low 8 Scheduled

MultiScan Performs the work ofthe AutoBalance andCollect jobssimultaneously.

RestripeMark

Low 4 Auto

PermissionRepair Corrects file anddirectory permissionsin the /ifs directory.

None Low 5 Manual

QuotaScan* Updates quotaaccounting fordomains created on anexisting file tree.Available only if youactivate aSmartQuotas license.

None Low 6 Auto

SetProtectPlus Applies a default filepolicy across thecluster. Runs only if aSmartPools license isnot active.

Restripe Low 6 Manual

ShadowStoreDelete Frees space that isassociated with ashadow store.

None Low 2 Scheduled

SmartPools* Enforces SmartPoolsfile policies. Availableonly if you activate aSmartPools license.

Restripe Low 6 Scheduled

SnapRevert Reverts an entiresnapshot back tohead.

None Low 5 Manual

SnapshotDelete Creates free spaceassociated withdeleted snapshots.

None Medium 2 Auto

System jobs



ImpactPolicy

Priority Operation

- - - - - -TreeDelete Deletes a specified file

path in the /ifsdirectory.

None Medium 4 Manual

* Available only if you activate an additional license

Job operationOneFS includes system maintenance jobs that run to ensure that your Isilon clusterperforms at peak health. Through the Job Engine, OneFS runs a subset of these jobsautomatically, as needed, to ensure file and data integrity, check for and mitigate driveand node failures, and optimize free space. For other jobs, for example, Dedupe, you canuse Job Engine to start them manually or schedule them to run automatically at regularintervals.

The Job Engine runs system maintenance jobs in the background and prevents jobswithin the same classification (exclusion set) from running simultaneously. Two exclusionsets are enforced: restripe and mark.

Restripe job types are:

u AutoBalance

u AutoBalanceLin

u FlexProtect

u FlexProtectLin

u MediaScan

u MultiScan

u SetProtectPlus

u SmartPools

Mark job types are:

u Collect

u IntegrityScan

u MultiScan

Note that MultiScan is a member of both the restripe and mark exclusion sets. Youcannot change the exclusion set parameter for a job type.

The Job Engine is also sensitive to job priority, and can run up to three jobs, of anypriority, simultaneously. Job priority is denoted as 1–10, with 1 being the highest and 10being the lowest. The system uses job priority when a conflict among running or queuedjobs arises. For example, if you manually start a job that has a higher priority than threeother jobs that are already running, Job Engine pauses the lowest-priority active job, runsthe new job, then restarts the older job at the point at which it was paused. Similarly, ifyou start a job within the restripe exclusion set, and another restripe job is alreadyrunning, the system uses priority to determine which job should run (or remain running)and which job should be paused (or remain paused).

Other job parameters determine whether jobs are enabled, their performance impact, andschedule. As system administrator, you can accept the job defaults or adjust theseparameters (except for exclusion set) based on your requirements.

System jobs

Job operation 339

When a job starts, the Job Engine distributes job segments—phases and tasks—acrossthe nodes of your cluster. One node acts as job coordinator and continually works withthe other nodes to load-balance the work. In this way, no one node is overburdened, andsystem resources remain available for other administrator and system I/O activities notoriginated from the Job Engine.

After completing a task, each node reports task status to the job coordinator. The nodeacting as job coordinator saves this task status information to a checkpoint file.Consequently, in the case of a power outage, or when paused, a job can always berestarted from the point at which it was interrupted. This is important because some jobscan take hours to run and can use considerable system resources.

Job performance impactThe Job Engine service monitors system performance to ensure that maintenance jobs donot significantly interfere with regular cluster I/O activity and other system administrationtasks. Job Engine uses impact policies that you can manage to control when a job can runand the system resources that it consumes.

Job Engine has four default impact policies that you can use but not modify. The defaultimpact policies are:

Impact policy Allowed to run Resource consumption- - -LOW Any time of day. Low

MEDIUM Any time of day. Medium

HIGH Any time of day. High

OFF_HOURS Outside of businesshours. Business hours aredefined as 9AM to 5pm,Monday through Friday.OFF_HOURS is pausedduring business hours.

Low

If you want to specify other than a default impact policy for a job, you can create a custompolicy with new settings.

Jobs with a low impact policy have the least impact on available CPU and disk I/Oresources. Jobs with a high impact policy have a significantly higher impact. In all cases,however, the Job Engine uses CPU and disk throttling algorithms to ensure that tasks thatyou initiate manually, and other I/O tasks not related to the Job Engine, receive a higherpriority.

Job prioritiesJob priorities determine which job takes precedence when more than three jobs ofdifferent exclusion sets attempt to run simultaneously. The Job Engine assigns a priorityvalue between 1 and 10 to every job, with 1 being the most important and 10 being theleast important.

The maximum number of jobs that can run simultaneously is three. If a fourth job with ahigher priority is started, either manually or through a system event, the Job Enginepauses one of the lower-priority jobs that is currently running. The Job Engine places the

System jobs


paused job into a priority queue, and automatically resumes the paused job when one ofthe other jobs is completed.

If two jobs of the same priority level are scheduled to run simultaneously, and two otherhigher priority jobs are already running, the job that is placed into the queue first is runfirst.

Managing system jobsThe Job Engine enables you to control periodic system maintenance tasks that ensureOneFS file system stability and integrity. As maintenance jobs run, the Job Engineconstantly monitors and mitigates their impact on the overall performance of the cluster.

As system administrator, you can tailor these jobs to the specific workflow of your Isiloncluster. You can view active jobs and job history, modify job settings, and start, pause,resume, cancel, and update job instances.

View active jobsIf you are noticing slower system response while performing administrative tasks, youcan view jobs that are currently running on your Isilon cluster.

Procedure


2. In the Active Jobs table, view status information about all currently running jobs, jobsettings, and progress details.

a. You can perform bulk actions on the active jobs by selecting the Status check box,then selecting an action from the Select a bulk action drop-down list.

View job historyIf you want to check the last time a critical job ran, you can view recent activity for aspecific job, or for all jobs.

Procedure

1. Click Cluster Management > Job Operations > Job Reports.

The Job Reports table displays a chronological list of the last ten job events that haveoccurred on the cluster. Event information includes the time the event occurred, thejob responsible for the event, and event results.

2. Filter reports by job type by selecting the job from the Filter by Job Type drop-down listand clicking Reset.

3. Click on View Details next to a job name to view recent events for only that job.

Recent events for the job appear in the View Job Report Details window, and includeinformation such as start time, duration, and whether or not the job was successful.

Start a jobBy default, only some system maintenance jobs are scheduled to run automatically.However, you can start any of the jobs manually at any time.

Procedure


2. In the Active Jobs window, select the job you want to start and click More.

System jobs

Managing system jobs 341

3. Click Start Running Job.

Pause a jobYou can pause a job temporarily to free up system resources.

Procedure


2. In the Active Jobs table, click More for the job that you want to pause.

3. Click Pause Running Job in the menu that appears.

The job remains paused until you resume it.

Resume a jobYou can resume a paused job.

Procedure


2. In the Active Jobs table, click More for the job that you want to pause.

3. Click Resume Running Job in the menu that appears.

Results

The job continues from the phase or task at which it was paused.

Cancel a jobIf you want to free up system resources, or for any reason, you can permanentlydiscontinue a running, paused, or waiting job.

Procedure


2. In the Active Jobs table, click More for the job that you want to cancel.

3. Click Cancel Running Job in the menu that appears.

Update a jobYou can change the priority and impact policy of a running, waiting, or paused job.

When you update a job, only the current instance of the job runs with the updatedsettings. The next instance of the job returns to the default settings for that job.

Note

To change job settings permanently, see "Modify job type settings."

Procedure


2. In the Active Jobs table, click View/Edit for the job that you want to update.

3. In the View Active Job Details window, click Edit Job.

a. Select a new priority level from the Priority drop-down list.

b. Select an impact policy level from the Impact Policy drop-down list.

System jobs



When you update a running job, the job automatically resumes. When you update apaused or idle job, the job remains in that state until you restart it.

Modify job type settingsYou can customize system maintenance jobs for your administrative workflow bymodifying the default priority level, impact level, and schedule for a job type.

Procedure


2. In the Job Types table, locate the row for the policy you want to modify and clickView / Edit.

The View Job Type Details window appears, displaying current default settings,schedule, current state, and recent activity.

3. Click Edit Job Type. The Edit Job Type Details window appears.

4. Modify the details you want to change. You can modify the default priority, the defaultimpact policy, whether the job is enabled, and whether the job runs manually or on aschedule.

5. Click Scheduled to modify a job schedule, then select the schedule option from thedrop-down list.


The modifications are saved and applied to all instances of that job type. The resultsare shown in the View Job Type Details window.

7. Click Close.

Managing impact policiesFor system maintenance jobs that run through the Job Engine service, you can create andassign policies that help control how jobs affect system performance.

As system administrator, you can create, copy, modify, and delete impact policies, andview their settings.

Create an impact policyThe Job Engine includes four impact policies, which you cannot modify or delete.However, you can create and configure new impact policies.

Procedure

1. Click Cluster Management > Job Operations > Impact Policies.

2. Click Add an Impact Policy.

The Create Impact Policy window appears.

3. In the Name text field, type a name for the policy. This field is required.

4. In the Description text field, type a comment about the impact policy.

Include information specific to the impact policy such as unique schedule parametersor logistical requirements that make the impact policy necessary.

5. Click Add an Impact Policy Interval.

System jobs

Modify job type settings 343

a. In the Add an Impact Policy Interval window, select the impact level and start andend times from the drop-down lists.

b. Click Add Impact Policy Interval.

The Add an Impact Policy Interval window disappears, and the settings you selectedappear in the Impact Schedule table.

6. Click Create Impact Policy.

Your copy of the impact policy is saved and is listed in alphabetical order in theImpact Policies table.

Copy an impact policyYou can use a default impact policy as the template for a new policy by making andmodifying a copy.

Procedure


2. In the Impact Policies table, locate the row for the policy you want to copy and clickMore. The Copy Impact Policy window appears.

3. In the Name field, type a name for the new policy.

4. In the Description text field, enter a description for the new policy.

Include information specific to the impact policy such as unique schedule parametersor logistical requirements that make the impact policy necessary.

5. Click Add an Impact Policy Interval.

a. In the Add an Impact Policy Interval window, select the impact level and start andend times from the drop-down lists.

b. Click Add Impact Policy Interval.

The Add an Impact Policy Interval window closes, and the settings you selectedappear in the Impact Schedule table.

6. Click Copy Impact Policy.

Your copy of the impact policy is saved and is listed in alphabetical order in theImpact Policies table.

Modify an impact policyYou can change the name, description, and impact intervals of a custom impact policy.

Before you begin

You cannot modify the default impact policies, HIGH, MEDIUM, LOW, and OFF_HOURS. Ifyou want to modify a policy, create and modify a copy of a default policy.

Procedure

1. Navigate to Cluster Management > Job Operations > Impact Policies.

2. In the Impact Policies table, click View / Edit for the policy you want to modify.

The Edit Impact Policy window appears.

3. Click Edit Impact Policy, and modify one or all of the following:

System jobs


Options Description

Policy description a. In the Description field, type a new overview for the impactpolicy.

b. Click Submit.

Impact schedule a. In the Impact Schedule area, modify the schedule of theimpact policy by adding, editing, or deleting impact intervals.

b. Click Save Changes.

The modified impact policy is saved and listed in alphabetical order in the ImpactPolicies table.

Delete an impact policyYou can delete impact policies that you have created.

You cannot delete default impact policies, HIGH, MEDIUM, LOW, and OFF_HOURS.

Procedure


2. In the Impact Policies table, click More next to the custom impact policy that you wantto delete.

3. Click Delete.

A confirmation dialog box appears.


View impact policy settingsYou can view the impact policy settings for any job.

Procedure


The Job Types table is displayed.

2. If necessary, scroll through the Job Types table to find a specific job.

The impact policy settings for the job are shown in the Job Types table.

Viewing job reports and statisticsYou can generate reports for system jobs and view statistics to better determine theamounts of system resources being used.

Most system jobs controlled by the Job Engine run at a low priority and with a low impactpolicy, and generally do not have a noticeable impact on cluster performance.

A few jobs, because of the critical functions they perform, run at a higher priority and witha medium impact policy. These jobs include FlexProtect and FlexProtect Lin, FSAnalyze,SnapshotDelete, and TreeDelete.

As a system administrator, if you are concerned about the impact a system job mighthave on cluster performance, you can view job statistics and reports. These tools enable

System jobs

Delete an impact policy 345

you to view detailed information about job load, including CPU and memory usage andI/O operations.

View statistics for a job in progressYou can view statistics for a job in progress.

Procedure


You can view jobs that are running in the Active Jobs area.

2. Click the View/Edit option to the right of the job entry.

Results

The View Active Jobs Details screen opens, where you can view statistics such asprocessed data, elapsed time, phase, and progress, including an estimate of the timeremaining for the job to complete.

View a report for a completed jobAfter a job finishes, you can view a report about the job.

Before you begin

A report for a job is not available until after the job is completed.

Procedure

1. Click Cluster Management > Job Operations > Job Reports.

The Job Reports screen appears with a list of the last 10 completed jobs.

2. Locate the job whose report you want to view.

If the job is not on the first page of the Job Reports screen, click the right-arrow icon topage through the list until you locate your job.

3. Click View Details.

The View Job Report Details screen appears, listing job statistics such as elapsedtime, CPU and memory usage, and total I/O operations.

4. When you are finished viewing the report, click Close.

System jobs


CHAPTER 20

Networking


u Networking overview........................................................................................... 348u About the internal network.................................................................................. 348u External client network overview......................................................................... 349u Configuring the internal network......................................................................... 353u Configuring an external network.......................................................................... 356u Managing external client connections with SmartConnect................................... 370u Managing network interface provisioning rules....................................................372

Networking 347

Networking overviewAfter you determine the topology of your network, you can set up and manage yourinternal and external networks.

There are two types of networks associated with an EMC Isilon cluster:

InternalNodes communicate with each other using a high speed low latency InfiniBandnetwork. You can optionally configure a second Infiniband network as a failover forredundancy and security.

ExternalClients connect to the cluster through the external network with Ethernet. The Isiloncluster supports standard network communication protocols, including NFS, SMB,HTTP, and FTP. The cluster includes various external Ethernet connections, providingflexibility for a wide variety of network configurations. External network speeds varyby product.

You can manage both the internal and external network settings from the OneFS webadministration interface and the command-line interface.

About the internal networkThe EMC Isilon cluster must connect to at least one high-speed, low-latency Infinibandswitch for internal communications and data transfer. The connection to the Infinibandswitch is also referred to as an internal network. The internal network is separate from theexternal network (typically Ethernet) by which users access the cluster.

Upon initial configuration of your cluster, OneFS creates a default internal network for theInfiniband switch called int-a. A second Infiniband switch can be added for redundancyand failover; the default name is int-b/failover.

Internal IP address rangesThe number of IP addresses assigned to the internal network determines how manynodes can be joined to the EMC Isilon cluster.

When you initially configure the cluster, you specify one or more IP address ranges for theinternal InfiniBand network. This range of addresses is used by the nodes tocommunicate with each other. It is recommended that you create a range of addresseslarge enough to accommodate adding additional nodes to your cluster. If the IP addressrange defined during the initial configuration is too restrictive for the size of the internalnetwork, you can add ranges to the int-a network and int-b network. For certainconfiguration changes, such as deleting an IP address assigned to a node, the clustermust be restarted.

While all clusters will have, at minimum, one internal Infiniband network (int-a), to enablea second internal network (int-b) you must assign another IP address range to it. Toenable internal network failover, assign an IP address range to the failover network. Thisrange is used to refer to the actual IP addresses in use to provide seamless internal IPaddress failover.

Networking


Internal network failoverYou can configure an internal switch as a failover network to provide redundancy for intra-cluster communications.

Enable an internal failover network by connecting the int-a interfaces of each node in thecluster to one switch, connecting the int-b ports on each node to another switch, andthen restarting the EMC Isilon cluster.

In addition to the IP address range assigned to the int-a internal network, if you enablefailover on a second Infiniband switch, you must assign an IP address range that pointsto actual IP addresses used by the cluster. These addresses enable seamless failover inthe event that either the int-a or int-b switches fail.

External client network overviewYou connect a client computer to the EMC Isilon cluster through the external network.OneFS supports network subnets, IP address pools, and features network provisioningrules to simplify configuration.

Subnets simplify external (front-end) network management and provide flexibility inimplementing and maintaining the cluster network. You can create IP address poolswithin subnets to partition your network interfaces according to workflow or node type.You can configure external network settings through provisioning rules and then thoserules are applied to nodes that are added to the cluster.

You must initially configure the default external IP subnet in IPv4 format. Afterconfiguration is complete, you can configure additional subnets using IPv4 or IPv6.

IP address pools can be associated with a node or a group of nodes as well as with theNIC ports on the nodes. For example, based on the network traffic that you expect, youmight decide to establish one subnet for storage nodes and another subnet foraccelerator nodes.

How you set up your external network subnets depends on your network topology. In abasic network topology where all client-node communication occurs through a singlegateway, only a single external subnet is required. If clients connect through multiplesubnets or internal connections, you must configure multiple external network subnets.

External network settingsA default external network subnet is created during the initial set up of your EMC Isiloncluster. You can make modifications to this subnet, create new subnets, and makeadditional configuration changes to the external network.

During initial cluster setup, OneFS performs the following actions:

u Creates a default external network subnet called subnet0, with the specifiednetmask, gateway, and SmartConnect service address.

u Creates a default IP address pool called pool0 with the specified IP address range,the SmartConnect zone name, and the external interface of the first node in thecluster as the only member.

u Creates a default network provisioning rule called rule0, which automatically assignsthe first external interface for all newly added nodes to pool0.

u Adds pool0 to subnet0 and configures pool0 to use the virtual IP of subnet0 as itsSmartConnect service address.

u Sets the global, outbound DNS settings to the domain name server list and DNSsearch list, if provided.

Networking

Internal network failover 349

Once the initial external network has been established, you can configure the followinginformation about your external network:

u Netmask

u IP address range

u Gateway

u Domain name server list (optional)

u DNS search list (optional)

u SmartConnect zone name (optional)

u SmartConnect service address (optional)

You can make modifications to the external network through the web administrationinterface and the command-line interface.

IP address poolsYou can partition EMC Isilon cluster nodes and external network interfaces into logical IPaddress pools. IP address pools are also utilized when configuring SmartConnect zonesand IP failover support for protocols such as NFS. Multiple pools for a single subnet areavailable only if you activate a SmartConnect Advanced license.

