HP Serviceguard A.11.20.00 for Linux Release Notes€¦ · NFS-mountedFileSystems AsofServiceguardA.11.20.00,NFS-mounted(imported)filesystemsaresupportedasshared...

HP Serviceguard A.11.20.00 for LinuxRelease Notes

HP Part Number: 701461-001aPublished: October 2014Edition: 2.0

© Copyright 2001–2014 Hewlett-Packard Development Company. All rights reserved. Reproduction, adaptation, or translation of this document

without prior written permission is prohibited, except as allowed under copyright laws.

© Copyright 1979, 1980, 1983, 1985-96 Regents of the University of California. This software is based in part on the Fourth Berkeley Software

Distribution under license from the regents of the University of California.

© Copyright 1986-1997 Sun Microsystems, Inc.

Confidential computer software. Valid license from HP required for possession, use or copying. Consistent with FAR 12.211 and 12.212, CommercialComputer Software, Computer Software Documentation, and Technical Data for Commercial Items are licensed to the U.S. Government undervendor's standard commercial license. The information contained herein is subject to change without notice. The only warranties for HP productsand services are set forth in the express warranty statements accompanying such products and services. Nothing herein should be construed asconstituting an additional warranty. HP shall not be liable for technical or editorial errors or omissions contained herein. UNIX is a registeredtrademark of The Open Group.

Acknowledgments

This product includes software developed by the Apache Software Foundation (http://www.apache.org).

This product includes software developed by the Independent JPEG Group (http://www.ijg.org).

Trademark Notices

HP Serviceguard® is a registered trademark of Hewlett-Packard Company, and is protected by copyright.

NFS® is a registered trademark of Sun Microsystems, Inc.

NIS™ is a trademark of Sun Microsystems, Inc.

Red Hat® is a registered trademark of Red Hat Software, Inc.

Linux® is a registered trademark of Linus Torvalds.

UNIX® is a registered trademark of The Open Group.

SUSE® is a registered trademark of SUSE AG, a Novell Business.

Warranty

The information in this document is subject to change without notice

Hewlett-Packard makes no warranty of any kind with regard to this manual, including, but not limited to, the implied warranties of merchantabilityand fitness for a particular purpose. Hewlett-Packard shall not be held liable for errors contained herein or direct, indirect, special, incidental orconsequential damages in connection with the furnishing, performance, or use of this material.

A copy of the specific warranty terms applicable to your Hewlett-Packard product and replacement parts can be obtained from your local Salesand Service Office.

Revision History

Publication dateEdition numberSupported versionsSupported operatingsystems

Manufacturing partnumber

June, 20121Red Hat 5.7Linux701461-001Red Hat 5.8Red Hat 6.1Red Hat 6.2

October, 20142Red Hat 5.7Linux701461-001aRed Hat 5.8Red Hat 6.1Red Hat 6.2

ContentsOverview................................................................................................................................5Serviceguard Bundled Components - Product Structure...................................................................5Supported Platforms and Linux Distributions..................................................................................5New and Changed Features in This Release.................................................................................5

New Features of Serviceguard...............................................................................................5New Features of Serviceguard Manager............................................................................5

Changed Features................................................................................................................6Features not Supported in this Version.....................................................................................6About the New Features.......................................................................................................6

Integration of Custom Defined Monitor in Serviceguard Using Generic Resources....................6Advantages of Generic Resources.................................................................................6

NFS-mounted File Systems................................................................................................7Halting a Node or the Cluster while Keeping Packages Running (Live Application Detach)........7Support for Virtualization with ESX as Hypervisor and Guests as Serviceguard Nodes..............7Improved Performance while Starting and Halting a Non-detached Multi-node Package...........7

Deprecated or Obsolete Features...........................................................................................7Defects Fixed in This Version......................................................................................................8Known Problems and Limitations.................................................................................................8

Known Problems and Limitations in Serviceguard.....................................................................8Known Problems and Limitations in Serviceguard Manager.......................................................8

Compatibility and Installation Requirements..................................................................................9Installing HP Serviceguard for Linux........................................................................................9

Pre-Installation Checklist for Serviceguard...........................................................................9Pre-Installation Checklist for Serviceguard Manager...........................................................10Considerations when Upgrading Serviceguard..................................................................10Considerations when Installing Serviceguard....................................................................11