IP address pools:

u Map available addresses to configured interfaces.

u Belong to external network subnets.

u Partition network interfaces on your cluster into pools.

u Can be to assigned to groups in your organization.

The IP address pool of a subnet consists of one or more IP address ranges and a set ofcluster interfaces. All IP address ranges in a pool must be unique.

A default IP address pool is configured during the initial cluster setup through thecommand-line configuration wizard. You can modify the default IP address pool at anytime. You can also add, remove, or modify additional IP address pools.

If you add external network subnets to your cluster through the subnet wizard, you mustspecify the IP address pools that belong to the subnet.

IP address pools are allocated to external network interfaces either dynamically orstatically. The static allocation method assigns one IP address per pool interface. The IPaddresses remain assigned, regardless of that interface's status, but the method doesnot guarantee that all IP addresses are assigned. The dynamic allocation methoddistributes all pool IP addresses, and the IP address can be moved depending on theinterface's status and connection policy settings.

Connection balancing with SmartConnectSmartConnect balances client connections to the EMC Isilon cluster.

The SmartConnect module is available in two modes:

BasicIf you have not activated a SmartConnect advanced license, SmartConnect operatesin Basic mode. Basic mode balances client connections by using a round robinpolicy. Basic mode is limited to static IP address allocation and to one IP addresspool per external network subnet. This mode is included with OneFS as a standardfeature.

Networking


AdvancedIf you activate a SmartConnect Advanced license, SmartConnect operates inAdvanced mode. Advanced mode enables client connection balancing based onround robin, CPU utilization, connection counting, or network throughput. Advancedmode supports IP failover and allows IP address pools to support multiple DNSzones within a single subnet.

The following information describes the SmartConnect DNS client-connection balancingpolicies:

Round RobinThis method selects the next available node on a rotating basis. This is the defaultstate (after SmartConnect is activated) if no other policy is selected.

Note

Round robin is the only connection policy available without activating aSmartConnect Advanced license.

Connection CountThis method determines the number of open TCP connections on each availablenode and optimizes the cluster usage.

Network ThroughputThis method determines the average throughput on each available node to optimizethe cluster usage.

CPU UsageThis method determines the average CPU utilization on each available node tooptimize the cluster usage.

Note

SmartConnect requires that you add a new name server (NS) record to the existingauthoritative DNS zone that contains the cluster and that you delegate the SmartConnectzone as a fully qualified domain name (FQDN).

External IP failoverYou can redistribute IP addresses for external IP failover if one or more node interfacesbecomes unavailable.

SmartConnect Basic does not support IP failover. You can enable dynamic IP allocationand IP failover in your EMC Isilon cluster if you activate a SmartConnect Advancedlicense.

Dynamic IP allocation ensures that all IP addresses in the IP address pool are assigned tomember interfaces. Dynamic IP allocation allows clients to connect to any IP addresses inthe pool and receive a response. If a node or an interface becomes unavailable, OneFSmoves the IP address to other member interfaces in the IP address pool.

IP failover ensures that all of the IP addresses in the pool are assigned to an availablenode. When an node interface becomes unavailable, the dynamic IP address of the nodeis redistributed among the remaining available node interfaces. Subsequent clientconnections are directed to the node interface that is assigned to that IP address.

Networking

External IP failover 351

If a SmartConnect Advanced license is active on the cluster, you may have enabled IPfailover when you configured your external network settings. You can also modify yoursubnet settings at any time to enable IP failover for selected IP address pools.

IP failover occurs when a pool has dynamic IP address allocation set. You can furtherconfigure IP failover for your network environment with the following options:

IP allocation methodThis method ensures that all of the IP addresses in the pool are assigned to anavailable node.

Rebalance policyThis policy controls how IP addresses are redistributed when the node interfacemembers for a given IP address pool become available after a period ofunavailability.

IP failover policyThis policy determines how to redistribute the IP addresses among remainingmembers of an IP address pool when one or more members are unavailable.

NIC aggregationNetwork interface card (NIC) aggregation, also known as link aggregation, is optional, andenables you to combine the bandwidth of a node's physical network interface cards intoa single logical connection. NIC aggregation provides improved network throughput.

Note

Configuring link aggregation is an advanced function of network switches. Consult yournetwork switch documentation before configuring your EMC Isilon cluster for linkaggregation.

NIC aggregation can be configured during the creation of a new external network subnet.Alternatively, you can configure NIC aggregation on the existing IP address pool of asubnet.

u OneFS provides support for the following link aggregation methods:

Link Aggregation Control Protocol (LACP)Supports the IEEE 802.3ad Link Aggregation Control Protocol (LACP). Thismethod is recommended for switches that support LACP and is the default modefor new pools.

Legacy Fast EtherChannel (FEC) modeThis method is compatible with aggregated configurations in earlier versions ofOneFS.

Etherchannel (FEC)This method is a newer implementation of the Legacy FEC mode.

Active / Passive FailoverThis method transmits all data through the master port, which is the first port inthe aggregated link. The next active port in an aggregated link takes over if themaster port is unavailable.

Round-RobinThis method balances outbound traffic across all active ports in the aggregatedlink and accepts inbound traffic on any port.

Networking


u Some NICs may allow aggregation of ports only on the same network card.

u For LACP and FEC aggregation modes, the switch must support IEEE 802.3ad linkaggregation. Since the trunks on the network switch must also be configured, thenode must be connected with the correct ports on the switch.

VLANsVirtual LAN (VLAN) tagging is an optional setting that enables an EMC Isilon cluster toparticipate in multiple virtual networks.

You can partition a physical network into multiple broadcast domains, or virtual localarea networks (VLANs). You can enable a cluster to participate in a VLAN which allowsmultiple cluster subnet support without multiple network switches; one physical switchenables multiple virtual subnets.

VLAN tagging inserts an ID into packet headers. The switch refers to the ID to identify fromwhich VLAN the packet originated and to which network interface a packet should besent.

DNS name resolutionYou can designate up to three DNS servers and up to six search domains for your externalnetwork.

You can configure the DNS server settings during initial cluster configuration with thecommand-line Configuration wizard. After the initial configuration, you can modify theDNS server settings through the web administration interface or through the isinetworks command.

IPv6 supportYou can configure dual stack support for IPv6.

With dual-stack support in OneFS, you can configure both IPv4 and IPv6 addresses.However, configuring an EMC Isilon cluster to use IPv6 exclusively is not supported. Whenyou set up the cluster, the initial subnet must consist of IPv4 addresses.

The following table describes important distinctions between IPv4 and IPv6.

IPv4 IPv6- -32-bit addresses 128-bit addresses

Subnet mask Prefix length

Address Resolution Protocol (ARP) Neighbor Discovery Protocol (NDP)

Configuring the internal networkYou can modify the internal network settings of your EMC Isilon cluster.

You can perform the following actions:

u Modify the IP address ranges of the internal network and the int-b/failover network

u Modify the internal network netmask

u Configure and enable an internal failover network

Networking

VLANs 353

u Disable internal network failover

You can configure the int-b/failover network to provide backup in the event of an int-anetwork failure. Configuration involves specifying a valid netmask and IP address rangefor the failover network.

Modify the internal IP address rangeEach internal Infiniband network requires an IP address range. The ranges should have asufficient number of IP addresses for present operating conditions as well as futureexpansion and addition of nodes. You can add, remove, or migrate IP addresses for boththe internal (int-a) and failback (int-b) networks.

Procedure

1. Click Cluster Management > Network Configuration.

2. In the Internal Networks Settings area, select the network that you want to add IPaddresses for.

l To select the int-a network, click int-a.

l To select the int-b/failover network, click int-b/Failover.

3. In the IP Ranges area, you can add, delete, or migrate your IP address ranges. Ideallythe new range is contiguous with the previous one. For example, if your current IPaddress range is 192.168.160.60 - 192.168.160.162, the new range should start with192.168.160.163.

4. Click Submit.

Modify the internal network netmaskYou can modify the netmask value for the internal network.

If the netmask is too restrictive for the size of the internal network, you must modify thenetmask settings. It is recommended that you specify a class C netmask, such as255.255.255.0, for the internal netmask. This netmask is large enough to accommodatefuture clusters.

Note

For the changes in netmask value to take effect, you must reboot the cluster.

Procedure

1. Click Cluster Configuration > Network Configuration.

2. In the Internal Network Settings area, select the network that you want to configurethe netmask for.

l To select the int-a network, click int-a.

l To select the int-b/Failover network, click int-b / Failover.

3. In the Netmask field, type a netmask value.

You cannot modify the netmask value if the change invalidates any node addresses.

4. Click Submit.

A dialog box prompts you to reboot the cluster.

5. Specify when you want to reboot the cluster.

l To immediately reboot the cluster, click Yes. When the cluster finishes rebooting,the login page appears.

Networking


l Click No to return to the Edit Internal Network page without changing the settingsor rebooting the cluster.

Configure and enable an internal failover networkYou can enable an internal failover network on your EMC Isilon cluster. By default, the int-b and internal failover networks are disabled.

Procedure


2. In the Internal Network Settings area, click int-b/Failover.

3. In the IP Ranges area, for the int-b network, click Add range.

4. On the Add IP Range dialog box, enter the IP address at the low end of the range in thefirst IP range field.

5. In the second IP range field, type the IP address at the high end of the range.

Ensure that there is no overlap of IP addresses between the int-a and int-b/failovernetwork ranges. For example, if the IP address range for the int-a network is192.168.1.1 - 192.168.1.100, specify a range of 192.168.2.1 - 192.168.2.100 for theint-b network.

6. Click Submit.

7. In the IP Ranges area for the Failover network, click Add range.

Add an IP address range for the failover network, ensuring there is no overlap with theint-a network or the int-b network.

The Edit Internal Network page appears, and the new IP address range appears in theIP Ranges list.

8. In the Settings area, specify a valid netmask. Ensure that there is no overlap betweenthe IP address range for the int-b network or for the failover network.

It is recommended that you use a class C netmask, such as 255.255.255.0, for theinternal network.

9. In the Settings area, for State, click Enable to enable the int-b and failover networks.

10.Click Submit.

The Confirm Cluster Reboot dialog box appears.

11.Restart the cluster by clicking Yes.

Disable internal network failoverYou can disable the int-b and failover internal networks.

Procedure


2. In the State area, click Disable.

3. Click Submit.

The Confirm Cluster Reboot dialog box appears.

4. Restart the cluster by clicking Yes.

Networking

Configure and enable an internal failover network 355

Configuring an external networkYou can configure all external network connections between the EMC Isilon cluster andclient computers.

Adding a subnetYou can add and configure an external subnet.

Adding a subnet to the external network encompasses these tasks:

Procedure

1. Configuring subnet settings.

2. Adding an IP address to a new subnet.

3. Optional: Configuring SmartConnect settings for a new subnet.

4. Selecting interface members for a new subnet.

Configure subnet settingsYou can add a subnet to the external network of a cluster using the web administrationinterface or the Isilon command line. This procedure describes using the webadministration interface to add a subnet.

Procedure


2. In the External Network Settings area, click Add subnet.

3. In the Basic section, in the Name field, type a unique name for the subnet.

The name can be up to 32 alphanumeric characters long and can include underscoresor hyphens, but not spaces or other punctuation.

4. Optional: In the Description field, type a descriptive comment about the subnet.

The comment can be no more than 128 characters.

5. Specify the IP address format for the subnet and configure an associated netmask orprefix length setting:

l For an IPv4 subnet, click IPv4 in the IP Family list. In the Netmask field, type adotted decimal octet (x.x.x.x) that represents the subnet mask.

l For an IPv6 subnet, click IPv6 in the IP family list. In the Prefix length field, type aninteger (ranging from 1 to 128) that represents the network prefix length.

6. In the MTU list, type or select the size of the maximum transmission units the clusteruses in network communication. Any numerical value is allowed, but must becompatible with your network and the configuration of all devices in the network path.Common settings are 1500 (standard frames) and 9000 (jumbo frames).

Although OneFS supports both 1500 MTU and 9000 MTU, using a larger frame size fornetwork traffic permits more efficient communication on the external networkbetween clients and cluster nodes.

Note

To benefit from using jumbo frames, all devices in the network path must beconfigured to use jumbo frames.

Networking


7. In the Gateway address field, type the IP address of the gateway server device throughwhich the cluster communicates with systems outside of the subnet.

8. In the Gateway priority field, type an integer for the priority of the subnet gateway fornodes assigned to more than one subnet.

You can configure only one default gateway per node, but each subnet can beassigned a gateway. When a node belongs to more than one subnet, this optionenables you to define the preferred default gateway. A value of 1 represents thehighest priority, and 10 represents the lowest priority.

9. If you plan to use SmartConnect for connection balancing, in the SmartConnectservice IP field, type the IP address that will receive all incoming DNS requests foreach IP address pool according to the client connection policy. You must have at leastone subnet configured with a SmartConnect service IP in order to use connectionbalancing.

10.Optional: In the Advancedsection, you can enable VLAN tagging if you want to enablethe cluster to participate in virtual networks.

Note

Configuring a VLAN requires advanced knowledge of network switches. Consult yournetwork switch documentation before configuring your cluster for a VLAN.

11.If you enable VLAN tagging, you must also type a VLAN ID that corresponds to the IDnumber for the VLAN set on the switch, with a value from 2 to 4094.

12.Optional: In the Hardware load balancing field, type the IP address for a hardwareload balancing switch using Direct Server Return (DSR). This routes all client traffic tothe cluster through the switch. The switch determines which node handles the trafficfor the client, and passes the traffic to that node.

13.Click Next.

The Step 2 of 4 -- IP Address Pool Settings dialog box appears.

After you finish

The next step in the process of adding a new subnet is adding an IP address pool.

Add an IP address pool to a new subnetYou can partition the external network interface of your cluster into groups, or pools, ofunique IP address ranges in a subnet.

Before you begin

You must specify basic subnet settings by completing the previous subnet wizard page.

Note

If your cluster is running SmartConnect Basic for connection balancing, you can configureonly one IP address pool per subnet. If you activate a SmartConnect Advanced license,you can configure unlimited IP address pools per subnet.

Procedure

1. In the Step 2 of 4 — IP Address Pool Settings dialog box, type a unique Name for theIP address pool. The name can be up to 32 alphanumeric characters long and caninclude underscores or hyphens, but no spaces or other punctuation.

2. Type a Description for the IP address pool. The description can contain up to 128characters.

Networking

Adding a subnet 357

3. In the Access zone list, click to select an access zone for the pool. OneFS includes adefault system access zone.

4. In the IP range (low-high) area, click New.

OneFS adds an IP address range with default Low IP and High IP values.

5. Click to select the default Low IP value. Replace the default value with the starting IPaddress of the subnet's IP address pool.

6. Click to select the default High IP value. Replace the default value with the ending IPaddress of the subnet's IP address pool.

7. Optional: Add IP address ranges to the IP address pool by repeating steps 3 through 6as needed.

8. Click Next.

The Step 3 of 4 — SmartConnect Settings dialog box appears.

After you finish

The next step in the process of adding a new subnet is configuring SmartConnectsettings, which is optional. If you do not wish to configure SmartConnect settings, thenext step is adding network interface members to the new subnet.

Configure SmartConnect settings for a new subnetYou can configure subnet connection balancing for a cluster's external network with thedefault SmartConnect Basic feature of OneFS. You can configure advanced settings if youactivate a SmartConnect Advanced license.

Before you begin

You must specify basic subnet settings and add at least one IP address pool range to thenew subnet by completing the previous subnet wizard pages.

You can configure SmartConnect as an optional module to balance client connections onthe external network of your cluster. A SmartConnect Advanced license must be active forcertain options. An active SmartConnect Advanced license adds additional advancedbalancing policies to evenly distribute CPU usage, client connections, or throughput. Anactive license also lets you define IP address pools to support multiple DNS zones in asubnet. In addition, SmartConnect supports IP failover, also known as NFS failover. Incontrast, with SmartConnect Basic you can only set a round robin balancing policy.

Note

SmartConnect requires that you add a new name server (NS) record to the existingauthoritative DNS zone that contains the cluster and that you delegate the SmartConnectzone as a fully qualified domain name (FQDN).

Procedure

1. In the Step 3 of 4 — SmartConnect Settings dialog box, type a Zone name for theSmartConnect zone that this IP address pool represents. The zone name must beunique among the pools served by the SmartConnect service subnet specified in Step3 below.

2. In the Connection policy list, select the type of connection balancing policy set by theIP address pool of this subnet. The connection balancing policy determines howSmartConnect distributes incoming DNS requests across the members of an IPaddress pool.

Networking


Options Description

Round Robin Selects the next available node on a rotating basis, and is thedefault policy if no other policy is selected.

Connection Count Determines the number of open TCP connections on eachavailable node to optimize the cluster usage.

Network Throughput Sets the overall average throughput volume on each availablenode to optimize the cluster usage.

CPU Usage Examines average CPU utilization on each available node tooptimize the cluster usage.

3. In the SmartConnect service subnet list, select the name of the external networksubnet whose SmartConnect service will answer DNS requests on behalf of the IPaddress pool. A pool can have only one SmartConnect service answering DNSrequests. If this option is left blank, the IP address pool the subnet belongs to isexcluded when SmartConnect answers incoming DNS requests for the cluster.

Note

If you have activated a SmartConnect Advanced license, complete the following stepsfor the options in the SmartConnect Advanced section of this wizard page.

4. In the IP allocation method list, select the method by which IP addresses are assignedto the member interfaces for this IP address pool:

Options Description

Static Select this IP allocation method to assign IP addresses when memberinterfaces are added to the IP pool. As members are added to the pool,this method allocates the next unused IP address from the pool to eachnew member. After an IP address is allocated, the pool member keeps theaddress indefinitely unless one of the following items is true:

l The member interface is removed from the network pool.

l The member node is removed from the cluster.

l The member interface is moved to another IP address pool.

Dynamic Select this IP allocation method to ensure that all IP addresses in the IPaddress pool are assigned to member interfaces, which allows clients toconnect to any IP addresses in the pool and be guaranteed a response. Ifa node or an interface becomes unavailable, their IP addresses areautomatically moved to other available member interfaces in the pool.

If you select the dynamic IP allocation method, you can specify the SmartConnectRebalance policy and the IP failover policy in the next two steps.

5. Select the type of SmartConnect Rebalance policy to redistribute IP addresses. IPaddress redistribution occurs when node interface members in an IP address poolbecome available. These options can only be selected if the IP allocation method isset to Dynamic.

Networking

Adding a subnet 359

Options Description

AutomaticFailback (default)

Automatically redistributes IP addresses. The automaticrebalance is triggered by a change to one of the following items.

l Cluster membership.

l Cluster external network configuration.

l A member network interface.

Manual Failback Does not redistribute IP addresses until you manually issue arebalance command through the command-line interface.

6. The IP failover policy—also known as NFS failover—determines how to redistribute theIP addresses among remaining members of an IP address pool when one or moremembers are unavailable. In order to enable IP failover, set the IP allocation methodto Dynamic, and then select an IP failover policy:

Options Description





7. Click Next to store the changes that you made to this wizard page.

The Step 4 of 4 — IP Address Pool members dialog box appears.

After you finish

The next step in the process of adding a new subnet is adding network interfacemembers.

Select interface members for a new subnetYou can select which network interfaces are in the IP address pool that belongs to theexternal network subnet.

Before you begin

You must specify basic subnet settings and add at least one IP address pool range to thenew subnet by completing previous subnet wizard pages.

Procedure

1. In the Step 4 of 4 — IP Address Pool Members dialog box, select which Availableinterfaces on which nodes you want to assign to the current IP address pool, and thenclick the right arrow button to move them to the Interfaces in current pool.

Alternatively, drag and drop the selected interfaces between the Available interfacestable and the Interfaces in current pool table.

Selecting an available interface for a node that has a Type designated Aggregationbonds together the external interfaces for the selected node.

Networking


In the case of aggregated links, choose the aggregation mode that corresponds to theswitch settings from the Aggregation mode drop-down.

Note

Configuring link aggregation requires advanced knowledge of how to configurenetwork switches. Consult your network switch documentation before configuringyour cluster for link aggregation.

2. When you have finished assigning external network interfaces to the IP address pool,click Submit.

The external subnet settings you configured by using the Subnet wizard appear on theEdit Subnet page.

Managing external network subnetsYou can configure subnets on an external network to manage connections between theEMC Isilon cluster and client computers.

Modify external subnet settingsYou can modify the subnet for the external network.

Note

Modifying an external network subnet that is in use can disable access to the cluster andthe web administration interface. OneFS displays a warning if deleting a subnet willterminate communication between the cluster and the web administration interface.

Procedure


2. In the External Network Settings area, click the name of the subnet you want tomodify.

3. In the Settings area, click Edit.

4. Modify the Basic subnet settings as needed.

Options Description

Description A descriptive comment that can be up to 128 characters.

Netmask The subnet mask for the network interface. This field appearsonly for IPv4 subnets.

MTU The maximum size of the transmission units the cluster uses innetwork communication. Any numerical value is allowed, butmight not be compatible with your network. Common settingsare 1500 (standard frames) and 9000 (jumbo frames).

Gateway address The IP address of the gateway server through which the clustercommunicates with systems outside of the subnet.

Gateway priority The priority of the subnet's gateway for nodes that are assignedto more than one subnet. Only one default gateway can beconfigured on each Isilon node, but each subnet can have itsown gateway. If a node belongs to more than one subnet, thisoption enables you to define the preferred default gateway. A

Networking

Managing external network subnets 361

Options Description

value of 1 is the highest priority, with 10 being the lowestpriority.

SmartConnectservice IP

The IP address that receives incoming DNS requests fromoutside the cluster. SmartConnect responds to these DNSrequests for each IP address pool according to the pool's client-connection policy. To use connection balance, at least onesubnet must be configured with a SmartConnect service IPaddress.

5. Optional: Modify the Advanced settings as needed.

Note

Configuring a virtual LAN requires advanced knowledge of network switches. Consultyour network switch documentation before configuring your cluster for a VLAN. If youare not using a virtual LAN, leave the VLAN options disabled.

Options Description

VLAN tagging You can enable VLAN tagging. VLAN tagging allows a cluster toparticipate in multiple virtual networks. VLAN support providessecurity across subnets that is otherwise available only bypurchasing additional network switches.

VLAN ID If you enabled VLAN tagging, type a VLAN ID that corresponds tothe ID number for the VLAN that is set on the switch, with a valuefrom 1 to 4094.

Hardware loadbalancing IPs

You can enter the IP address for a hardware load balancing switchthat uses Direct Server Return (DSR).

6. Click Submit.

Remove an external subnetYou can delete an external network subnet that you no longer need.

Deleting an external network subnet that is in use can prevent access to the cluster andthe web administration interface. OneFS displays a warning if deleting a subnet willterminate communication between the cluster and the web administration interface.

Procedure


2. In the External Network Settings area, click the name of the subnet you want todelete.

The Edit Subnet page appears for the subnet you specified.

3. Click Delete subnet.

A confirmation dialogue box appears.

4. Click Yes to delete the subnet.

If the subnet you are deleting is used to communicate with the web administrationinterface, the confirmation message will contain an additional warning.

Networking


Create a static routeYou can create a static route to connect to networks that are unavailable through thedefault routes.

You configure a static route on a per-pool basis. A static route can be configured only withthe command-line interface and only with the IPv4 protocol.

Procedure


2. Create a static route by running the following command: isi networks modifypool --name <subnetname>:<poolname> --add-static-routes <subnet>/<netmask>-<gateway>


Modifying pool 'subnet0:pool0': Saving: OK OK

3. To verify that the static route was created, run the following command: isi networksls pools -v.

Remove a static routeYou can remove static routes.

Procedure


2. Remove a static route by running the following command: isi networks modifypool --name <subnetname>:<poolname> --remove-static-routes <subnet>/<netmask>-<gateway>


Modifying pool 'subnet0:pool0':

Saving: OK

3. To ensure that the static route was created, run the following command: isinetworks ls pools -v.

Enable or disable VLAN taggingYou can configure a cluster to participate in multiple virtual private networks, also knownas virtual LANs or VLANs. You can also configure a VLAN when creating a subnet using theSubnet wizard.

Procedure


2. In the External Network Settings area, click the name of the subnet that contains theIP address pool that you want to add interface members to.

3. In the Settings area for the subnet, click Edit.

4. In the VLAN tagging list, select Enabled or Disabled.

If you select Enabled, proceed to the next step. If you select Disabled, proceed to Step6.

5. In the VLAN ID field, type a number between 2 and 4094 that corresponds to the VLANID number set on the switch.

Networking

Managing external network subnets 363

6. Click Submit.

Managing IP address poolsIP address pools allow you to manage the IP addresses clients use to connect to an EMCIsilon cluster. You can also add network interfaces to IP address pools. Each IP addresspool is associated with a subnet.

Add an IP address poolYou can add an IP address pool to a external network subnet.

Procedure


2. Click the name of the subnet to which you are adding an IP address pool.

3. In the IP Address Pools area, click Add pool.

4. Specify basic settings for the new IP address pool and click Next.

5. Optional: Specify SmartConnect settings and click Next.

6. Optional: Add network interface members and click Submit.

Modify an IP address poolYou can use the web interface to modify IP address pool settings.

Procedure


2. Click the name of the subnet containing the pool you want to modify.

3. In the Basic Settings area, click Edit for the IP address pool you want to modify.

4. Modify the address pool settings and click Submit.

Delete an IP address poolYou can use the web interface to delete IP address pool settings.

Procedure


2. Click the name of the subnet containing the pool you want to delete.

3. Click Delete pool by the pool you want to delete.

Modify a SmartConnect zoneYou can modify the settings of a SmartConnect zone that you created for an externalnetwork subnet using the Subnet wizard.

Procedure


2. Click the name of the external network subnet that contains the SmartConnect zoneyou want to modify.

3. In the SmartConnect settings area of the pool containing the SmartConnect settingsyou want to modify, click Edit.

4. Modify the Zone name, and then click Submit.

The SmartConnect zone should be a fully-qualified domain name (FQDN).

Networking


Disable a SmartConnect zoneYou can remove a SmartConnect zone from an external network subnet.

Procedure


2. Click the name of the external network subnet that contains the SmartConnect zoneyou want to disable.

3. In the SmartConnect settings area for the pool containing the SmartConnect zone youwant to delete, click Edit.

4. To disable the SmartConnect zone, delete the name of the SmartConnect zone fromthe Zone name field and leave the field blank.

5. Click Submit to disable the SmartConnect zone.

Configure IP failoverYou can configure IP failover to reassign an IP address from an unavailable node to afunctional node, which enables clients to continue communicating with the cluster, evenafter a node becomes unavailable.

Procedure

1. Click Cluster Management > Network Configuration

2. In the External Network Settings area, click the name of the subnet for which youwant to set up IP failover.

3. Expand the area of the pool you want to modify and click Edit in the SmartConnectSettings area.

4. Optional: In the Zone name field, enter a name for the zone, using no more than 128characters.

5. In the Connection Policy list, select a balancing policy:

Options Description

Round Robin Selects the next available node on a rotating basis, and is thedefault state if no other policy is selected.


Network Throughput Uses the overall average throughput volume on each availablenode to optimize the cluster usage.


6. If you purchased a license for SmartConnect Advanced, you will also have access tothe following lists:

IP allocation methodThis setting determines how IP addresses are assigned to clients. Select eitherDynamic or Static.

Rebalance PolicyThis setting defines the client redirection policy for when a node becomesunavailable. The IP allocation list must be set to Dynamic in order for rebalancepolicy options to be selected.

Networking

Managing IP address pools 365

IP failover policyThis setting defines the client redirection policy when an IP address becomesunavailable.

Allocate IP addresses to accommodate new nodesYou can expand capacity by adding new nodes to your Isilon cluster.

After the hardware installation is complete, you can allocate IP addresses for a new nodeon one of the cluster's existing external network subnets, and then add the node'sexternal interfaces to the subnet's IP address pool.

You can also use network provisioning rules to automate the process of configuring theexternal network interfaces for new nodes when they are added to a cluster, although youmay still need to allocate more IP addresses for the new nodes, depending on how manyare already configured.

Procedure


2. In the External Network Settings area, click the name of the subnet that contains theIP address pool that you want to allocate more IP addresses to in order toaccommodate the new nodes.

3. In the Basic settings area, click Edit.

4. Click New to add a new IP address range using the Low IP and High IP fields. or clickthe respective value in either the Low IP or High IP columns and type a new beginningor ending IP address.

5. Click Submit.

6. In the Pool members area, click Edit.

7. In the Available Interfaces table, select one or more interfaces for the newly addednode, and then click the right arrow button to move the interfaces into the Interfacesin current pool table.

8. Click Submit to assign the new node interfaces to the IP address pool.

Managing network interface membersYou can assign nodes and network interfaces to specific IP address pools.

You can also aggregate network interfaces and specify the aggregation method.

Modify the interface members of a subnetYou can use the web interface to modify interface member settings.

Procedure


2. In the External Network Settings area, click the subnet containing the interfacemembers you want to modify.

3. Click Edit next to the Pool members area.

Networking


Remove interface members from an IP address poolYou can use the web interface to remove interface members from an IP address pool.

Procedure


2. In the External Network Settings area, click the name of the subnet containing the IPaddress pool that you want to remove the interface member(s) for.


4. Remove a node's interface from the IP address pool by clicking the node's name in theInterfaces in current pool column, and then click the left arrow button. You can alsodrag and drop node interfaces between the Available Interfaces list and the Interfacesin current poolcolumn.

5. When you have finished removing node interfaces from the IP address pool, clickSubmit.

Configure NIC aggregationYou can configure cluster IP address pools to use NIC aggregation.

Before you begin

You must enable NIC aggregation on the cluster before you can enable NIC aggregationon the switch. If the cluster is configured but the switch is not configured, then the clustercan continue to communicate. If the switch is configured, but the cluster is notconfigured, the cluster cannot communicate, and you are unable to configure the clusterfor NIC aggregation.

This procedure describes how to configure network interface card (NIC) aggregation for anIP address pool belonging to an existing subnet. You can also configure NIC aggregationwhile configuring an external network subnet using the Subnet wizard.

Configuring NIC aggregation means that multiple, physical external network interfaces ona node are combined into a single logical interface. If a node has two external GigabitEthernet interfaces, both will be aggregated. On a node with both Gigabit and 10 GigabitEthernet interfaces, both types of interfaces can be aggregated, but only with interfacesof the same type. NIC aggregation cannot be used with mixed interface types.

An external interface for a node cannot be used by an IP address pool in both anaggregated configuration and an individual interface. You must remove the individualinterface for a node from the Interfaces in current pool table before configuring anaggregated NIC. Otherwise, the web administration interface displays an error messagewhen you click Submit.

Note

Configuring link aggregation requires advanced knowledge of network switches. Consultyour network switch documentation before configuring your cluster for NIC aggregation.

Procedure


2. In the External Network Settings area, click the name of the subnet that contains theIP address pool that you want to add aggregated interface members to.


In the case of multiple IP address pools, expand the pool that you want to add theaggregated interfaces to, and then click Edit in the Pool members area.

Networking

Managing network interface members 367

4. In the Available interfaces table, click the aggregated interface for the node, which isindicated by a listing of AGGREGATION in the Type column.

For example, if you want to aggregate the network interface card for Node 2 of thecluster, click the interface named ext-agg, Node 2 under Available interfaces, andthen click the right-arrow button to move the aggregated interface to the Interfaces incurrent pool table.

5. From the Aggregation mode drop-down, select the appropriate aggregation mode thatcorresponds to the network switch settings.

Note

Consult your network switch documentation for supported NIC aggregation modes.

OneFS supports the following NIC aggregation modes:

Options Description

Link AggregationControl Protocol(LACP)

Supports the IEEE 802.3ad Link Aggregation ControlProtocol (LACP). This method is recommended for switchesthat support LACP and is the default mode for new pools.

Legacy FastEtherChannel (FEC)mode

This method is compatible with aggregated configurationsin earlier versions of OneFS.

Etherchannel (FEC) This method is the newer implementation of the Legacy FECmode.

Active / PassiveFailover

This method transmits all data transmits through the masterport, which is the first port in the aggregated link. The nextactive port in an aggregated link takes over if the masterport is unavailable.

Round-Robin Tx This method balances outbound traffic across all activeports in the aggregated link and accepts inbound traffic onany port.

6. Click Submit.

NIC and LNI aggregation optionsNetwork interface card (NIC) and logical network interface (LNI) mapping options can beconfigured for aggregation.

The following list provides guidelines for interpreting the aggregation options.

u Nodes support multiple network card configurations.

u LNI numbering corresponds to the physical positioning of the NIC ports as found onthe back of the node. LNI mappings are numbered from left to right.

u Aggregated LNIs are listed in the order in which they are aggregated at the time theyare created.

u NIC names correspond to the network interface name as shown in command-lineinterface tools such as ifconfig and netstat.

Networking


LNI NIC Aggregated LNI Aggregated NIC Aggregated NIC(Legacy FEC mode

- - - - -ext-1

ext-2

em0

em1

ext-agg = ext-1 + ext-2 lagg0 fec0

ext-1

ext-2

ext-3

ext-4

em2

em3

em0

em1

ext-agg = ext-1 + ext-2

ext-agg-2 = ext-3 + ext-4

ext-agg-3 = ext-3 + ext-4 + ext-1 +ext-2

lagg0

lagg1

lagg2

fec0

fec1

fec2

ext-1

ext-2

10gige-1

10gige-1

em0

em1

cxgb0

cxgb1

ext-agg = ext-1 + ext-2

10gige-agg-1 = 10gige-1 + 10gige-2

lagg0

lagg1

fec0

fec1

Remove an aggregated NIC from an IP address pool

You can remove an aggregated NIC configuration from an IP address pool if your networkenvironment has changed. However, you must first replace the aggregated setting withsingle-NIC settings in order for the node to continue supporting network traffic.

Procedure


2. In the External Network Settings area, click the name of the subnet that contains theIP address pol with the NIC aggregation settings you want to remove.


4. Select the name of the aggregated NIC for the node that you want to remove in theInterfaces in current pool table, and then click the left arrow button to move the nameinto the Available interfaces table.

5. Select one or more individual interfaces for the node in the Available interfaces table,and then click the right arrow button to move the interfaces into the Interfaces incurrent pool table.

6. When you have completed modifying the node interface settings, click Submit.

Move nodes between IP address poolsYou can move nodes between IP address pools in the event of a network reconfigurationor installation of a new network switch.

The process of moving nodes between IP address pools involves creating a new IPaddress pool and then assigning it to the nodes so that they are temporarily servicingmultiple subnets. After testing that the new IP address pool is working correctly, the oldIP address pool can safely be deleted.

Procedure

1. Create a new IP address pool with the interfaces belonging to the nodes you want tomove.

2. Verify that the new IP address pool functions properly by connecting to the nodes youwant to move with IP addresses from the new pool.

Networking

Managing network interface members 369

3. Delete the old IP address pool.

Reassign a node to another external subnetYou can move a node interface to a different subnet.

Nodes can be reassigned to other subnets.

Procedure


2. In the External Settings area, click the subnet containing the node that you want tomodify.

3. In the IP Address Pools area, click Edit next to the Pool members area.

4. Reassign the interface members that you want to move by dragging and droppingthem from one column to other, or by clicking on an interface member and using theleft arrow and right arrow buttons.

Configure DNS settingsYou can configure the domain name servers (DNS) and DNS search list to resolve hostnames for the EMC Isilon cluster.

Procedure

1. Click Cluster Management > Networking Configuration.

2. In the DNS Settingsarea, click Edit.

3. In the Domain name server(s) field, enter up to three domain name server IPaddresses. You can specify domain name server addresses in IPv4 or IPv6 format.

4. In the DNS search list field, enter up to six DNS search suffixes. DNS search suffixesare appended to unqualified host names .

5. Optional: In the DNS resolver options field, enter advanced DNS configurationvariables.

Note

Setting DNS resolver options may change how OneFS performs DNS lookups. Do notset DNS resolver options unless directed to do so by Isilon Technical Support.

6. Click Submit.

Managing external client connections with SmartConnectYou can manage settings that determine how IP addresses are allocated to clientconnection requests.

With the Basic SmartConnect module, you can specify connection balancing and IPallocation policies. If you have activated a Advanced SmartConnect license, you can alsospecify settings that manage IP failover and rebalancing.

Configure client connection balancingYou can configure connection balancing for your cluster's external network connectionswith SmartConnect.

Before you begin

You must first enable SmartConnect by setting up a SmartConnect service address on theexternal network subnet that answers incoming DNS requests.

Networking


You might have already configured SmartConnect while setting up an external networksubnet using the Subnet wizard. However, you can configure or modify connectionbalancing settings at any time as your networking requirements change.

Procedure


2. In the External Network Settings area, click the link for the subnet that you want toconfigure for connection balancing.

3. In the Settings area, verify that the SmartConnect service IP was configured.

If the SmartConnect service IP field reads Not set, click Edit, and then specify the IPaddress that DNS requests are directed to.

4. In the SmartConnect settings area , click Edit.

5. In the Zone name field, type a name for the SmartConnect zone that this IP addresspool represents. The zone name must be unique among the pools served by theSmartConnect service subnet that is specified in Step 7 below.