Ports Requirements.............................................................................................................11Ports Needed for Authentication...........................................................................................12Ports Required by Serviceguard Manager..............................................................................12System Firewalls.................................................................................................................12Installing the Quorum Server...............................................................................................13Setting Permissions on the /root/.rhosts File...........................................................................13Rolling Software Upgrades..................................................................................................14

Requirements................................................................................................................14Limitations ........................................................................................................................15Preparation.......................................................................................................................15Procedure — Performing the Rolling Upgrade........................................................................16

Keeping Kernels Consistent.............................................................................................18Example .....................................................................................................................18

Step 1. Halt First Node..............................................................................................19Step 2. Upgrade First Node.......................................................................................19Step 3. Restart Cluster on First Node...........................................................................19Step 4. Repeat the Process on Next Node...................................................................19Step 5. Move Package Back to Original Node.............................................................20

Rolling Upgrade from A.11.19.X to A.11.20.00........................................................................20Performing Offline Upgrade from A.11.18.X to A.11.20.00........................................................21Rolling Upgrade in Serviceguard Manager...........................................................................21

Operating System and Version Compatibility..............................................................................21Hardware Requirements..........................................................................................................21Related Information.................................................................................................................22

Further Information.............................................................................................................22Legacy Packages....................................................................................................................22

Contents 3

Quorum Server Upgrade Required if You Are Using an Alternate Address......................................22ipnodes Entries Needed in /etc/nsswitch.conf............................................................................23High Availability Consulting Services........................................................................................23Release Notes Revisions..........................................................................................................23Software Availability in Native Languages.................................................................................23

4 Contents

OverviewThis document provides information about HP Serviceguard A.11.20.00 for Linux.

Serviceguard Bundled Components - Product StructureFor information about the product structure and licensing information on Serviceguard LinuxA.11.20.00, see Quick Specs document.

Supported Platforms and Linux DistributionsServiceguard A.11.20.00 is available on HP ProLiant Servers for the following Linux distributions:• Red Hat Enterprise Linux 5 server or Advanced Platform

• Red Hat Enterprise Linux 6 server or Advanced Platform

NOTE: For full details of supported errata, etc., see the HP Serviceguard for Linux CertificationMatrix at www.hp.com/info/sglx.

New and Changed Features in This ReleaseThis section describes the new and changed features for HP Serviceguard A.11.20.00 for Linux.

New Features of Serviceguard• “Integration of Custom Defined Monitor in Serviceguard Using Generic Resources” (page 6)

• “NFS-mounted File Systems” (page 7)

• “Halting a Node or the Cluster while Keeping Packages Running (Live Application Detach)”(page 7)

• “Support for Virtualization with ESX as Hypervisor and Guests as Serviceguard Nodes” (page7)

• Support for ext3 and ext4 file systems on logical volumes created out of volume groups in RedHat configured as part of Serviceguard packages.

NOTE: Only RHEL 6.1 and RHEL 6.2 are supported in Serviceguard A.11.20.00.

• Support for Quorum Server on Red Hat 6.0

• “Improved Performance while Starting and Halting a Non-detached Multi-node Package”(page 7)

New Features of Serviceguard Manager

• Support for configuration, administration, and monitoring for up to 4 nodes in a cluster.

• Support for ext3 file system on RHEL 5 (x86_64) and RHEL 6.

• Support for ext4 file system on RHEL 6 (x86_64).

• Support for Live Application Detach of Serviceguard modular packages for reduced planneddowntime.

• Support for configuration, administration, and monitoring of Oracle Single Instance toolkit onRHEL 5.

• Support for configuration, administration, and monitoring of Network File System (NFS) toolkitpackage.

Overview 5

• Support for configuration, administration, and monitoring of Generic Resources module in aServiceguard package.

• Support for configuration, administration, and monitoring of Linux Contributed Toolkit Suite.