6. In the Connection policy drop-down list, select the type of connection balancingpolicy that is configured for the IP address pool for this zone. The policy determineshow SmartConnect distributes incoming DNS requests across the members of an IPaddress pool.

Note

Round robin is the only connection policy available if you have not activated aSmartConnect Advanced license.

Options Description





7. In the SmartConnect service subnet list, select the name of the external networksubnet whose SmartConnect service answers DNS requests on behalf of the IPaddress pool. A pool can have only one SmartConnect service answering DNSrequests. If this option is left blank, the IP address pool that the SmartConnect servicebelongs to is excluded when SmartConnect answers incoming DNS requests for thecluster.

If you have activated a SmartConnect Advanced license, complete the following stepsin the SmartConnect Advanced area.

8. In the IP allocation method list, select the method by which IP addresses are assignedto the member interfaces for this IP address pool.

Networking

Configure client connection balancing 371

Managing network interface provisioning rulesYou can configure provisioning rules to automate the configuration of external networkinterfaces.

Provisioning rules specify how new nodes are configured when they are added to an EMCIsilon cluster

If the new node type matches the type defined in a rule, the new node's interface name isadded to the subnet and the IP address pool specified in the rule.

For example, you can create a provisioning rule that configures new Isilon storage nodes,and another rule that configures new accelerator nodes.

OneFS automatically checks for multiple provisioning rules when new rules are added toensure there are no conflicts.

Create a node provisioning ruleConfigure one or more provisioning rules to automate the process of adding new nodes toyour Isilon cluster. All Isilon nodes support provisioning rules.

Before you begin

External network subnets and IP address pools must be configured before creating nodeprovisioning rules. You must also verify that the IP address pool included in theprovisioning rule has sufficient IP addresses to accommodate the new node's clientconnections.

Procedure


2. In the Provisioning Rules area, click Add rule.

3. In the Name field, type a unique name for the provisioning rule. The rule name can bea maximum of 32 characters and can include spaces or other punctuation.

4. Optional: In the Description field, type a descriptive comment about the provisioningrule.

5. In the If node type is list, select the type of node to which you want to apply the rule:

Options Description

Any Apply the provisioning rule to all types of Isilon nodes that jointhe cluster.

Storage-i Apply the provisioning rule only to Isilon i-Series storage nodesthat join the cluster.

Accelerator-i Apply the provisioning rule only to Isilon i-Series performanceaccelerator nodes that join the cluster.

Storage Apply the provisioning rule only to Isilon storage nodes thatjoin the cluster.

Accelerator Apply the provisioning rule only to performance-acceleratornodes that join the cluster.

Networking


Options Description

Backup-Accelerator Apply the provisioning rule only to Isilon backup-acceleratornodes that join the cluster.

6. In the then assign interface list, assign one of the following interfaces to the externalnetwork subnet and IP address pool for the node specified in the rule:

Options Description

ext-1 The first external Gigabit Ethernet interface on the cluster.

ext-2 The second external Gigabit Ethernet interface on the cluster.

ext-3 The third external Gigabit Ethernet interface on the cluster.

ext-4 The fourth external Gigabit Ethernet interface on the cluster.

ext-agg The first and second external Gigabit Ethernet interfaces aggregatedtogether.

ext-agg-2 The third and fourth external Gigabit Ethernet interfaces aggregatedtogether.

ext-agg-3 The first four external Gigabit Ethernet interfaces aggregatedtogether.

ext-agg-4 All six Gigabit Ethernet interfaces aggregated together.

10gige-1 The first external 10 Gigabit Ethernet interface on the cluster.

10gige-2 The second external 10 Gigabit Ethernet interface on the cluster.

10gige-agg-1 The first and second external 10 Gigabit Ethernet interfacesaggregated together.

7. In the Subnet list, select the external subnet that the new node will join.

8. In the Pool list, select the IP address pool of the subnet that should be used by thenew node.

9. Click Submit.

Modify a node provisioning ruleYou can modify node provisioning rules.

Procedure

1. Click Cluster Configuration > Network Configuration.

2. In the Provisioning Rules area, click the name of the rule you want to modify.

3. Modify the provisioning rule settings as needed.

4. Click Submit.

Networking

Modify a node provisioning rule 373

Delete a node provisioning ruleYou can delete a provisioning rule that is no longer necessary.

Procedure


2. In the Provisioning Rules area, click Delete next to the rule you want to delete.

A confirmation dialog box appears.

3. Click Yes to delete the rule, or click No to keep the rule.

Networking


CHAPTER 21

Hadoop


u Hadoop overview................................................................................................ 376u OneFS Hadoop support....................................................................................... 376u Hadoop cluster integration..................................................................................377u Managing HDFS...................................................................................................377u Securing HDFS connections through Kerberos..................................................... 379

Hadoop 375

Hadoop overviewHadoop is a flexible, open-source framework for large-scale distributed computation.

The OneFS file system can be configured for native support of the Hadoop Distributed FileSystem (HDFS) protocol, enabling your cluster to participate in a Hadoop system.

HDFS integration requires you to activate a separate license. To obtain additionalinformation or to enable HDFS support for your EMC Isilon cluster, contact your EMC Isilonsales representative.

OneFS Hadoop supportAn HDFS implementation adds HDFS to the list of protocols that can be used to accessthe OneFS file system. Implementing HDFS on an Isilon cluster does not create a separateHDFS file system. The cluster can continue to be accessed through NFS, SMB, FTP, andHTTP.

The HDFS implementation from Isilon is a lightweight protocol layer between the OneFSfile system and HDFS clients. Unlike with a traditional HDFS implementation, files arestored in the standard POSIX-compatible file system on an Isilon cluster. This means filescan be accessed by the standard protocols that OneFS supports, such as NFS, SMB, FTP,and HTTP as well as HDFS.

Files that will be processed by Hadoop can be loaded by using standard Hadoopmethods, such as hadoop fs -put, or they can be copied by using an NFS or SMBmount and accessed by HDFS as though they were loaded by Hadoop methods. Also, filesloaded by Hadoop methods can be read with an NFS or SMB mount.

The supported versions of Hadoop are as follows:

u Apache Hadoop 0.20.203.0

u Apache Hadoop 0.20.205

u Apache Hadoop 1.0.x

u Apache Hadoop 1.2.1

u Apache Hadoop 2.0.x

u CDH 3

u CDH 4

u CDH 4.2

u Greenplum HD 1.1

u Greenplum HD 1.2

u Pivotal HD 1.0.1

u HAWQ 1.1.0.1

u Hortonworks Data Platform 1.3

Hadoop


Hadoop cluster integrationTo enable native HDFS support in OneFS, you must integrate the Isilon cluster with acluster of Hadoop compute nodes.

This process requires configuration of the Isilon cluster as well as each Hadoop computenode that needs access to the cluster.

Managing HDFSTo keep the HDFS service performing efficiently on an Isilon cluster, you will need to befamiliar with the user and system configuration options available as part of an HDFSimplementation.

You can manage an HDFS implementation by using the following methods:

u Hadoop client machines are configured directly through their Hadoop installationdirectory.

u A secure shell (SSH) connection to a node in the Isilon cluster is used to configure theHDFS service.

Configure the HDFS protocolYou can specify which HDFS distribution to use, and you can set the logging level, theroot path, the Hadoop block size, and the number of available worker threads.

By default, subnet0:pool0 binds to the /default-rack. As a result, a Hadoop client connectsover HDFS to the datanodes with interfaces that are assigned to the pool. You configureHDFS by running the isi hdfs command in the OneFS command-line interface.

Procedure

1. Open a secure shell (SSH) connection to any node in the cluster and log in by usingthe root account.

Note

You can combine multiple options with a single isi hdfs command. For command

usage and syntax, run the isi hdfs -h command.

2. To set the default logging level for the Hadoop daemon across the cluster, run the isihdfs command with the --log-level option.

Valid values are listed below, in descending order from the highest to the lowestlogging level. The default value is NOTICE. The values are case-sensitive.

l EMERG: A panic condition. This is normally broadcast to all users.

l ALERT: A condition that should be corrected immediately, such as a corruptedsystem database.

l CRIT: Critical conditions, such as hard device errors.

l ERR: Errors.

l WARNING: Warning messages.

l NOTICE: Conditions that are not error conditions, but may need special handling.

l INFO: Informational messages.

Hadoop

Hadoop cluster integration 377

l DEBUG: Messages that contain information typically of use only when debugging aprogram.

For example, the following command sets the log level to WARNING:isi hdfs --log-level=WARNING

3. To set the path on the cluster to present as the HDFS root directory, run the isihdfs command with the --root-path option.

Valid values include any directory path beginning at /ifs, which is the default HDFSroot directory.For example, the following command sets the root path to /ifs/hadoop:isi hdfs --root-path=/ifs/hadoop

4. To set the Hadoop block size, run the isi hdfs command with the --block-sizeoption.

Valid values are 4KB to 1GB. The default value is 64MB.For example, the following command sets the block size to 32 MB:isi hdfs --block-size=32MB

5. To tune the number of worker threads that HDFS uses, run the isi hdfs commandwith the --num-threads option.

Valid values are 1 to 256 or auto, which is calculated as twice the number of cores.The default value is auto.For example, the following command specifies 8 worker threads:

isi hdfs --num-threads=86. To allocate IP addresses from an IP address pool, run the isi hdfs racks

modify command.

The following command allocates IP addresses from a pool named pool1, which is inthe subnet0 subnet.isi hdfs racks modify --name=/default-rack --add-ip-pool=subnet0:pool1

Note

All pools are assigned to the /default-rack by default.

After you finish

To access files on OneFS by using the HDFS protocol, you must first create a local Hadoopuser that maps to a user on a Hadoop client. The user can be any local user.

Create a local userTo access files on OneFS by using the HDFS protocol, you must first create a local Hadoopuser that maps to a user on a Hadoop client.

Procedure

1. Open a secure shell (SSH) connection to any node in the cluster and log in by usingthe root user account.

2. At the command prompt, run the isi auth users create command to create alocal user.

For example, isi auth users create --name="user1".

Hadoop


Enable or disable the HDFS serviceThe HDFS service, which is enabled by default after you activate an HDFS license, can beenabled or disabled by running the isi services command.

Procedure

1. Open a secure shell (SSH) connection to any node in the cluster and log in by usingthe root user account.

2. At the command prompt, run the isi service command to enable or disable theHDFS service, isi_hdfs_d.

l To enable the HDFS service, run the following command:

isi services isi_hdfs_d enablel To disable the HDFS service, run the following command:

isi services isi_hdfs_d disable

Securing HDFS connections through KerberosYou can secure HDFS connections on a cluster with either MIT Kerberos or MicrosoftActive Directory with Kerberos.

Configuring HDFS authentication with MIT KerberosYou can configure HDFS to authenticate through MIT Kerberos on an EMC Isilon cluster.

Before you begin, confirm the following checklist items:

u The cluster is running properly with HDFS in simple security authentication mode.

u A SmartConnect zone is configured on the Isilon cluster and hostname resolution iscorrect .

u System clocks on the Isilon nodes and the Hadoop clients are synchronized with aformal time source such as Active Directory or NTP.

u You know how to configure Linux and Unix systems to work with Kerberos.

u You know how to manage an EMC Isilon cluster through the command-line interface.

You must perform the following tasks in sequence to configure HDFS authenticationthrough Kerberos.

1. Configure the realm, hostnames, and domain for HDFS.

2. Configure OneFS for Kerberos.

3. Configure the krb5.conf file.

4. Configure cluster settings for HDFS.

5. Modify Hadoop configuration files for Kerberos authentication.

For more information about authenticating HDFS connections through Kerberos, see thewhite paper EMC Isilon Best Practices for Hadoop Data Storage.

Hadoop

Enable or disable the HDFS service 379

Configure the realm, hostnames, and domain for HDFSYou must add your realm, set the host names of the KDC and the KDC admin server, andspecify the default realm as part of setting up Kerberos authentication over HDFS.

Before you begin

These instructions assume the following prerequisites:

u You have already configured a Kerberos system with a resolvable hostname for theKDC and a resolvable hostname for the KDC admin server.

u You know how to configure client computers to work with Kerberos authentication.

u You have established an SSH connection to a node in the cluster and are logged in asthe root user.

Realm names must be uppercase.

Procedure

1. Run the following command to add the realm and set the KDC and KDC admin server,where <REALM>, <kdc-hostname>, and <kdc-admin-server-hostname> are placeholders foryour realm, KDC, and KDC admin server, respectively:

isi auth krb5 add realm --realm=<REALM> \ --kdc=<kdc-hostname> \ --admin-server=<kdc-admin-server-hostname>

2. Then run the following command to set the default realm, where <REALM> is the namethat you specified for the realm name:

isi auth krb5 modify default --default-realm=<REALM>3. Next, write a krb5.conf file by running the following command:

isi auth krb5 write4. Run the following command to verify that you added the realm correctly:

kadmin -p <admin-principal-name>kadmin: quit

Configure OneFS for KerberosYou must configure cluster settings and then create and copy a keytab file to nodes onthe cluster.

Before you begin




u You have created a SmartConnect zone for your Isilon cluster. If you have not, seeManaging external client connections with SmartConnect.

You configure the cluster to use the principal that matches the name of the SmartConnectzone that you created. This principal must be added to the KDC and exported to thekeytab file, which is then copied to the nodes in the cluster.

Procedure

1. Run the following command to set the HDFS principal to be used and return theprincipal that you must add to your KDC, where <SmartConnect-zone-name> is the name

Hadoop


of the SmartConnect zone that you created for your cluster. You must be logged on asthe root user to run this command:

isi hdfs krb5 --kerb-instance=<SmartConnect-zone-name>2. Run the following commands on the cluster with the kadmin utility to create a

principal on the KDC, where <admin-principal-name> is the value for the admincredentials for your KDC, and the principal that is being created is the one returnedfrom the previous command:

kadmin -p <admin-principal-name>add_principal -randkey hdfs/<SmartConnect-zone-name>

3. In the kadmin utility, run the following command to add the principal to a keytab fileand export the keytab file with a unique file name:

ktadd -k <keytab /path/filename> hdfs/<SmartConnect-zone-name>kadmin: quit

4. In the OneFS command-line interface, securely copy the keytab file to all the nodes inyour cluster. You can script this copying task if you have several nodes. The followingcommand is just one example of a script that copies the keytab files to every node:

for ip in `isi_nodes %{internal}`; do scp <keytab /path/filename> $ip:/etc/; done

Configure the HDFS daemon to use your keytab fileYour final step in configuring Kerberos authentication over HDFS is to configure the HDFSdaemon to use your keytab file.

Before you begin

You must be logged in as the root user to run this command.

You must set the OneFS HDFS daemon to use the keytab file that you exported:

Procedure

1. In the OneFS command-line interface, configure the OneFS HDFS daemon to use thekeytab file that you exported by running the following command, where <filename> is aplaceholder for the name of your keytab file:

isi hdfs krb5 --keytab=/etc/<filename>After you finish

You must modify the Hadoop configuration files on your Hadoop clients as the final stepin configuring HDFS authentication through Kerberos.

Configuring HDFS authentication with Active Directory KerberosYou can authenticate and authorize HDFS connections with Kerberos and Microsoft ActiveDirectory.

Before you begin, confirm the following checklist items:

u The cluster is running properly with HDFS in simple security authentication mode.

u A SmartConnect zone is configured on the Isilon cluster and hostname resolution iscorrect.

u You know how to configure Linux and Unix systems to work with Kerberos and ActiveDirectory.

u You know how to manage an EMC Isilon cluster through the command-line interface.

You must perform the following tasks in sequence to configure HDFS authenticationthrough Active Directory Kerberos.

Hadoop

Configuring HDFS authentication with Active Directory Kerberos 381

1. Create HDFS user accounts in Active Directory.

2. Create keytab files for HDFS user accounts.

3. Configure the krb5.conf file.

4. Configure cluster settings for HDFS.

5. Modify Hadoop configuration files for Kerberos authentication.

For more information about authenticating HDFS connections through Kerberos, see thewhite paper EMC Isilon Best Practices for Hadoop Data Storage.

Create HDFS user accounts in Active DirectoryYou must create HDFS-specific user accounts in Active Directory before you configureKerberos authentication.

You must create user accounts in Active Directory for HDFS, the JobTracker, and theTaskTracker with a service principal name.

Note

After you configure Kerberos authentication, you cannot use a local user on a Hadoopcompute client to access HDFS data on an EMC Isilon cluster; you must use an account inActive Directory.

Procedure

1. With standard Active Directory tools, create HDFS user accounts, similar to thefollowing patterns for the three service principal names, where <jt-user-name>@<DOMAIN> and <tt-user-name>@<DOMAIN> are placeholders for JobTracker andTaskTracker user names:

hdfs@<DOMAIN><jt-user-name>@<DOMAIN><tt-user-name>@<DOMAIN>

Create keytab files for HDFS user accountsYou must create keytab files for the HDFS user accounts with Microsoft's ktpass utilityand distribute them to Hadoop compute clients.

After the keytab files are created, you must distribute them to the Hadoop computeclients that need them. To avoid account name conflicts after you configure Kerberos, youmay need to delete any local accounts on your Isilon cluster if you created them for theHadoop services; that is, user accounts such as hdfs, mapred, and hadoop, inaddition to the local group named hadoop.

Procedure

1. Run a command similar to the following example to create keytab files for ActiveDirectory users with the ktpass utility:

ktpass princ <user>/<computername>@<DOMAIN> \ mapuser<user>@<DOMAIN> +rndPass -out Documents/user.keytab \ /ptype KRB5_NT_PRINCIPAL

2. Establish a secure connection such as SCP and distribute the keytab files to theHadoop compute clients that need them.

Hadoop


Note

The Kerberos keytab file contains an encrypted, local copy of the host's key, which ifcompromised may potentially allow unrestricted access to the host computer. It iscrucial to protect the keytab file with file-access permissions.

Configure cluster settings for HDFSYou must configure settings on the cluster as part of setting up HDFS authentication withActive Directory and Kerberos.

Before you begin



u You have established an SSH connection to a node in the cluster and are logged in asthe root user.

Procedure

1. Run the following commands to join the cluster to the Active Directory domain, where<DOMAIN> is a placeholder for your domain name.

Note

You can skip this command if you are already joined to a domain:

isi auth ads create <DOMAIN> administratorisi auth ads modify --provider-name=<DOMAIN> --assume-default-domain=true

2. Next, run the following commands, where <SmartConnect-zone-name> is a placeholderfor the SmartConnect zone that the Hadoop compute clients are connecting to:

isi hdfs krb5 --kerb-instance=<SmartConnect-zone-name>isi hdfs krb5 --keytab=DYNAMIC:/usr/lib/ \kt_isi_pstore.so:hdfs:<DOMAIN> isi auth ads spn create --spn=hdfs/<SmartConnect-zone-name> \ --domain=<DOMAIN> --user=administrator

After you finish

You must modify the Hadoop configuration files on your Hadoop clients as the final stepin configuring HDFS authentication with Kerberos through Active Directory.

Modifying Hadoop configuration files for Kerberos authenticationYou must modify four files when you make changes to Kerberos authentication overHDFS.

To configure Kerberos authentication, you must modify four Hadoop configuration files onyour Hadoop compute clients:

u krb5.conf

u mapred-site.xml

u hdfs-site.xml

u core-site.xml

You must restart the Hadoop daemons on the compute clients to apply the changes.

Hadoop

Modifying Hadoop configuration files for Kerberos authentication 383

Configure the krb5.conf fileYou must configure Hadoop compute client settings to use Active Directory.

Before you begin

You must be logged in as the root user to run the commands in this task.

When you configure the krb5.conf file on the Hadoop client, you must include thedefault domain and server settings for the KDC.

Procedure

1. On the Hadoop compute client, open the krb5.conf file in a text editor and provideinformation similar to the following example, where values in angle brackets areplaceholders for values specific to your environment:

[logging] default = FILE:/var/log/krb5libs.log kdc = FILE:/var/log/krb5kdc.log admin_server = FILE:/var/log/kadmind.log

[libdefaults] default_realm = <DOMAIN> dns_lookup_realm = false dns_lookup_kdc = false ticket_lifetime = 24h renew_lifetime = 7d forwardable = true

[realms] <DOMAIN> = { kdc = <kdc-hostname> admin_server = <kdc-admin-server-hostname> }

[domain_realm] .<DOMAIN> = <DOMAIN> <DOMAIN> = <DOMAIN>

2. Save and close the krb5.conf.

3. Make sure that both the Isilon cluster and the Hadoop compute clients use eitherActive Directory or the same NTP server as their time source.

The Kerberos standard requires that system clocks be no more than 5 minutes apart.

Modify the hdfs-site.xml fileYou must provide the values for namenodes and datanodes in the hdfs-site.xml file.

On the Hadoop compute client, add the following properties to the hdfs-site.xmlfile.

Procedure

1. Open the file hdfs-site.xml with a text editor and make the following changesaccording to the following example, where <SmartConnect-zone-name> is a placeholderfrom the resolvable hostname for the EMC Isilon cluster:

<?xml version="1.0"?><configuration> <property> <name>dfs.namenode.keytab.file</name> <value>/etc/krb5.keytab</value> </property> <property> <name>dfs.namenode.kerberos.principal</name>

Hadoop


<value>hdfs/<hostname>@<REALM | DOMAIN></value> </property> <property> <name>dfs.datanode.keytab.file</name> <value>/etc/krb5.keytab</value> </property> <property> <name>dfs.datanode.kerberos.principal</name> <value>hdfs/<hostname>@<REALM | DOMAIN></value> </property></configuration>

2. Save the changes to the file.

3. Restart the Hadoop daemons on the compute client to apply the changes in theHadoop configuration file.

Modify the core-site.xml file for authentication and authorizationYou must provide the values for authentication and authorization in the core-site.xml file.

On the Hadoop compute client, add the following authentication and authorizationproperties to the core-site.xml file.

Procedure

1. Open the file core-site.xml with a text editor and make the following changes,using the following example for configuring two compute clients with a serviceprincipal name (SPN):

<?xml version="1.0"?><configuration> <property> <name>fs.default.name</name> <value>hdfs://<cluster-SmarConnect-name>:8020</value>  </property> <property> <name>hadoop.rpc.protection</name> <value>authentication</value> <description> This field sets the quality of protection for secured sasl connections. Possible values are authentication, integrity and privacy. Authentication means authentication only and no integrity or privacy; integrity implies authentication and integrity are enabled; and privacy implies all of authentication, integrity and privacy are enabled. </description> </property> <property><property> <name>hadoop.security.authentication</name> <value>kerberos</value> </property> <property> <name>hadoop.security.authorization</name> <value>true</value> </property></configuration>



Hadoop

Modifying Hadoop configuration files for Kerberos authentication 385

Modify the mapred-site.xml file.You must provide the values for JobTracker and TaskTracker in the mapred-site.xmlfile.

On the Hadoop compute client, add the following authentication and authorizationproperties in the mapred-site.xml file.

Note

To run Hadoop jobs or distcp, you must make sure that the principals that the Hadoop

daemons are using, the value of the TaskTracker’s and JobTracker’s Kerberos principal inmapred-site.xml, map to users on the Isilon cluster and can be resolved on the

cluster by using either OneFS local users or users from LDAP or Active Directory.

Procedure

1. Open the file mapred-site.xml with a text editor.

2. Add the principal for the JobTracker and the location of its keytab file to mapred-site.xml, using the following example for configuring two compute clients with aservice principal name (SPN):

<?xml version="1.0"?><configuration> <property> <name>mapreduce.jobtracker.kerberos.principal</name> <value><jt-user-name>/_HOST@<REALM | DOMAIN></value> </property> <property> <name>mapreduce.jobtracker.keytab.file</name> <value>/etc/<jt-user-name>.keytab</value> </property>

Next, add the principal for the TaskTracker and the location of its keytab file, using thefollowing example:

<property> <name>mapreduce.tasktracker.kerberos.principal</name> <value><tt-user-name>/_HOST@<REALM | DOMAIN></value> </property> <property> <name>mapreduce.tasktracker.keytab.file</name> <value>/etc/<tt-user-name>.keytab</value> </property></configuration>



Test the Kerberos connection to the clusterYou should verify that the connection to your cluster through Kerberos is working.

Before you begin



Test the connection from a compute client and then run a sample MapReduce job toverify your configuration.

Hadoop


Procedure

1. On a compute client, run the following commands to validate the connection to thecluster:su hdfskinit hdfs@<REALM | DOMAIN>$HADOOP_HOME/bin/hadoop fs -ls /

2. Run commands similar to the following example to initiate a MapReduce job:passwd hdfs <password>su – hdfs$HADOOP_HOME/sbin/start-yarn.sh$HADOOP_HOME/bin/hadoop jar \ $HADOOP_EXAMPLES/hadoop-mapreduce-examples.jar pi 100 1000000

Sample commands for configuring MIT Kerberos authentication over HDFSThese command-line examples move through the commands to set up OneFS to workwith an MIT Kerberos 5 KDC.

The following command example configures an Isilon SmartConnect zone named wai-kerb-sc:

kdc-demo-1# isi networks modify subnet --name=subnet0 \ --sc-service-addr=192.0.2.17Modifying subnet 'subnet0':Saving: OK

kdc-demo-1# isi networks modify pool --name=subnet0:pool0 \ --zone=wai-kerb-sc --sc-subnet=subnet0Modifying pool 'subnet0:pool0': Saving: OK

The following command example runs on an Isilon node named kdc-demo-1. The name ofthe MIT Kerberos 5 KDC and admin server is york.east.example.com.

kdc-demo-1# isi auth krb5 add realm \ --realm=EAST.EXAMPLE.COM --kdc=york.east.example.com \ --admin-server=york.east.example.com

kdc-demo-1# isi auth krb5 modify default \ --default-realm=EAST.EXAMPLE.COM

kdc-demo-1# isi auth krb5 writekdc-demo-1# kadmin -p root/[email protected] as principal root/[email protected] with password.Password for root/[email protected]:

kadmin: quit

kdc-demo-1# ping wai-kerb-scPING wai-kerb-sc.east.example.com (192.0.2.11): 56 data bytes64 bytes from 192.0.2.11: icmp_seq=0 ttl=64 time=0.561 ms

kdc-demo-1# isi hdfs krb5 --kerb-instance=wai-kerb-scAdd this principal to your KDC: \ hdfs/wai-kerb-sc.east.example.com@<YOUR-REALM.COM>

kdc-demo-1# kadmin -p root/adminAuthenticating as principal root/admin with password.Password for root/[email protected]:

kadmin: add_principal -randkey hdfs/wai-kerb-sc.east.example.com WARNING: no policy specified for hdfs/[email protected]; defaulting to no policyPrincipal "hdfs/[email protected]"created.kadmin: ktadd -k /ifs/hdfs.keytab hdfs/wai-kerb-sc.east.example.com Entry for principal hdfs/wai-kerb-sc.east.example.com with kvno 3,

Hadoop

Sample commands for configuring MIT Kerberos authentication over HDFS 387

encryption type AES-256 CTS mode with 96-bit SHA-1 HMAC added tokeytab WRFILE:/ifs/hdfs.keytab.Entry for principal hdfs/wai-kerb-sc.east.example.com with kvno 3,encryption type ArcFour with HMAC/md5 added to keytab WRFILE:/ifs/hdfs.keytab.Entry for principal hdfs/wai-kerb-sc.east.example.com with kvno 3,encryption type Triple DES cbc mode with HMAC/sha1 added to keytabWRFILE:/ifs/hdfs.keytab.Entry for principal hdfs/wai-kerb-sc.east.example.com with kvno 3,encryption type DES cbc mode with CRC-32 added to keytab WRFILE:/ifs/hdfs.keytab.kdc-demo-1# for ip in `isi_nodes %{internal}`;do scp /ifs/hdfs.keytab$ip:/etc/;done Password:hdfs.keytab 100% 666 0.7KB/s 0.7KB/s 00:00 Max throughput: 0.7KB/s

kdc-demo-1# kinit -k -t /etc/hdfs.keytab \ hdfs/wai-kerb-sc.east.example.comkdc-demo-1# klistTicket cache: FILE:/tmp/krb5cc_0Default principal: hdfs/[email protected] Valid starting Expires Service principal01/28/14 15:15:34 01/29/14 01:15:34 krbtgt/[email protected] until 01/29/14 15:13:46 kdc-demo-1# kdestroy

kdc-demo-1# isi hdfs krb5 --keytab=/etc/hdfs.keytab

Troubleshooting Kerberos authenticationKerberos authentication problems can be difficult to diagnose, but you can check thefollowing settings to troubleshoot your configuration.

Follow these steps to troubleshoot authentication problems:

u Check all the configuration parameters, including the location and validity of thekeytab file.

u Check user and group accounts for permissions. Make sure that there are noduplicate accounts across systems, such as a local hdfs account on OneFS and anhdfs account in Active Directory.

u Make sure that the system clocks on the Isilon nodes and the Hadoop clients aresynchronized with a formal time source such as Active Directory or NTP. The Kerberosstandard requires that system clocks be no more than 5 minutes apart.

u Check to confirm that the service principal name of a Hadoop service, such asTaskTracker, is not mapped to more than one object in Active Directory.

Hadoop


CHAPTER 22

Antivirus


u Antivirus overview............................................................................................... 390u On-access scanning............................................................................................ 390u Antivirus policy scanning.....................................................................................391u Individual file scanning....................................................................................... 391u Antivirus scan reports......................................................................................... 391u ICAP servers........................................................................................................ 392u Supported ICAP servers....................................................................................... 392u Anitvirus threat responses...................................................................................392u Configuring global antivirus settings................................................................... 393u Managing ICAP servers........................................................................................ 395u Create an antivirus policy.................................................................................... 397u Managing antivirus policies.................................................................................397u Managing antivirus scans....................................................................................398u Managing antivirus threats..................................................................................399u Managing antivirus reports..................................................................................401

Antivirus 389

Antivirus overviewYou can scan the files you store on an Isilon cluster for computer viruses and othersecurity threats by integrating with third-party scanning services through the InternetContent Adaptation Protocol (ICAP). OneFS sends files through ICAP to a server runningthird-party antivirus scanning software. These servers are referred to as ICAP servers.ICAP servers scan files for viruses.

After an ICAP server scans a file, it informs OneFS of whether the file is a threat. If a threatis detected, OneFS informs system administrators by creating an event, displaying nearreal-time summary information, and documenting the threat in an antivirus scan report.You can configure OneFS to request that ICAP servers attempt to repair infected files. Youcan also configure OneFS to protect users against potentially dangerous files bytruncating or quarantining infected files.

Before OneFS sends a file to be scanned, it ensures that the scan is not redundant. If afile has already been scanned and has not been modified, OneFS will not send the file tobe scanned unless the virus database on the ICAP server has been updated since the lastscan.

Note

Antivirus scanning is available only if all nodes in the cluster are connected to theexternal network.

On-access scanningYou can configure OneFS to send files to be scanned before they are opened, after theyare closed, or both. Sending files to be scanned after they are closed is faster but lesssecure. Sending files to be scanned before they are opened is slower but more secure.

If OneFS is configured to ensure that files are scanned after they are closed, when a usercreates or modifies a file on the cluster, OneFS queues the file to be scanned. OneFS thensends the file to an ICAP server to be scanned when convenient. In this configuration,users can always access files without any delay. However, it is possible that after a usermodifies or creates a file, a second user might access the file before the file is scanned. Ifa virus was introduced to the file from the first user, the second user will be able toaccess the infected file. Also, if an ICAP server is unable to scan a file, the file will still beaccessible to users.

If OneFS ensures that files are scanned before they are opened, when a user attempts todownload a file from the cluster, OneFS first sends the file to an ICAP server to bescanned. The file is not sent to the user until the scan is complete. Scanning files beforethey are opened is more secure than scanning files after they are closed, because userscan access only scanned files. However, scanning files before they are opened requiresusers to wait for files to be scanned. You can also configure OneFS to deny access to filesthat cannot be scanned by an ICAP server, which can increase the delay. For example, ifno ICAP servers are available, users will not be able to access any files until the ICAPservers become available again.

If you configure OneFS to ensure that files are scanned before they are opened, it isrecommended that you also configure OneFS to ensure that files are scanned after theyare closed. Scanning files as they are both opened and closed will not necessarilyimprove security, but it will usually improve data availability when compared to scanningfiles only when they are opened. If a user wants to access a file, the file may have already

Antivirus


been scanned after the file was last modified, and will not need to be scanned again ifthe ICAP server database has not been updated since the last scan.

Antivirus policy scanningYou can create antivirus scanning policies that send files from a specified directory to bescanned. Antivirus policies can be run manually at any time, or configured to runaccording to a schedule.

Antivirus policies target a specific directory on the cluster. You can prevent an antiviruspolicy from sending certain files within the specified root directory based on the size,name, or extension of the file. Antivirus policies do not target snapshots. Only on-accessscans include snapshots. Antivirus scans are handled by the OneFS job engine, andfunction the same as any system job.

Individual file scanningYou can send a specific file to an ICAP server to be scanned at any time.

If a virus is detected in a file but the ICAP server is unable to repair it, you can send thefile to the ICAP server after the virus database had been updated, and the ICAP servermight be able to repair the file. You can also scan individual files to test the connectionbetween the cluster and ICAP servers.

Antivirus scan reportsOneFS generates reports about antivirus scans. Each time that an antivirus policy is run,OneFS generates a report for that policy. OneFS also generates a report every 24 hoursthat includes all on-access scans that occurred during the day.

Antivirus scan reports contain the following information:

u The time that the scan started.

u The time that the scan ended.

u The total number of files scanned.

u The total size of the files scanned.

u The total network traffic sent.

u The network throughput that was consumed by virus scanning.

u Whether the scan succeeded.

u The total number of infected files detected.

u The names of infected files.

u The threats associated with infected files.

u How OneFS responded to detected threats.

u The name and IP address of the user that triggered the scan.

This information is not included in reports triggered by antivirus scan policies.

Antivirus

Antivirus policy scanning 391

ICAP serversThe number of ICAP servers that are required to support an Isilon cluster depends on howvirus scanning is configured, the amount of data a cluster processes, and the processingpower of the ICAP servers.

If you intend to scan files exclusively through antivirus scan policies, it is recommendedthat you have a minimum of two ICAP servers per cluster. If you intend to scan files onaccess, it is recommended that you have at least one ICAP server for each node in thecluster.

If you configure more than one ICAP server for a cluster, it is important to ensure that theprocessing power of each ICAP server is relatively equal. OneFS distributes files to theICAP servers on a rotating basis, regardless of the processing power of the ICAP servers. Ifone server is significantly more powerful than another, OneFS does not send more files tothe more powerful server.

Supported ICAP serversOneFS supports ICAP servers running the following antivirus scanning software:

u Symantec Scan Engine 5.2 and later.u Trend Micro Interscan Web Security Suite 3.1 and later.u Kaspersky Anti-Virus for Proxy Server 5.5 and later.u McAfee VirusScan Enterprise 8.7 and later with VirusScan Enterprise for Storage 1.0

and later.

Anitvirus threat responsesYou can configure the system to repair, quarantine, or truncate any files that the ICAPserver detects viruses in.

OneFS and ICAP servers react in one or more of the following ways when threats aredetected:

AlertAll threats that are detected cause an event to be generated in OneFS at the warninglevel, regardless of the threat response configuration.

RepairThe ICAP server attempts to repair the infected file before returning the file to OneFS.

QuarantineOneFS quarantines the infected file. A quarantined file cannot be accessed by anyuser. However, a quarantined file can be removed from quarantine by the root user ifthe root user is connected to the cluster through secure shell (SSH).

If you backup your cluster through NDMP backup, quarantined files will remainquarantined when the files are restored. If you replicate quarantined files to anotherIsilon cluster, the quarantined files will continue to be quarantined on the targetcluster. Quarantines operate independently of access control lists (ACLs).

TruncateOneFS truncates the infected file. When a file is truncated, OneFS reduces the size ofthe file to zero bytes to render the file harmless.

Antivirus


You can configure OneFS and ICAP servers to react in one of the following ways whenthreats are detected:

Repair or quarantineAttempts to repair infected files. If an ICAP server fails to repair a file, OneFSquarantines the file. If the ICAP server repairs the file successfully, OneFS sends thefile to the user. Repair or quarantine can be useful if you want to protect users fromaccessing infected files while retaining all data on a cluster.

Repair or truncateAttempts to repair infected files. If an ICAP server fails to repair a file, OneFStruncates the file. If the ICAP server repairs the file successfully, OneFS sends the fileto the user. Repair or truncate can be useful if you do not care about retaining alldata on your cluster, and you want to free storage space. However, data in infectedfiles will be lost.

Alert onlyOnly generates an event for each infected file. It is recommended that you do notapply this setting.

Repair onlyAttempts to repair infected files. Afterwards, OneFS sends the files to the user,whether or not the ICAP server repaired the files successfully. It is recommended thatyou do not apply this setting. If you only attempt to repair files, users will still beable to access infected files that cannot be repaired.