Changed FeaturesThis section provides information about the features modified in the A.11.20.00 release of HPServiceguard for Linux.For Serviceguard:

• Support for Modular Packages:

◦ Performance improvement of modular packages

◦ Online modification for various package parameters that can be configured

For Serviceguard Manager:

• Support for VMware virtual machines as nodes in a cluster (supporting newer version of ESXthat is, 4.1 and 5.0)

Features not Supported in this Version• Red Hat 4

• SUSE Linux Enterprise Server (SLES) 10 and 11

• Red Hat GFS and reiserfs file systems

• More than 4 nodes in a cluster

• XDC

• Xen

• Hardware support X86 (32 bit)

• CLX is not supported with Serviceguard A.11.20.00

About the New FeaturesThe subsections that follow discuss the major new capabilities introduced in the A.11.20.00 release.Information on using these capabilities is in the latest version of Managing HP ServiceguardA.11.20.00 for Linux, available at http://www.hp.com/go/linux-serviceguard-docs.

Integration of Custom Defined Monitor in Serviceguard Using Generic ResourcesThe feature, generic resources is a resource monitoring mechanism in Serviceguard that allowsyou to monitor critical resources for a package through custom monitors, by configuring thesemonitors as user-defined services. It provides integration of custom, user-defined monitors inServiceguard by configuring the generic resource parameters as part of package configuration.For more information, see the latest edition of the Managing HP Serviceguard A.11.20.00 forLinux manual.

Advantages of Generic Resources

• Custom defined monitors can also be integrated.

• Provides better control, options, and flexibility in terms of getting and setting the status of aresource.

• With generic resources, different kind of monitoring mechanism can be used and these canco-exist in a single package.

6

NFS-mounted File SystemsAs of Serviceguard A.11.20.00, NFS-mounted (imported) file systems are supported as sharedstorage in packages. The same package can mount more than one NFS-imported file system, andcan use both cluster-local shared storage and NFS imports.For more information, see “Planning for NFS-mounted File Systems” in Chapter 4 of Managing HPServiceguard A.11.20.00 for Linux manual.

Halting a Node or the Cluster while Keeping Packages Running (Live Application Detach)There may be circumstances in which you want to do maintenance that involves halting a node,or the entire cluster, without halting or failing over the affected packages. Such maintenance mightconsist of anything short of rebooting the node or nodes, but a likely case is networking changesthat will disrupt the heartbeat.New command options in Serviceguard A.11.20.00 (collectively known as Live Application Detach(LAD)) allow you to do this kind of maintenance while keeping the packages running. The packagesare no longer monitored by Serviceguard, but the applications continue to run. Packages in thisstate are called detached packages.This capability applies only to modular failover packages and modular multi-node packages. Formore information, see “Halting a Node or the Cluster while Keeping Packages Running” in chapter7 of Managing HP Serviceguard A.11.20.00 for Linux manual.

Support for Virtualization with ESX as Hypervisor and Guests as Serviceguard NodesHP Serviceguard for Linux is certified for deployment on Linux Virtual machines created on VMwareESX server. On Serviceguard for Linux clusters, virtualization is supported for various configurationsusing physical machine and VMware virtual machines running on ESX server, so as to providehigh availability for applications.

NOTE: Serviceguard is certified on VMware ESX guests, not on ESX hosts, and provides highavailability for applications, not for the virtual machines themselves.

For more information, see the white paper Using Serviceguard for Linux with VMware VirtualMachines, Managing HP Serviceguard A.11.20.00 for Linux manual at http://www.hp.com/go/linux-serviceguard-docs, and HP Serviceguard for Linux Certification Matrix at www.hp.com/info/sglx.

Improved Performance while Starting and Halting a Non-detached Multi-node PackageThe commands cmrunpkg (1m) and cmhaltpkg (1m) have been modified so that you willnow see a performance improvement while starting or halting a non-detached multi-node package.When cmrunpkg (1m) is run on a non-detached multi-node package, the package is startedsimultaneously on all the nodes or the list of the nodes specified, thus reducing the down time instarting the packages. Similarly, when cmhaltpkg (1m) is run on a non-detached multi-nodepackage, the package is halted simultaneously on all the nodes or the list of the nodes specified,thus reducing the down time in halting the packages.This performance improvement is directly dependent on the number of nodes on which the packageneeds to be started or halted; the performance benefit increases as the number of nodes increases.cmrunpkg (1m) and cmhaltpkg (1m) defaults to this behavior.

Deprecated or Obsolete FeaturesThe following features are deprecated or obsolete in this version.cmviewconf is obsolete as of Serviceguard A.11.20.00. Use cmviewcl (1m) to obtaininformation about the cluster.The Cluster Object Manager (COM) is obsolete as of Serviceguard A.11.20.00.