QuarantineQuarantines all infected files. It is recommended that you do not apply this setting. Ifyou quarantine files without attempting to repair them, you might deny access toinfected files that could have been repaired.

TruncateTruncates all infected files. It is recommended that you do not apply this setting. Ifyou truncate files without attempting to repair them, you might delete dataunnecessarily.

Configuring global antivirus settingsYou can configure global antivirus settings that are applied to all antivirus scans bydefault.

Exclude files from antivirus scansYou can prevent files from being scanned by antivirus policies.

Procedure

1. Click Data Protection > Antivirus > Settings.

2. In the File size restriction area, specify whether to exclude files from being scannedbased on size.

l Click Scan all files regardless of size.

l Click Only scan files smaller than the maximum file size and specify a maximumfile size.

3. In the Filename restrictions area, specify whether to exclude files from being scannedbased on file names and extensions.

Antivirus

Configuring global antivirus settings 393

l Click Scan all files.

l Click Only scan files with the following extensions or filenames.

l Click Scan all files except those with the following extensions or filenames.

4. Optional: If you chose to exclude files based on file names and extensions, specifywhich files will be selected.

a. In the Extensions area, click Edit list, and specify extensions.

b. In the Filenames area, click Edit list, and specify filenames.

You can specify the following wild cards:


For example, specifying m* would match movies and m123.

[ ] Matches any characters contained in the brackets, or a range of characters separatedby a dash.

For example, specifying b[aei]t would match bat, bet, and bit.

For example, specifying 1[4-7]2 would match 142, 152, 162, and 172.


For example, specifying b[!ie] would match bat but not bit or bet.


For example, specifying [[c]at would match cat, and [at.

You can match a dash within a bracket if it is either the first or last character.

For example, specifying car[-s] would match cars, and car-.

? Matches any character in place of the question mark.

For example, specifying t?p would match tap, tip, and top.

5. Click Submit.

Configure on-access scanning settingsYou can configure OneFS to automatically scan files as they are accessed by users. On-access scans operate independently of antivirus policies.

Procedure


2. In the On Access Scans area, specify whether you want files to be scanned as they areaccessed.

l To require that all files be scanned before they are opened by a user, select Scanfiles when they are opened, and then specify whether you want to allow access tofiles that cannot be scanned.

l To scan files after they are closed, select Scan files when they are closed.

3. In the Directories to be scanned area, specify the directories that you want to applyon-access settings to.

Antivirus


If no directories are specified, on-access scanning settings are applied to all files. Ifyou specify a directory, only files from the specified directories will be scanned asthey are accessed.

4. Click Submit.

Configure antivirus threat response settingsYou can configure how OneFS responds to detected threats.

Procedure


2. In the Action on detection area, specify how you want OneFS to react to potentiallyinfected files.

Configure antivirus report retention settingsYou can configure how long OneFS retains antivirus reports before automatically deletingthem.

Procedure


2. In the Reports area, specify how long you want OneFS to keep reports for.

Enable or disable antivirus scanningYou can enable or disable all antivirus scanning. This procedure is available only throughthe web administration interface.

Procedure

1. Click Data Protection > Antivirus > Summary.

2. In the Service area, click Enable or Disable.

Managing ICAP serversBefore you can send files to be scanned on an ICAP server, you must configure OneFS toconnect to the server. You can test, modify, and remove an ICAP server connection. Youcan also temporarily disconnect and reconnect to an ICAP server.

Add and connect to an ICAP serverYou can add and connect to an ICAP server. After a server is added, OneFS can send filesto the server to be scanned for viruses.

Procedure


2. In the ICAP Servers area, click Add server.

3. In the Add ICAP Server dialog box, in the ICAP URL field, type the IP address of an ICAPserver.

4. Optional: In the Description field, type a description of this ICAP server.

5. Click Submit.

The ICAP server is displayed in the ICAP Servers table.

Antivirus

Configure antivirus threat response settings 395

Test an ICAP server connectionYou can test the connection between the cluster and an ICAP server. This procedure isavailable only through the web administration interface.

Procedure


2. In the ICAP Servers table, in the row for the ICAP server, click Test connection.

If the connection test succeeds, the Status column displays a green icon. If theconnection test fails, the Status column displays a red icon.

Modify ICAP connection settingsYou can modify the IP address and optional description of ICAP server connections.

Procedure


2. In the ICAP Servers table, in the row for an ICAP server, click Edit.

3. Modify settings, and then click Submit.

Temporarily disconnect from an ICAP serverIf you want to prevent OneFS from sending files to an ICAP server, but want to retain theICAP server connection settings, you can temporarily disconnect from the ICAP server.

Procedure


2. In the ICAP Servers table, in the row for an ICAP server, click Disable.

Reconnect to an ICAP serverYou can reconnect to an ICAP server that you have temporarily disconnected from.

Procedure


2. In the ICAP Servers table, in the row for an ICAP server, click Enable.

Remove an ICAP serverYou can permanently disconnect from the ICAP server.

Procedure


2. In the ICAP Servers table, in the row for an ICAP server, click Delete.

Antivirus


Create an antivirus policyYou can create an antivirus policy that causes specific files to be scanned for viruseseach time the policy is run.

Procedure

1. Click Data Protection > Antivirus > Policies.

2. Click Add policy.

3. In the Name field, type a name for the antivirus policy.

4. Click Add directory and select a directory that you want to scan.

Optionally, repeat this step to specify multiple directories.

5. In the Restrictions area, specify whether you want to enforce the file size and namerestrictions specified by the global antivirus settings.

l Click Enforce file size and filename restrictions.

l Click Scan all files within the root directories.

6. In the Run policy area, specify whether you want to run the policy according to aschedule or manually.

Scheduled policies can also be run manually at any time.

Options Description

Run the policy onlymanually.

Click Manually

Run the policy accordingto a schedule.

a. Click Scheduled.

b. In the Interval area, specify on what days you wantthe policy to run.

c. In the Frequency area, specify how often you wantthe policy to run on the specified days.

7. Click Submit.

Managing antivirus policiesYou can modify and delete antivirus policies. You can also temporarily disable antiviruspolicies if you want to retain the policy but do not want to scan files.

Modify an antivirus policyYou can modify an antivirus policy.

Procedure

1. Click Data Protection > Antivirus > Policy.

2. In the Policies table, click the name of the antivirus policy that you want to modify.

3. Modify settings, and then click Submit.

Antivirus

Create an antivirus policy 397

Delete an antivirus policyYou can delete an antivirus policy.

Procedure


2. In the Policies table, in the row for an antivirus policy, click Delete.

Enable or disable an antivirus policyYou can temporarily disable antivirus policies if you want to retain the policy but do notwant to scan files.

Procedure


2. In the Policies table, in the row for an antivirus policy, click Enable or Disable.

View antivirus policiesYou can view antivirus policies.

Procedure


2. In the Policies table, view antivirus policies.

Managing antivirus scansYou can scan multiple files for viruses by manually running an antivirus policy, or scan anindividual file without an antivirus policy. You can also stop antivirus scans.

Scan a fileYou can manually scan an individual file for viruses. This procedure is available onlythrough the command-line interface (CLI).

Procedure


2. Run the isi avscan manual command.

For example, the following command scans /ifs/data/virus_file:isi avscan manual /ifs/data/virus_file

Manually run an antivirus policyYou can manually run an antivirus policy at any time. This procedure is available onlythrough the web administration interface.

Procedure


2. In the Policies table, in the row for a policy, click Start.

Antivirus


Stop a running antivirus scanYou can stop a running antivirus scan. This procedure is available only through the webadministration interface.

Procedure


2. In the Currently Running table, in the row for an antivirus scan, click Cancel.

Managing antivirus threatsYou can repair, quarantine, or truncate files in which threats are detected. If you thinkthat a quarantined file is no longer a threat, you can rescan the file or remove the file fromquarantine.

Manually quarantine a fileYou can quarantine a file to prevent the file from being accessed by users.

Procedure

1. Click Data Protection > Antivirus > Detected Threats.

2. In the Detected Threats table, in the row of a file, click Quarantine.

Rescan a fileYou can rescan the file for viruses if, for example, you believe that a file is no longer athreat.

Procedure


2. In the Detected Threats table, in the row of a file, click Rescan.

Remove a file from quarantineYou can remove a file from quarantine if, for example, you believe that the file is nolonger a threat.

Procedure


2. In the Detected Threats table, in the row of a file, click Restore.

Manually truncate a fileIf a threat is detected in a file, and the file is irreparable and no longer needed, you cantruncate the file.

Procedure


2. In the Detected Threats table, in the row for a file, click Truncate.

The Confirm dialog box appears.

3. Click Yes.

Antivirus

Stop a running antivirus scan 399

View threatsYou can view files that have been identified as threats by an ICAP server.

Procedure


2. In the Detected Threats table, view potentially infected files.

Antivirus threat informationYou can view information about the antivirus threats that are reported by an ICAP server.

StatusThe color of the icon indicates the status of the potentially infected file.

RedOneFS did not take any action on the file.

OrangeOneFS truncated the file.

YellowOneFS quarantined the file.

ThreatDisplays the name of the detected threat as it is recognized by the ICAP server.

FilenameDisplays the name of the potentially infected file.

DirectoryDisplays the directory in which the file is located.

RemediationIndicates how OneFS responded to the file when the threat was detected. If OneFSdid not quarantine or truncate the file, Infected appears.

DetectedDisplays the time that the file was detected.

PolicyDisplays the name of the antivirus policy that caused the threat to be detected. If thethreat was detected as a result of a manual antivirus scan of an individual file,Manual scan appears.

CurrentlyDisplays the current state of the file.

File sizeDisplays the size of the file in bytes. Truncated files display a size of zero bytes.

Antivirus


Managing antivirus reportsIn addition to viewing antivirus reports through the web administration interface, you canexport reports to a comma-separated values (CSV) file. You can also view events that arerelated to antivirus activity.

Export an antivirus reportYou can export an antivirus report to a comma separated values (CSV) file.

Procedure

1. Click Data Protection > Antivirus > Reports.

2. In the Reports table, in the row for a report, click Export.

3. Save the CSV file.

View antivirus reportsYou can view antivirus reports.

Procedure

1. Click Data Protection > Antivirus > Reports.

2. In the Reports table, in the row for a report, click View Details.

View antivirus eventsYou can view events that relate to antivirus activity.

Procedure

1. Click Dashboard > Events > Event History.

2. In the Event History table, view all events.

All events related to antivirus scans are classified as warnings. The following eventsare related to antivirus activities:

Anti-Virus scan found threatsA threat was detected by an antivirus scan. These events refer to specific reportson the Antivirus Reports page but do not provide threat details.

No ICAP Servers availableOneFS is unable to communicate with any ICAP servers.

ICAP Server Unresponsive or InvalidOneFS is unable to communicate with an ICAP server.

Antivirus

Managing antivirus reports 401

CHAPTER 23

iSCSI


u iSCSI overview.....................................................................................................404u iSCSI targets and LUNs........................................................................................404u iSNS client service...............................................................................................405u Access control for iSCSI targets........................................................................... 405u iSCSI considerations and limitations................................................................... 406u Supported SCSI mode pages............................................................................... 406u Supported iSCSI initiators................................................................................... 407u Configuring the iSCSI and iSNS services.............................................................. 407u Managing iSCSI targets....................................................................................... 409u Configuring iSCSI initiator access control............................................................ 411u Creating iSCSI LUNs.............................................................................................415u Managing iSCSI LUNs.......................................................................................... 418

iSCSI 403

iSCSI overviewThe Isilon iSCSI module enables customers to provide block storage for MicrosoftWindows, Linux, and VMware systems over an IP network. To access the iSCSI module,you must activate a separate license.

Note

The iSCSI module is supported by exception only and is limited to certain use cases. It isrecommended that you find other solutions for your block storage requirements.

You can create and manage iSCSI targets on a cluster. The targets are available as SCSIblock devices on which you can store structured and unstructured data. iSCSI targetscontain one or more logical units, each uniquely identified by a logical unit number (LUN).You can format and connect to targets, such as physical disk devices, on the local filesystem.

You can configure each target to limit access to a list of initiators. You can also requireinitiators to authenticate with a target by using the Challenge-Handshake AuthenticationProtocol (CHAP).

The iSCSI module includes the following features:

u Support for Microsoft Internet Storage Name Service (iSNS) server

u Isilon SmartConnect Advanced dynamic IP allocation

u Isilon FlexProtect

u Data mirroring from 2x to 8x

u LUN cloning

u One-way CHAP authentication

u Initiator access control

iSCSI targets and LUNsA logical unit is a storage object (such as a disk or disk array) that is accessible by aniSCSI target on an Isilon cluster. Each logical unit is uniquely identified by a logical unitnumber (LUN). Although a LUN is an identifier for a logical unit, the terms are often usedinterchangeably. A logical unit must be associated with a target, and each target cancontain one or more logical units.

The following table describes the three types of LUNs that the Isilon iSCSI modulesupports:

LUN Type Description- -Normal A normal LUN is the default LUN type for clone and imported LUNs, and the only type

available for newly created LUNs. Normal LUNs are either writeable or read-only.

Snapshot A snapshot LUN is a copy of a normal LUN or another snapshot LUN. Althoughsnapshot LUNs require little time and disk space to create, they are read-only. Youcan create snapshot LUNs by cloning existing normal or snapshot LUNs, but youcannot create snapshot clones of clone LUNs.

Clone A clone LUN is a copy of a normal, snapshot, or clone LUN. Clone LUNs areimplemented using overlay and mask files in conjunction with a snapshot. Clone

iSCSI


LUN Type Description- -

LUNs require little time and disk space to create, and the LUN is fully writeable. Youcan create clone LUNs by cloning or importing existing LUNs.

SmartConnect and iSCSI targetsYou can specify a SmartConnect service IP or virtual IP address for an initiator whenconnecting to iSCSI targets.

When an initiator connects to a target with the SmartConnect service, the iSCSI session isredirected to a node in the cluster, based on the SmartConnect connection policysettings. The default connection policy setting is round robin. If the SmartConnect serviceis configured for dynamic IP allocation, connections are redirected to other nodes in caseof failure. Dynamic IP allocation is available only if you activate a SmartConnectAdvanced license.

iSNS client serviceiSCSI initiators can discover and connect to iSCSI targets through the Microsoft InternetStorage Name Service (iSNS) protocol.

The iSNS server establishes a repository of active iSCSI nodes. The nodes can beinitiators or targets. In addition, you can enable, disable, configure, and test the iSNSclient service through the iSCSI module in OneFS.

Access control for iSCSI targetsThe iSCSI module supports Challenge-Handshake Authentication Protocol (CHAP) andinitiator access control for connections to individual targets.

The CHAP and initiator access control security options can be implemented together orused separately.

CHAP authenticationYou can authenticate initiator connections to iSCSI targets with the Challenge-HandshakeAuthentication Protocol (CHAP).

You can restrict initiator access to a target by enabling CHAP authentication and thenadding user:secret pairs to the target's CHAP secrets list. If you enable CHAPauthentication, initiators are required to provide a valid user:secret pair to authenticatetheir connections to the target. CHAP authentication is disabled by default.

Note

The Isilon iSCSI module does not support mutual CHAP authentication.

iSCSI

SmartConnect and iSCSI targets 405

Initiator access controlYou can control which initiators are allowed to connect to a target by enabling initiatoraccess control and configuring the target's initiator access list. By default, initiator accesscontrol is disabled, and all initiators are allowed to access the target.

You can restrict access to a target by enabling access control and then adding initiatorsto the target's initiator access list. If you enable access control but leave the initiatoraccess list empty, no initiators are able to access the target.

iSCSI considerations and limitationsWhen planning your iSCSI deployment, be aware of the following limitations andconsiderations.

u Multipath I/O (MPIO) is recommended only for iSCSI workflows that have primarilyread-only operations. The node must invalidate the data cache on all other nodesduring file-write operations and performance decreases in proportion to the numberof write operations. If all MPIO sessions are connected to the same node,performance should not decrease.

u The Isilon iSCSI module supports one-way Challenge-Handshake AuthenticationProtocol (CHAP). The authentication configuration is shared by all of the nodes, so atarget authenticates its initiator regardless of the node the initiator is connectingthrough.

u The Isilon iSCSI module supports the importing of normal LUNs only. Importingsnapshot LUNs and clone LUNs is not supported. You cannot back up and thenrestore a snapshot or clone LUN, or replicate snapshot or clone LUNs to anothercluster. It is recommended that you deploy a backup application to back up iSCSILUNs on the iSCSI client, as the backup application ensures that the LUN is in aconsistent state at the time of backup.

u The Isilon iSCSI module does not support the following:

l Internet Protocol Security (IPsec)

l Multiple connections per session (MCS)

l iSCSI host bus adaptors (HBAs)

l Mutual CHAP authentication

Supported SCSI mode pagesThe SCSI Mode Sense command is used to obtain device information from mode pages ina target device. The Mode Select command is used to set new values.

OneFS supports the following mode pages:

Mode page name Page code Subpage code- - -Caching mode page* 08h 00h

Return all mode pages only 3Fh 00h

Control mode page** 0Ah 00h

Informational exceptions control mode page 1Ch 00h

iSCSI


* For the caching mode page, OneFS supports the write cache enable (WCE) parameteronly.

** OneFS supports querying this mode page through the Mode Sense command, butdoes not support changing the fields of this page through the Mode Select command.

Supported iSCSI initiatorsOneFS 7.0.0 or later is compatible with the following iSCSI initiators.

Operating System iSCSI Initiator- -Microsoft Windows 2003 (32-bit and 64-bit) Microsoft iSCSI Initiator 2.08 or later (Certified)

Microsoft Windows 2008 (32-bit and 64-bit) Microsoft iSCSI Initiator (Certified)

Microsoft Windows 2008 R2 (64-bit only) Microsoft iSCSI Initiator (Certified)

Red Hat Enterprise Linux 5 Linux Open-iSCSI Initiator (Supported)

VMware ESX 4.0 and ESX 4.1 iSCSI Initiator (Certified)

VMware ESXi 4.0 and ESXi 4.1 iSCSI Initiator (Certified)

VMware ESXi 5.0 iSCSI Initiator (Certified)

Configuring the iSCSI and iSNS servicesYou can disable or enable and configure the iSCSI service and the iSNS client service. Youcan find targets for iSCSI initiators through the iSNS client service. The settings for theiSCSI and iSNS services are applied to all of the nodes in the cluster. You cannot modifythese settings for individual nodes.

Configure the iSCSI serviceYou can enable or disable the iSCSI service for all the nodes in a cluster.