New and Changed Features in This Release 7

Defects Fixed in This VersionThe following table lists the known problems and fixes in this release of Serviceguard for Linux.

Table 1 Fixes in Serviceguard Manager B.03.31.00 for Linux

DescriptionDefect ID

Null pointer on operation log screen when no operation log file exist.QXCR1001204646

Should not allow to proceed without network probing, in cross subnet cluster.QXCR1001214558

If network probing is not selected, heartbeat option is not listing.QXCR1001214565

Not possible to uncheck the node for negative scenario in cross subnet.QXCR1001214572

Roles configured are not displayed on Serviceguard Manager pages.QXCR1001217581

Known Problems and LimitationsThis section provides a list of known problems and limitations as known to HP at the time ofpublication. If workarounds are available, they are included.

Known Problems and Limitations in Serviceguard• QXCR1001207446: cmcld core, cdb_reform_is_hung

• QXCR1001222620: Package halt behavior for modular packages

• QXCR1001225435: Unable to connect to QS at IP address and port 1238

• QXCR1001220269: Apply of iptables rule to open hacl-hb (udp) portfails. iptables rule to open the UDP port hacl-hb fails. Withoutopening the hacl-hb(udp) port by applying the iptables rule, thecluster will not come up on a Red Hat 6 system running with defaultfirewall rules configured.

Work around: The following entry needs to be populated in the/etc/services on the system to which iptables rules have to beconfigured.

hacl-hb 5300/udp # High Availability (HA) Cluster heartbeat

• QXCR1001207290: After rolling upgrade all the nodes to ServiceguardA.11.20.00 completes, any attempt to change the quorum disk to pointto a different physical volume will fail with the following message.

Online quorum disk change is not supported until second cmapplyconfcommand after Rolling Upgrade. cmapplyconf: Unable to apply theconfiguration

Work around: After rolling upgrade all the nodes to Serviceguard 11.20completes,run a cmgetconf and then run cmapplyconf on theconfiguration file obtained from cmgetconf.

Next modify the cluster configuration file, to make quorum diskpoint to new disk and then do an cmapplyconf of the new clusterconfiguration file.

Known Problems and Limitations in Serviceguard Manager• QXCR1001220958: Easy Install script is not showing Serviceguard

Manger-tomcat related errors. User need to open easy install log

8

file to confirm if Serviceguard Manager related products areinstalled successfully.

• QXCR1001222830: Pre-requisite rpms are not enforced duringinstallation of sgmgr-tomcat.Java (1.6 or later) and Tomcat (5.x or6.x) are the pre-requisites for the functioning ofserviceguard-manager-tomcat which is not enforced during theinstallation of serviceguard-manager-tomcat rpm.

Compatibility and Installation RequirementsRead this entire document and any other Release Notes you may have before you begin aninstallation.

NOTE: All nodes must be on the same distribution, including errata level and architecture version.Using different distribution versions within the same cluster is supported only during rolling upgrade.Servers using processors from Intel and AMD can be mixed in a cluster. For supported rollingupgrade paths, see Rolling Software Upgrades (page 14). For information about supportedServiceguard versions, Linux distributions, related software, and hardware, see the documentslisted under “Related Information” (page 22).

Installing HP Serviceguard for LinuxHP provides an automated installation script that takes you through the process of installing andconfiguring Serviceguard for Linux, including Serviceguard Manager and ancillary software. Forinformation and installation instructions, see the latest version of HP Serviceguard for LinuxDeployment Guide, at http://www.hp.com/go/linux-serviceguard-docs.The cmeasyinstall script automates the installation of Serviceguard and Serviceguard Managercomponent rpms that are shipped by HP. This script must be run on each of the node participatingin the cluster. The script checks for the pre-requisite rpms that are required for Serviceguard andServiceguard Manager. If they are present, the component rpms are installed. For more information,see the README_install.txt that is bundled along with the cmeasyinstall script.

Pre-Installation Checklist for Serviceguard

• Before you begin installing Serviceguard, make sure that all components of the cluster havebeen upgraded to their latest firmware versions.

• Serviceguard for Linux depends on the xinetd service. Make sure that the xinetd rpm hasbeen installed from the distribution source (for example, your Linux installation CD) and isenabled.To find out if xinetd is running, use a command such as:ps -ef | grep xinetd

You should see output similar to the following:root 3954 1 0 Jan17? 00:00:05 /usr/sbin/xinetd root 15734 15513 0 13:38 pts/100:00:00 grep xinetd

If xinetd is not running, check that it is installed as follows:rpm -qa xinetd

You can use the chkconfig command to enable xinetd if you need to:/sbin/chkconfig --level 35 xinetd on

• You may need to install additional software from the distribution, depending on whichServiceguard components you are installing. The following lists all the software you will need

Compatibility and Installation Requirements 9

for each distribution if you install all the Serviceguard components. You may want to installall of this software now, to make sure all dependencies are met:

◦ For Red Hat 5 and Red Hat 6:– xinetd

– gcc

– glibc

– libn1

– net-snmp

– lm_sensors

– tog-pegasus

– kernel-headers

– compat-libstdc++-296 (rhel 5)

– compat-libstdc++-33 (rhel 5)

– sg3_utils-libs

– sg3_utils

– libgcc

– kernel-devel

– krb5-libs

– zlib

– e2fsprogs-libs (rhel 5)

– libblkid (rhel 6)

– glibc-devel

– glibc-headers

– libstdc++

– libXp

– libgomp

– authd

Pre-Installation Checklist for Serviceguard Manager1. Ensure Java JDK 1.6 or greater is installed, if not download recent Java JDK for your Linux

distribution and install. Make sure /usr/bin/java points to the JAVA 1.6 or later.2. Ensure tomcat 5.x or tomcat6.x is installed before proceeding the installation of Serviceguard

Manager.

Considerations when Upgrading Serviceguard

• .rhosts

If you relied on .rhosts for access in the previous version of the cluster, you must nowconfigure Access Control Policies for the cluster users. For instructions on how to proceed, see

10

the subsection “Allowing Root Access to an Unconfigured Node” under “Configuring Root-LevelAccess” in Chapter 5 of the Managing Serviceguard for Linux manual.

• cmclnodelist

When you upgrade from an earlier version, Serviceguard converts entries into new entrieswritten into the cluster configuration file during the upgrade, as follows:USER_NAME <user_name>

USER_HOST <host_node>

USER_ROLE Monitor

A wild-card + (plus) is converted as follows:USER_NAME ANY_USER

USER_HOST ANY_SERVICEGUARD_NODE

USER_ROLE Monitor

After you complete the upgrade, use cmgetconf to create and save a copy of the newconfiguration. If you do a cmapplyconf, you want to be sure it applies the newly migratedAccess Control Policies.

Considerations when Installing ServiceguardWhen you install Serviceguard for the first time on a node, the node is not yet part of a cluster,and so there is no Access Control Policy. For instructions on how to proceed, see the subsection“Allowing Root Access to an Unconfigured Node” under “Configuring Root-Level Access” in Chapter5 of the Managing Serviceguard for Linux manual.

Ports RequirementsServiceguard uses the ports listed below. Before installing, make sure that no other program usesthese ports.Ports Needed on Red Hat:

• icmp 8/icmp

• hacl-hb 5300/TCP High Availability (HA) Cluster heartbeat

• hacl-hb 5300/UDP High Availability (HA) Cluster heartbeat

• hacl-cfg 5302/TCP HA Cluster TCP configuration

• hacl-cfg 5302/UDP HA Cluster UDP configuration

• hacl-local 5304/TCP HA Cluster CommandsIf you are using SNMP:• snmp 161/UDP

• snmptrap 162/UDPIf you are using the WBEM provider (used by Serviceguard Manager):• wbem-http TCP/5988

• wbem-https TCP/5989

Compatibility and Installation Requirements 11

If you are using the Quorum Server

• hacl-qs 1238/TCP HA Quorum ServerIf you are using the appserver utility:• hacl-poll 5315/TCP

Ports Needed for AuthenticationThe ports reserved for authentication are also used by Serviceguard:

• auth 113/TCP authentication

• auth 113/UDP authentication

Ports Required by Serviceguard ManagerIf you are using Serviceguard Manager via HP SMH, make sure the following ports are open inaddition to the ports listed above:• compaq-https 2381/tcp

• compaq-https 2381/udp

• cpq-wbem 2301/tcp

• cpq-wbem 2301/udp

• cpq-wbem1188 (tcp/udp)

System FirewallsWhen using a system firewall with Serviceguard for Linux, you must leave open the ports listedabove. For detailed instructions, see Configuring firewall rules for HP Serviceguard on Red HatWhite Paper available at www.hp.com/go/linux-serviceguard-docs—> White papers.Serviceguard also uses some dynamic ports for some cluster services; these also need to be openin the firewall. They are typically in the range 32768-61000 for Red Hat. To determine the rangeon a given system, check the contents of the file /proc/sys/net/ipv4/ip_local_port_range. If you have adjusted the dynamic port range using kernel tunableparameters, alter your firewall rules accordingly.