Before you disable the iSCSI service, be aware of the following considerations:

u All of the current iSCSI sessions will be terminated for all the nodes in the cluster.

u Initiators cannot establish new sessions until the iSCSI service is re-enabled.

Procedure

1. Click File System Management > iSCSI > Settings.

2. In the iSCSI Service area, set the service state that you want:

l If the service is disabled, you can enable it by clicking Enable.

l If the service is enabled, you can disable it by clicking Disable.

iSCSI

Supported iSCSI initiators 407

Configure the iSNS client serviceYou can configure and enable or disable the Internet Storage Name Service (iSNS), whichiSCSI initiators use to discover targets.

Procedure

1. Click File System Management > iSCSI > Settings.

2. In the iSNS Client Service area, configure the iSNS client service settings:

l iSNS server address: Type the IP address of the iSNS server with which you want toregister iSCSI target information.

l iSNS server port: Type the iSNS server port number. The default port number is3205.

3. Click Test connection to validate the iSNS configuration settings.

If the connection to the iSNS server fails, check the iSNS server address and the iSNSserver port number.

4. Click Submit.

5. Change the service to the state that you want:

l If the service is disabled, you can enable it by clicking Enable. Enabling the serviceallows OneFS to register information about iSCSI targets.

l If the service is enabled, you can disable it by clicking Disable. Disabling theservice prevents OneFS from registering information about iSCSI targets.

View iSCSI sessions and throughputIf the iSCSI service is enabled on the cluster, you can view a summary of current iSCSIsessions and current throughput.

Note

To view historical iSCSI throughput data, you must obtain the EMC Isilon InsightIQ virtualappliance, which requires you to activate a separate license. For more information,contact your EMC Isilon representative.

Procedure

1. Click File System Management > iSCSI > Summary.

2. Review the current throughput data and current session information.

l The Current Throughput area displays a chart that illustrates overall inbound andoutbound throughput across all iSCSI sessions during the past hour, measured inkilobits per second (Kbps). This chart automatically updates every 15 seconds.

l The Current Sessions area displays information about each current connectionbetween an initiator and a target, including the client and target IP addresses;node, target, and LUN; operations per second; and the inbound, outbound, andtotal throughput in bits per second. You can view details about a target by clickingthe target name.

iSCSI


Managing iSCSI targetsYou can configure one or more targets for an iSCSI server, and each target can containone or more logical units.

iSCSI initiators on clients establish connections to the targets. Targets define connectionendpoints and serve as container objects for logical units on an iSCSI server.

You can control access to the target by configuring SmartConnect pools, initiator accesscontrol, and authentication with the Challenge-Handshake Authentication Protocol(CHAP). The iSCSI module discovers targets through a server running Microsoft InternetStorage Name Service (iSNS).

Create an iSCSI targetYou can configure one or more iSCSI targets, each with its own settings for initiatoraccess control and authentication. A target is required as a container object for one ormore logical units.

Procedure

1. Click File System Management > iSCSI > Targets & Logical Units.

2. In the Targets area, click Add target.

3. In the Name field, type a name for the target.

The name must begin with a letter and can contain only lowercase letters, numbers,and hyphens (-).

4. In the Description field, type a descriptive comment for the target.

5. In the Default path field, type the full path of the directory, beginning with /ifs,where the logical unit number (LUN) directory is created, or click Browse to select adirectory.

Note

This directory is used only if no other directory is specified during LUN creation or if aLUN is not created. The directory must be in the /ifs directory tree. The full path tothe directory is required, and wildcard characters are not supported.

6. Add one or more SmartConnect pools for the target to connect with. This settingoverrides any global default SmartConnect pools that are configured for iSCSI targets.

a. For the SmartConnect pool(s) setting, click Edit list.

b. Move pools between the Available Pools and Selected Pools lists by clicking apool and then clicking the right or left arrow. To remove all selected pools at once,click clear.

c. Click OK.

7. Click Submit.

8. Optional: In the Initiator Access Control area, enable and configure the settings forinitiator access control.

a. Click Enable to restrict target access to initiators that are added to the initiatoraccess control settings.

b. Click Add initiator.

iSCSI

Managing iSCSI targets 409

c. In the Initiator name field, type the name of the initiator that you want to allow toaccess this target, or click Browse to select from a list of initiators. An initiatorname must begin with an iqn. prefix.

d. Click OK.

Note

To continue adding initiators, click OK and add another. When you are finished addinginitiators, click OK.

9. Optional: In the CHAP Authentication area, enable and configure Challenge-Handshake Authentication Protocol (CHAP) settings.

Note

If CHAP authentication is enabled and the CHAP secrets list is empty, no initiators canaccess the target.

a. Click Enable to require initiators to authenticate with the target.

b. Click Add username.

c. In the Username field, type the name that the initiator will use to authenticate withthe target. You can specify an initiator's iSCSI qualified name (IQN) as theusername. Depending on whether you specify an IQN, valid usernames differ in thefollowing ways:

– If you specify an IQN as the username, the Username value must begin with aniqn. prefix. The characters that are allowed after the iqn. prefix arealphanumeric characters, periods (.), hyphens (-), and colons (:).

– All other usernames can use alphanumeric characters, periods (.), hyphens (-),and underscores (_).

Note

CHAP usernames and passwords are case sensitive.

d. In the Secret and Confirm secret fields, type the secret that the initiator will use toauthenticate with the target. A CHAP secret must be 12 to 16 characters long andcan contain any combination of letters, numbers, and symbols.

e. Click OK.

10.Click Submit.

Modify iSCSI target settingsYou can modify a target's description, change the path where logical unit directories arecreated, and modify the list of SmartConnect pools that the target uses. You can alsomanage the target's settings for initiator access control and authentication.

Procedure


2. In the Targets area, under Actions, click Edit for the target that you want to modify.

3. Modify the target's settings as needed.

iSCSI


Note

l Changing the default path does not affect existing logical units.

l Changing the security settings does not affect existing connections.

4. Click Submit.

Delete an iSCSI targetWhen you delete a target, all of the logical unit numbers (LUNs) that are contained in thetarget are also deleted, and all the data that is stored in the LUNs is deleted. Additionally,any iSCSI sessions that are connected to the target are terminated. This operation cannotbe undone.

Procedure


2. In the Targets area, under Actions, click Delete for the target that you want to delete.

3. In the confirmation dialog box, click Yes.

The target and all LUNs and LUN data that are contained in the target are deleted, andany iSCSI sessions on the target are terminated.

View iSCSI target settingsYou can view information about a target, including its iSCSI qualified name (IQN), defaultLUN directory path, capacity, and SmartConnect pool settings. You can also view thelogical units that are associated with the target as well as the settings for initiator accesscontrol and authentication.

Procedure


2. In the Targets area, click the name of a target.

3. Review the following sections for information on the target. To modify these settings,click Edit target.

l Target Details: Displays the target name, IQN, description, default path, capacity,and SmartConnect pool settings. The name and IQN cannot be modified.

l Logical Units: Displays any logical units that are contained in the target. You canadd or import a logical unit, or manage existing logical units. You can also selectthe columns to display or hide.

l Allowed Initiators: Displays the target's initiator access control status, and liststhe names of any initiators that are allowed to access the target when accesscontrol is enabled.

l CHAP Authentication: Displays the target's CHAP authentication status, and listsall user:secret pairs for the target.

Configuring iSCSI initiator access controlYou can configure access control to specify which initiators are allowed to connect to atarget. Access control is disabled by default.

If you enable initiator access control for an iSCSI target, access to that target is limited toa specified list of allowed initiators. If you modify a target's access control settings, the

iSCSI

Delete an iSCSI target 411

changes are applied to subsequent connection requests. Current connections areunaffected.

Configure iSCSI initiator access controlYou can configure access control to specify which initiators are allowed to connect to atarget.

If initiator access control is enabled for an iSCSI target, access is limited to a list ofinitiators. Access control is disabled by default.

Note

Modifications to a target's access control settings are applied to subsequent connectionrequests. Current connections are not affected.

Procedure


2. In the Targets area, under Actions, click Edit for the target whose initiator access stateyou want to change.

3. In the Initiator Access Control area, configure the access control state.

l If access control is disabled, click Enable to restrict target access to initiators thatyou add to the initiator access list.

Note

If you disable access control and the initiator access list is empty, no initiators areable to connect to the target.

l If access control is enabled, click Disable to allow all initiators access to thetarget.

Note

If you disable access control, the list of allowed initiators is ignored.

4. Add initiators by clicking Add initiator.

Control initiator access to a targetYou can control access to a target by adding initiators to its initiator access list. If youenable initiator access control, the initiator access list specifies which initiator names areallowed to access the target. However, the initiator access list is ignored unless initiatoraccess control is enabled.

Procedure


2. In the Targets area, under Actions, click Edit for the target that you want to allow aninitiator to access.

3. In the Initiator Access Control area, click Add initiator.

4. In the Initiator name field, type the name of the initiator that you want to allow toaccess the target, or click Browse to select from a list of known initiators. An initiatorname requires the iqn. prefix.

5. Click OK.

iSCSI


6. To continue adding initiators, click OK and add another.

7. When you are finished adding initiators, click OK.

Modify initiator nameYou can rename or replace an initiator that is allowed to connect to a target when accesscontrol is enabled.

Procedure



3. In the Initiator Access Control area, click Edit for the initiator that you want to modify.

4. Modify the initiator name.

5. Click OK.

Remove an initiator from the access listYou can remove an initiator from a target's initiator access list so that the initiator is nolonger able to connect to a target when access control is enabled.

Note

u If you remove all of the allowed initiators for a target and access control is enabled,the target will deny new connections until you disable access control.

u Removing an allowed initiator for a target does not affect the initiator's access toother targets.

Procedure



3. In the Initiator Access Control area, under Actions, click Delete for the initiator thatyou want to remove from the access list.


Create a CHAP secretTo use CHAP authentication, you must create user:secret pairs in the target's CHAPsecrets list and enable CHAP authentication. Initiators must then authenticate with thetarget by providing a user:secret pair.

Procedure


2. In the Targets area, under Actions, click Edit for the target that you want to create aCHAP secret for.

3. In the CHAP Authentication area, click Add username.

4. In the Username field, type the name that the initiator uses to authenticate with thetarget.

You can specify an initiator's iSCSI qualified name (IQN) as the username.

iSCSI

Modify initiator name 413

l If you specify an IQN as the username, the Username value must begin with aniqn. prefix. The characters that are allowed after the iqn. prefix are alphanumericcharacters, periods (.), hyphens (-), and colons (:).

l All other usernames can use alphanumeric characters, periods (.), hyphens (-), andunderscores (_).

Note

CHAP usernames and passwords are case sensitive.

5. In the Secret and Confirm secret fields, type the secret that the initiator will use toauthenticate with the target. A CHAP secret must be 12 to 16 characters long and cancontain any combination of letters, numbers, and symbols.

6. Click OK.

Modify a CHAP secretYou can modify the settings for a CHAP user:secret pair.

Procedure


2. In the Targets area, under Actions, click Edit for the target that you want to modify aCHAP user:secret pair for.

3. In the CHAP Authentication area, under Actions, click Edit for the username whosesettings you want to modify.

4. Make the changes that you want, and then click OK.

Delete a CHAP secretYou can delete a CHAP user:secret pair that is no longer needed.

Note

If you delete all of a target's CHAP secrets and CHAP authentication is enabled, noinitiators are able to access the target until you disable CHAP authentication.

Procedure


2. In the Targets area, under Actions, click Edit for the target that you want to delete aCHAP user:secret pair for.

3. In the CHAP Authentication area, under Actions, click Delete for the CHAP user:secretpair that you want to delete.


Enable or disable CHAP authenticationYou can enable or disable CHAP authentication for a target.

Note

Modifications to a target's CHAP authentication status are applied to subsequentconnection requests. Current connections are unaffected.

iSCSI


Procedure


2. In the Targets area, under Actions, click Edit for the target whose CHAP authenticationstate you want to modify.

3. In the CHAP Authentication area, configure the initiator's CHAP authentication state.

l If CHAP authentication is disabled, you can click Enable to require initiators toauthenticate with the target.

Note

If CHAP authentication is enabled and the CHAP secrets list is empty, no initiator isable to access the target.

l If CHAP authentication is enabled, click Disable to stop authenticating initiatorswith the target.

Note

If CHAP authentication is disabled, the CHAP secrets list is ignored.

4. Add CHAP user:secret pairs by clicking Add username.

Creating iSCSI LUNsYou can create a new LUN or you can clone an existing LUN to form a new LUN. LUNcloning, like LUN creation, is asynchronous.

When you create a LUN, you can set its target assignment, LUN number, directory path,size, provisioning policy, access state, write access, protection settings, and I/Ooptimization settings. The LUN number uniquely identifies the logical unit.

When you clone an existing LUN, you can set that LUN to be part of the same target or adifferent target. A cloned LUN is inaccessible by iSCSI initiators until the cloning iscomplete.

To clone a LUN, you must enable the Isilon SnapshotIQ module, which requires you toactivate a separate license. For more information, contact your EMC Isilon salesrepresentative.

Create an iSCSI LUNYou can create a logical unit and assign it to an iSCSI target for access.

Note

When you create a logical unit, you must assign it to an existing iSCSI target. Each targetcan contain one or more logical units.

Procedure


2. In the Logical Units area, click Add logical unit.

3. In the Add Logical Unit area, in theDescription field, type a descriptive comment forthe logical unit.

4. From the Target list, select the target that will contain the logical unit.

iSCSI

Creating iSCSI LUNs 415

5. Select one of the LUN number options.

l To assign the next available number to the logical unit, click Automatic. This is thedefault setting.

l To manually assign a number to the logical unit, click Manual and then, in theNumber field, type an integer value. The value must be within the range 0-255and must not be assigned to another logical unit within the target.

By default, the LUN number forms part of the directory name that is created for storingthe LUN data.

6. To manually specify the path where the LUN directory is created, in the Path field, typethe full path of the directory, beginning with /ifs, or click Browse to select thedirectory.

The directory must be in the /ifs directory tree. You must specify the full path to thedirectory, and wildcard characters are not allowed. The default path is /ifs/iscsi/ISCSI.LUN.<TargetName>.<LUNnumber>, where <TargetName> is the Targetvalue and <LUNnumber> is the LUN number.

7. In the Size field, specify the LUN capacity by typing an integer value and thenselecting a unit of measure from the list (MB, GB, or TB).

The minimum LUN size is 1 MB. The maximum LUN size is determined by the OneFSfile system. After you create a LUN, you can increase its size, but you cannot decreaseit.

8. Select one of the Provisioning options.

l To specify that blocks are unallocated until they are written, click Thin provision.

l To immediately allocate all the blocks, click Pre-allocate space. This is the defaultsetting.

Note

Allocation of all the blocks for a large LUN can take hours or even days.

9. Select one of the LUN access options.

l To make the LUN accessible, click Online. This is the default setting.

l To make the LUN inaccessible, click Offline.

10.Select one of the Write access options.

l To allow iSCSI initiators to write to the LUN, click Read-Write. This is the defaultsetting.

l To prevent iSCSI initiators from writing to the LUN, click Read-Only.

11.Under Protection Settings, from the Disk pool list, select the disk pool to contain thelogical unit.

12.From the SSD strategy list, select to specify a strategy to use if solid-state drives(SSDs) are available.

l Metadata read acceleration (Recommended): Writes metadata and all user data onhard disk drives (HDDs) and additionally creates a mirror backup of the metadataon an SSD. Depending on the global namespace acceleration setting, the SSDmirror may be an extra mirror in addition to the number required to satisfy theprotection level.

l Metadata read/write acceleration with performance redundancy (Requires moreSSD space): Writes all metadata on an SSD and writes all user data on HDDs.

iSCSI


l Data on SSDs (Requires most SSD space): Similar to metadata acceleration, butalso writes one copy of the file's user data (if mirrored) or all of the data (if notmirrored) on SSDs. Regardless of whether global namespace acceleration isenabled, any SSD blocks reside on the file's target pool if there is room. This SSDstrategy does not create additional mirrors beyond the normal protection level.

l Avoid SSDs (Reduces performance): Never uses SSDs; writes all associate file dataand metadata to HDDs only.

13.From the Protection level list, select a protection policy for the logical unit. Select UseiSCSI default (2x), which is the recommended setting for best performance, or one ofthe mirrored options, such as 2x to 8x.

14.Select one of the Write Cache options.

l To prevent write caching for files that contain LUN data, click Disabled. This is therecommended setting for LUNs.

l To allow write caching for files that store LUN data, click Enable.

Note

The Write Cache option controls whether file writes are sent to the coalescer or theendurant cache. With Write Cache disabled, which is the default and recommendedsetting, all file writes are sent to the endurant cache. The endurant cache is acommitted data guarantee. If Write Cache is enabled, all file writes are sent to thecoalescer. Write caching can improve performance, but can lead to data loss if a nodeloses power or crashes while uncommitted data is in the write cache.

15.Select one of the Data access pattern options.

l To select a random access pattern, click Random. This is the recommended settingfor LUNs.

l To select a concurrent access pattern, click Concurrency.

l To select a streaming access pattern, click Streaming. Streaming access patternscan improve performance in some workflows.

16.Click Submit.

Clone an iSCSI LUNYou can clone an existing LUN to a create a new LUN.

Procedure


2. In the Logical Units area, under Actions, click Clone for the logical unit that you wantto clone.

The page updates and the clone options appear. Most of the fields for these optionsare populated with information from the logical unit you selected to clone.

3. From the LUN type list, select Normal, Clone, or Snapshot.

4. Modify the other settings as needed.

Note

The settings for the clone vary according to the source LUN type.

5. Click Submit.

iSCSI

Clone an iSCSI LUN 417

iSCSI LUN cloning operationsDepending on the clone LUN type, the contents (or blocks) of a source LUN are eithercopied or referenced, and the attributes may or may not be copied. In general, clone andsnapshot type clone operations are fast, whereas normal type clones can take severalminutes or even hours to create, depending on the size of the LUN.

The following table describes the result of each cloning operation.

Source LUNtype

Clone LUNtype

Result

- - -Normal Normal A snapshot of the source LUN is created. The clone LUN is then created by copying the LUN data

from the snapshot. After completing the copy, the snapshot is deleted. The copy process may takeseveral hours to complete for large LUNs if the source LUN has a pre-allocated provisioning policy.The copy process may also take several minutes for thinly provisioned LUNs that are significantlyused.

Normal Snapshot A snapshot of the source LUN is created. The clone LUN is configured to reference the data fromthe snapshot. The snapshot is deleted when the clone is deleted.

Normal Clone A snapshot of the source LUN is created. The system then creates a clone LUN that references datafrom the snapshot.