• To enable intra-cluster communications, each HEARTBEAT_IP network on every node in thecluster must allow the following communications in both directions with all other nodes in thecluster:

◦ TCP on port numbers 5300 and 5302 — and allow only packets with the SYN flag

◦ UDP on port numbers 5300 and 5302

◦ TCP and UDP on dynamic ports

• If you use a quorum server, all nodes in the cluster must allow the following communicationto the quorum server IP address:

◦ TCP on port 1238 — and allow only packets with the SYN flag

Any node providing quorum service for another cluster must allow the following communicationfrom that cluster’s nodes:

◦ TCP on port 1238 — and allow only packets with the SYN flag

• Running the cmscancl command requires the ssh port be open.

12

There are additional firewall requirements to enable execution of Serviceguard commands fromnodes outside the cluster, such as those listed in cmclnodelist. To allow execution of Serviceguardcommands, follow these guidelines:All nodes in the cluster must allow the following communications:• from the remote nodes:

◦ TCP on ports 5302 — and allow only packets with the SYN flag

◦ UDP on port 5302

• to the remote nodes:

◦ TCP and UDP on dynamic ports

The remote nodes must allow the following communications:

• from the cluster nodes

◦ TCP and UDP on dynamic ports

• to the cluster nodes

◦ TCP on ports 5302 — and allow only packets with the SYN flag

◦ UDP on port 5302

Authentication communication must allow the following ports:

• from the cluster nodes:

◦ TCP and UDP on port 113

• to the cluster nodes:

◦ TCP and UDP on port 113

NOTE: If you suspect that the firewall is blocking communications, you can add -j LOG beforethe last line in your iptables file (for example /etc/sysconfig/iptables) to log any blockedports. Consult your Linux distribution’s documentation on firewalls for information on iptables.

Installing the Quorum ServerIf you choose to use a Quorum Server rather than a lock LUN for tie-breaking, install the QuorumServer software on a system outside the cluster. The Quorum Server software is on the Serviceguardfor Linux CD. Updated versions may be available from http://www.hp.com/go/softwaredepot.

CAUTION: If you are upgrading both the Quorum Server and Serviceguard for Linux, upgradethe Quorum Server before you upgrade Serviceguard. This is especially important if you are usingan alternate quorum server subnet. See “Quorum Server Upgrade Required if You Are Using anAlternate Address” (page 22).

For more information about Linux installation and configuration process, seeQuorum Server VersionA.04.00 Release Notes at www.hp.com/go/hpux-serviceguard-docs —> HP Serviceguard QuorumServer Software.For more information about the Quorum Server, see Managing HP Serviceguard A.11.20.00 forLinux manual at www.hp.com/go/linux-serviceguard-docs.

Setting Permissions on the /root/.rhosts FileThe /root/.rhosts file must not allow write access by group or other. If /root/.rhostsfile write permission is enabled for other or group, HP Serviceguard for Linux commands will

Compatibility and Installation Requirements 13

fail, logging a “Permission denied for user” message. This situation can arise when theServiceguard remote security file,$SGCONF/cmclnodelist, is not used and remote node securityis resolved by means of .rhosts.

NOTE: /etc/cmcluster.conf contains the mappings that resolve the symbolic referencesto$SGCONF, $SGROOT, etc., used in these Release Notes. See “Understanding the Location ofServiceguard Files” in Chapter 5 of the Managing HP Serviceguard for Linux manual for details.

For more information about configuring security for prospective cluster nodes, as well as for anexisting cluster, see Chapter 5 of Managing HP Serviceguard for Linux.

Rolling Software UpgradesYou can upgrade the Linux operating system and the Serviceguard software one node at a timewithout bringing down your clusters. This process can also be used any time when a node in thecluster needs to be taken offline for hardware maintenance or patch installations. Until the processof upgrade is complete on all nodes, you cannot change the cluster configuration files, and youwill not be able to use any of the features of the new Serviceguard release.

Requirements

CAUTION:• Special considerations apply to a rolling or non-rolling upgrade to Serviceguard A.11.20.00.

• If you are using a Quorum Server, make sure you read the latest version of HP ServiceguardQuorum Server A.04.00.01 Release Notes available at www.hp.com/go/hpux-serviceguard-docs—>HP Serviceguard Quorum Server SoftwareIf you are using an Quorum Server Upgrade Required if You Are Using an Alternate Address(page 22), you must upgrade the Quorum Server to version A.04.00 before you proceed;see “Quorum Server Upgrade Required if You Are Using an Alternate Address” (page 22).

Rolling upgrade is allowed from SGLX 11.18.xx to SGLX 11.19.xx and from SGLX 11.19xx toSGLX 11.20.00. Rolling upgrade from SGLX 11.18 to SGLX 11.20.00 is not allowed.To upgrade a Linux Serviceguard node to a newer Serviceguard release, you need to make sureof the following:

• The node must be running with a supported version of Linux (Red Hat 5.0, Red Hat 6.0 etc.,)

• The node must be running a supported release of Serviceguard (For example, A.11.19)

NOTE: If the target version of Serviceguard does not support the version of the operating systemcurrently on the node, then you must upgrade the operating system before upgrading Serviceguard.

• The OS upgrade for the node must be from and to the same Linux distribution (For example,RHEL 4 to RHEL 5, not RHEL 5 to SLES 11).

• All members of the cluster must be upgraded to the same version of OS and the Serviceguard.

• Make sure that all applications that run on the current OS will also be supported with the newOS.

• Verify that the new OS supports the current cluster hardware configuration and drivers (networkinterfaces, bonding driver, shared disk storage).

NOTE: If there is no HP bonding driver for the Linux version you are using, use the bondingdriver supplied with the Linux distribution.

• Do a full backup on each node.

14

NOTE: HP recommends you use the rolling upgrade process if possible, because it keeps yourmission-critical applications running; and also that you update the software (rather than do a coldinstall) if possible, because updating preserves the current OS and cluster configuration. But makesure you read the “Limitations ” (page 15) before you start the upgrade. The entire upgrade processmust complete successfully before you can use any Serviceguard commands.

CAUTION: If a failure occurs on one node while you are upgrading another, packages — andthe applications they contain — may not be able fail over to the node being upgraded.

LimitationsThe following limitations apply to rolling upgrades:

• During rolling upgrade, you should issue Serviceguard commands (other than cmrunnodeand cmhaltnode) only on a node containing the latest revision of the software. Performingtasks on a node containing an earlier revision of the software will not work or will causeinconsistent results.

• You cannot modify the cluster or package configuration until the upgrade is complete. Youcannot modify the hardware configuration—including the cluster’s networkconfiguration—during rolling upgrade. This means that you must upgrade all nodes to thenew release before you can modify the configuration file and copy it to all nodes.

• None of the features of the newer release of Serviceguard are allowed until all nodes havebeen upgraded.

• Binary configuration files may be incompatible between releases of Serviceguard. Do notmanually copy configuration files between nodes. The Serviceguard binary configuration file(cmclconfig) must be in the correct directory or the cluster will not behave properly.

• Within a Serviceguard cluster, no more than two versions of Serviceguard can be runningwhile the rolling upgrade is in progress.

• All nodes must be running the same releases of Linux and Serviceguard before the upgrade.

• Rolling upgrades are not intended as a means of using mixed releases of Serviceguard orLinux within the cluster. It is highly recommended that you upgrade all cluster nodes as quicklyas possible to the new release level.

• You cannot delete Serviceguard software (via rpm -e) from a node while the cluster is in theprocess of rolling upgrade.

• This procedure depends on the upgrade or re-install keeping the same device namingconvention and general system configuration. It is possible for devices to change names orbe changed in the scan order in a way that cannot be corrected. If this happens, the clusterneeds to be re-created rather than to be upgraded.