Snapshot Normal The clone LUN is created by copying the LUN data from the snapshot. The copy process may takeseveral minutes to complete for large LUNs if the source LUN has a pre-allocated provisioningpolicy. The copy process may also take several minutes for thinly provisioned LUNs that are heavilyused.

Snapshot Snapshot The clone LUN is configured to reference the data from the same snapshot that the source LUNreferences. The underlying snapshot is not deleted when a LUN is deleted unless the LUN beingdeleted is the last LUN referencing the snapshot.

Snapshot Clone The clone LUN is configured to reference the data from the same snapshot that the source LUNreferences. The underlying snapshot is not deleted when a LUN is deleted unless the LUN beingdeleted is the only LUN referencing the snapshot.

Clone Normal A snapshot of the source LUN is created. The clone LUN is then created by copying the LUN datafrom the snapshot. After completing the copy, the snapshot is deleted. The copy process may takeseveral minutes to complete for large LUNs if the source LUN has a pre-allocated provisioningpolicy. The copy process may also take several minutes for thinly provisioned LUNs that are heavilyused

Clone Snapshot Not allowed.

Clone Clone A clone of the clone LUN is created. The clone LUN is configured to reference data from thesnapshot.

Managing iSCSI LUNsYou can manage a LUN in the following ways.

u Modify a LUN

u Delete a LUN

u Migrate a LUN to another target

iSCSI


u Import a LUN

u View LUN settings

Modify an iSCSI LUNYou can modify only certain settings for a logical unit.

Procedure


2. In the Logical Units area, under Actions, click Edit for the logical unit that you want tomodify.

3. Modify the logical unit's settings.

4. Click Submit.

Delete an iSCSI LUNDeleting a logical unit permanently deletes all data on the logical unit. This operationcannot be undone.

Procedure


2. In the Logical Units area, under Actions, click Delete for the logical unit that you wantto delete.


Migrate an iSCSI LUN to another targetYou can move a logical unit from one target to another, change the value of its logical unitnumber (LUN), or update the path to the LUN directory. You cannot modify the path of asnapshot LUN.

The name of a logical unit comprises its target name and its LUN value. The two parts ofthe name are separated by a colon (such as "mytarget:0").

Procedure


2. In the Logical Units area, under Actions, click Move for the logical unit that you wantto move.

3. In the To target list, click to select a new target for the logical unit.

4. Click one of the To LUN number options.


l To manually assign a number to the logical unit, click Manual and then, in theNumber box, type an integer value. The value must be within the range 0-255 andmust not be assigned to another logical unit.

5. To configure the path where the LUN directory is created, in the To path box, type thefull path of the directory, or click Browse to select the directory. If a path is notspecified, the LUN directory is unchanged from the original directory where that LUNwas created.

6. Click Submit.

iSCSI

Modify an iSCSI LUN 419

Import an iSCSI LUNYou can recreate logical units that were replicated to a remote cluster or that were backedup and then restored to a remote cluster.

The iSCSI module does not support replicating or restoring logical unit snapshots or clonecopies.

Procedure


2. In the Logical Units area, click Import logical unit.

3. In the Description field, type a descriptive comment for the logical unit.

4. In the Source path field, type the full path (beginning with /ifs) of the directory thatcontains the logical unit that you want to import, or click Browse to select thedirectory.

5. From the Target list, select the target that will contain the logical unit.

6. Select one of the LUN number options.


l To manually assign a number to the logical unit, click Manual, and then in theNumber field, type an integer value. The value must be within the range 0-255 andmust not be assigned to another logical unit.

7. Select one of the LUN access options.

l To make the LUN accessible, click Online. This is the default setting.

l To make the LUN inaccessible, click Offline.

8. Select one of the Write access options.

l To allow iSCSI initiators to write to the LUN, click Read-Write. This is the defaultsetting.

l To prevent iSCSI initiators from writing to the LUN, click Read-Only.

9. Select one of the caching options.

l To allow write caching for files storing LUN data, click Enabled.

l To prevent write caching for files storing LUN data, click Disabled.

10.Click Submit.

View iSCSI LUN settingsYou can view information about a logical unit, including its logical unit number (LUN),iSCSI target, LUN type, LUN directory path, iSCSI qualified name, and other settings.

Procedure


2. In the Logical Units area, under Target:LUN, click the name of the logical unit that youwant to view.

3. The following settings display:

l LUN: Displays the numerical identifier of the logical unit. You can modify the LUNvalue by using the move operation.

iSCSI


l Target: Displays the name of the iSCSI target that contains the logical unit. You canmodify the target by using the move operation.

l Description: Displays an optional description for the logical unit. You can modifythe description by clicking Edit LUN.

l Type: Displays the LUN type (normal, clone, or snapshot). You cannot modify thissetting.

l Size: Displays the LUN capacity. You can increase the size of normal or snapshotLUNs by clicking Edit LUN, but you cannot decrease the size. You cannot modifythe size of snapshot LUNs.

l Status: Displays the connection status (online or offline) and write accesspermissions (read-only or read-write) of the LUN. You can modify write-accesssettings for normal or clone LUNs by clicking Edit LUN. You cannot modify write-access settings for snapshot LUNs.

l Path: Displays the path to the directory where the LUN files are stored. You canchange the path for normal or snapshot LUNs by using the move operation. Youcannot modify the path for snapshot LUNs.

l Disk pool: Displays the disk pool of the LUN. You can modify the disk pool byclicking Edit LUN.

l Protection level: Displays the mirroring level (such as 2x, 3x, 4x, and so on) orFlexProtect protection policy for the LUN. You can modify the protection policy fornormal or clone LUNs by clicking Edit LUN. You cannot modify these settings forsnapshot LUNs.

l Write Cache: Displays whether SmartCache is enabled or disabled. You can changethis setting for normal or clone LUNs by clicking Edit LUN. You cannot modify thesesettings for snapshot LUNs.

l Data access pattern: Displays the access pattern setting (Random, Concurrency, orStreaming) for the LUN. You can change the access pattern for normal or cloneLUNs by clicking Edit LUN. You cannot modify these settings for snapshot LUNs.

l SCSI name: Displays the iSCSI qualified name (IQN) of the LUN. You cannot modifythis setting.

l EUI: Displays the extended unique identifier (EUI), which uniquely identifies theLUN. You cannot modify this setting.

l NAA: Displays the LUN's T11 Network Address Authority (NAA) namespace. Youcannot modify this setting.

l Serial number: Displays the serial number of the LUN. You cannot modify thissetting.

iSCSI

View iSCSI LUN settings 421

CHAPTER 24

VMware integration


u VMware integration overview...............................................................................424u VAAI.................................................................................................................... 424u VASA...................................................................................................................425u Configuring VASA support................................................................................... 426u Disable or re-enable VASA...................................................................................427

VMware integration 423

VMware integration overviewOneFS integrates with VMware infrastructures, including vSphere, vCenter, and ESXi.VMware integration enables you to view information about and interact with Isilonclusters through VMware applications.

OneFS interacts with VMware infrastructures through VMware vSphere API for StorageAwareness (VASA) and VMware vSphere API for Array Integration (VAAI).

OneFS integrates with VMware vCenter through the Isilon for vCenter plug-in. The Isilonfor vCenter plug-in enables you to locally backup and restore virtual machines on anIsilon cluster. For more information about Isilon for vCenter, see the following documents:

u Isilon for vCenter Release Notes

u Isilon for vCenter Installation Guide

u Isilon for vCenter User Guide

VAAIOneFS uses VMware vSphere API for Array Integration (VAAI) to support offloadingspecific virtual machine storage and management operations from VMware ESXihypervisors to an Isilon cluster.

VAAI support enables you to accelerate the process of creating virtual machines andvirtual disks. For OneFS to interact with your vSphere environment through VAAI, yourVMware environment must include ESXi 5.0 or later hypervisors.

If you enable VAAI capabilities for an Isilon cluster, when you clone a virtual machineresiding on the cluster through VMware, OneFS clones the files related to that virtualmachine. For more information on file clones, see Clones.

VAAI support for block storageOneFS support for VMware vSphere API for Array Integration (VAAI) for block storage isenabled by default.

Note

The iSCSI module is supported by exception only and is limited to certain use cases. It isrecommended that you find other solutions for your block storage requirements.

OneFS supports the following VAAI primitives for block storage:

u Hardware Assisted Locking

u Full Copy

u Block Zeroing

Note

OneFS does not support the thin provisioning block reclaim mechanism.

VMware integration


VAAI support for NASTo enable OneFS to use VMware vSphere API for Array Integration (VAAI) for NAS, youmust install the VAAI NAS plug-in for Isilon on the ESXi server.

For more information on the VAAI NAS plug-in for Isilon, see the VAAI NAS plug-in forIsilon Release Notes.

VASAOneFS communicates with VMware vSphere through VMware vSphere API for StorageAwareness (VASA).

VASA support enables you to view information about Isilon clusters through vSphere,including Isilon-specific alarms in vCenter. VASA support also enables you to integratewith VMware profile driven storage by providing storage capabilities for Isilon clusters invCenter. For OneFS to communicate with vSphere through VASA, your VMwareenvironment must include ESXi 5.0 or later hypervisors.

Isilon VASA alarmsIf the VASA service is enabled on an Isilon cluster and the cluster is added as a VMwarevSphere API for Storage Awareness (VASA) vendor provider in vCenter, OneFS isgenerates alarms in vSphere.

The following table describes the alarm that OneFS generates:

Alarm name Description- -Thin-provisioned LUNcapacity exceeded

There is not enough available space on the cluster to allocate space forwriting data to thinly provisioned LUNs. If this condition persists, you willnot be able to write to the virtual machine on this cluster. To resolve thisissue, you must free storage space on the cluster.

VASA storage capabilitiesOneFS integrates with VMware vCenter through VMware vSphere API for StorageAwareness (VASA) to display storage capabilities of Isilon clusters in vCenter.

The following storage capabilities are displayed through vCenter:

ArchiveThe Isilon cluster is composed of Isilon NL-Series nodes. The cluster is configured formaximum capacity.

PerformanceThe Isilon cluster is composed of Isilon i-Series, Isilon X-Series, or Isilon S-Seriesnodes. The cluster is configured for maximum performance.

Note

The Isilon I-Series and X-Series nodes contain Solid State Drives (SSDs). If a clusteris composed of i-Series, X-Series , or S-Series nodes, but does not contain SSDs, thecluster is recognized as a capacity cluster.

VMware integration

VAAI support for NAS 425

CapacityThe Isilon cluster is composed of Isilon X-Series nodes that do not contain SSDs. Thecluster is configured for a balance between performance and capacity.

HybridThe Isilon cluster is composed of nodes associated with two or more storagecapabilities. For example, if the cluster contained both Isilon S-Series and NL-Seriesnodes, the storage capability of the cluster is displayed as Hybrid.

Configuring VASA supportTo enable VMware vSphere API for Storage Awareness (VASA) support for a cluster, youmust enable the VASA daemon on the cluster, download the Isilon vendor providercertificate and add the Isilon vendor provider in vCenter.

Enable VASAYou must enable an Isilon cluster to communicate with VMware vSphere API for StorageAwareness (VASA) by enabling the VASA daemon.

Procedure


2. Enable VASA by running the following command:

isi services isi_vasa_d enable

Download the Isilon vendor provider certificateTo add an Isilon cluster VASA vendor provider in VMware vCenter, you must use a vendorprovider certificate. You can download a vendor provider certificate from an Isilon cluster.

Procedure

1. In a supported web browser, connect to an Isilon cluster at https://<IPAddress>, where <IPAddress> is the IP address of the Isilon cluster.

2. Retrieve the security certificate and save the certificate to a location on your machine.

For more information about exporting a security certificate, see the documentation ofyour browser.

Note

Record the location of where you saved the certificate. You will need this file pathwhen adding the vendor provider in vCenter.

Add the Isilon vendor providerYou must add an Isilon cluster as a vendor provider in VMware vCenter before you canview information about the storage capabilities of the cluster through vCenter.

Before you begin

Download a vendor provider certificate.

Procedure

1. In vCenter, navigate to the Add Vendor Provider window.

VMware integration


2. Fill out the following fields in the Add Vendor Provider window:

NameType a name for this VASA provider. Specify as any string. For example, type EMCIsilon Systems.

URLType https://<IPAddress>:8081/vasaprovider, where <IPAddress> is the IPaddress of a node in the Isilon cluster.

LoginType root.

PasswordType the password of the root user.

Certificate locationType the file path of the vendor provider certificate for this cluster.

3. Select the Use Vendor Provider Certificate box.

4. Click OK.

Disable or re-enable VASAYou can disable or re-enable an Isilon cluster to communicate with VMware vSpherethrough VMware vSphere API for Storage Awareness (VASA).

To disable support for VASA, you must disable both the VASA daemon and the Isilon webadministration interface. You will not be able to administer the cluster through aninternet browser while the web interface is disabled. To re-enable support for VASA, youmust enable both the VASA daemon and the web interface.

Procedure


2. Disable or enable the web interface by running one of the following commands:

l isi services apache2 disablel isi services apache2 enable

3. Disable or enable the VASA daemon by running one of the following commands:

l isi services isi_vasa_d disablel isi services isi_vasa_d enable

VMware integration

Disable or re-enable VASA 427

CHAPTER 25

File System Explorer


u File System Explorer overview..............................................................................430u Browse the file system........................................................................................ 430u Create a directory................................................................................................ 430u Modify file and directory properties..................................................................... 431u View file and directory properties........................................................................ 431u File and directory properties................................................................................431

File System Explorer 429

File System Explorer overviewThe File System Explorer is a web-based interface that enables you to manage the contentstored on the cluster. You can use the File System Explorer to navigate the Isilon filesystem (/ifs), add directories, and manage file and directory properties including dataprotection, I/O optimization, and UNIX permissions. The File System Explorer is availableonly if you are logged in through the root user account.

Isilon file system directory permissions are initially set to allow full access for all users.Any user can delete any file, regardless of the permissions on the individual file.Depending on your environment, you want to establish permission restrictions throughthe File System Explorer.

You can view and configure file and directory properties from within Windows clients thatare connected to the cluster. However, because Windows and UNIX permissions differfrom one another, you must be careful not to make any unwanted changes that affect fileand directory access.

Note

The File System Explorer displays up to 1000 files in a directory. If more than 1000 filesexist within a directory, the files are displayed without additional information, such as filesize and last modified date.

Browse the file systemYou can browse the Isilon file system (/ifs) through the File System Explorer.

Procedure

1. Navigate to File System Management > File System Explorer.

2. View files and directories.

l You can expand and collapse directories in the Directories pane.

l The contents of the selected directory are displayed in the right pane. You can viewthe contents of another directory by clicking the directory in the Directories pane.

Create a directoryYou can create a directory under /ifs through the File System Explorer.

Procedure


2. In the Directories pane, specify where you want to create the directory.

3. Click Add Directory.

4. In the New Directory Properties dialog box, in the Directory name field, type a namefor the directory.

5. From the User list, select the owner of the directory.

6. From the Group list, select the group for the directory.

7. From the Permissions table, specify the basic permissions for the directory.

8. Click Submit.



Modify file and directory propertiesYou can modify the data protection, I/O optimization, and UNIX permission properties offiles and directories through the File System Explorer.

Procedure


2. In the Directories pane, click the directory that contains the file or directory that youwant to modify permissions for.

3. In the right pane, in the row of the file or directory you want to modify permissions for,click Properties.

4. In the Properties dialog box, specify the properties of the file or directory.

5. Click Submit.

View file and directory propertiesYou can view the data protection, I/O optimization, and UNIX permission properties offiles and directories through the File System Explorer.

Procedure


2. In the Directories pane, click the directory that contains the file or directory that youwant to view permissions for.

3. In the right pane, in the row of the file or directory you want to view permissions for,click Properties.

4. In the Properties dialog box, view the properties of the file or directory.

File and directory propertiesEach file and directory is assigned specific data protection, I/O optimization, and UNIXpermission properties that you can view through the File System Explorer.

The following properties are displayed in the Properties dialog box of the File SystemExplorer:

Protection Settings

Settings managementSpecifies whether protection settings are managed manually or by SmartPools. If youmodify either or both protection settings, this property automatically refreshes toManually managed. If you specify Managed by SmartPools, the protectionsettings will automatically refresh to match the SmartPools specifications the nexttime the SmartPools job is run.

Disk poolThe disk pool whose requested protection is applied if SmartPools is configured tomanage protection settings. This property is available only if SmartPools is licensedand enabled on the cluster.


Modify file and directory properties 431

SSDThe SSD strategy that will be used for user data and metadata if solid-state drives(SSDs) are available. The following SSD strategies are available:

Metadata accelerationOneFS creates a mirror backup of file metadata on an SSD and writes the rest ofthe metadata plus all user data to hard disk drives (HDDs). Depending on theglobal namespace acceleration setting, the SSD mirror might be an extra mirrorin addition to the number required to satisfy the protection level.

Avoid SSDsOneFS does not write data or metadata to SSDs. OneFS writes all data andmetadata to HDDs only.

Data on SSDsSimilar to metadata acceleration, OneFS creates a mirror backup of filemetadata on an SSD and writes the rest of the metadata plus all user data tohard disk drives. However, OneFS also writes one copy of the file user data (ifmirrored) or all of the data (if not mirrored) to SSDs. All SSD blocks reside on thefile target pool if there is adequate space available, regardless of whetherglobal namespace acceleration is enabled. OneFS does not create additionalmirrors beyond the normal protection level.

Actual protectionThe FlexProtect or data-mirroring requested protection for this file or directory. IfSmartPools is licensed and enabled on the cluster, the default requested protectionfor files and directories is inherited from the specified disk pool.

I/O Optimization Settings

Settings ManagementSpecifies whether I/O Optimization Settings are managed manually or bySmartPools. If you modify either or both I/O optimization settings, this propertyautomatically refreshes to Manually managed. If you specify Managed bySmartPools, the I/O optimization settings values will automatically refresh to matchthe SmartPools specifications the next time the SmartPools job is run.

SmartCacheSpecifies whether write caching with SmartCache is enabled for this file or directory.



Data access patternThe optimization settings for accessing data. The following data access patterns areavailable:

ConcurrencyFile or directory is optimized to support many clients simultaneously.

StreamingFile or directory is optimized for high-speed streaming of a single file. Forexample, this pattern can be useful if a single client needs to read very quicklyfrom a single file.

RandomFile or directory is optimized for unpredictable access.

The default data access pattern of iSCSI LUNs is the random access pattern. Thedefault data access pattern of other files and directories is the concurrent accesspattern.

UNIX Permissions

UserThe owner of the file or directory.

GroupThe group of the file or directory.

PermissionsThe basic permissions for the file or directory.


File and directory properties 433