Preparation

IMPORTANT: Make sure there is a supported upgrade path from your current Linux andServiceguard versions to the new versions. See the Serviceguard, Compatibility and Feature Matrixat http://www.hp.com/go/linux-serviceguard-docsThere is no upgrade path between some Linux OS releases. In such cases you will need to do anew OS installation (cold install).

Compatibility and Installation Requirements 15

CAUTION: No package can be in maintenance mode, whether it is running or not, when youperform an upgrade from Serviceguard A.11.19 to any later version, including an upgrade fromthe initial release of A.11.19 to the July 2009 or later patch.This means that:• You must make sure no packages are in maintenance mode when you start the upgrade.

• You must not put any package in maintenance mode until all the nodes have been upgraded.Breaking this rule will leave the cluster in an inconsistent state. To recover, you will need to haltthe cluster and then upgrade all the nodes.For more information about this feature, see “Package Maintenance Mode” section in the latestversion of Managing HP Serviceguard A.11.20.00 for Linux manual.

Do the following before you start:1. Record the hostname and its entire network interface IP addresses. Record each MAC address

of each interface and its network assignment (Example: eth1: HWaddr00:0B:CD:69:F4:68))

2. Record all network information, such as network mask, gateway address, DNS server address,its broadcast address, etc. This information can be useful if you are installing a new OS.

NOTE: Make sure that all your network and storage interfaces are supported by the newOS.

3. Record the storage configuration, such as all LVM information, and if possible, collect a listof hardware disks configured, for example, sfdisk -l.

4. Back up the following files on media that can be easily recovered by the node after its upgradeor a new OS installation:• Host files: /root/.rhosts, /etc/hosts, /etc/profile, and the network

information (including the bonding configurations):

◦ Red Hat: /etc/sysconfig/network-scripts/ifcfg*

5. Make sure you have the latest versions of the software listed under Pre-Installation Checklistand (if you are using or intend to use Serviceguard Manager) “Pre-Installation Checklist forServiceguard Manager”.

• SG files:$SGCONF/*: all current package control and configuration files, including their logfiles.

Procedure — Performing the Rolling Upgrade

NOTE: Warning messages may appear during rolling upgrade while a node is determining whatversion of software is running. This is a normal occurrence and not a cause for concern.

1. Halt the node you want to upgrade (cmhaltnode). This will cause the node's packages tostart up on an adoptive node.

2. Edit the /usr/local/cmcluster/conf/cmcluster.rc file to include the following line:AUTOSTART_CMCLD = 0

If you need to upgrade only Serviceguard, and not the Linux OS, skip to step 6 below.3. Upgrade the node to the new OS, or if necessary, do a complete installation with a new OS.4. After completing the OS upgrade or installation for the node, restore all its previously saved

Host files: /root/.rhosts, /etc/hosts, /etc/profile, /etc/profile, /etc/

16

sysconfig/network/ifcfg*(or /etc/sysconfig/ network-scripts/ifcfg*)and bonding files.Verify that the network configurations are the same prior to the upgrade or the new installation.Check the current interface eth0 has the same corresponding Mac address before and after.

5. Verify that all disks and their filesystems are the same prior to this OS upgrade or newinstallation.Check and compare with the disk layout collected before the upgrade. Use command vgscanto make sure the node with new OS sees all its previously configured LVM disks.

6. Install the new Serviceguard release from the CD.The examples that follow show only the Serviceguard rpm. Generally you also need to install:• cmsnmpd

• sgproviders

• SGMgrPI (B03.31.00). For example, (Serviceguard rpm for Red Hat 5x86_64):rpm -iRedHat6/Serviceguard/x86_64/serviceguard-A.11.20.00-0.rhel6.x86_64.rpm

For Serviceguard upgrade only, you can choose to install with one of the following options:For example,

• (Serviceguard rpm for Red Hat 6 x86_64):For a node not configured into a cluster: rpm -i --force

• For a node not configured into a cluster:rpm -i --forceRedHat6/Serviceguard/x86_64/serviceguard-A.11.20.00-0.rhel6.x86_64.rpm

• For a node already configured into a cluster:rpm -U --replacefilesRedHat6/Serviceguard/x86_64/serviceguard-A.11.20.00-0.rhel6.x86_64.rpm

A message such as the following is expected as a response to the command and doesnot indicate an error since this node is already configured into a cluster:Cannot uninstall Serviceguard while a cluster is configured error:%preun(serviceguard-<old_version> scriptlet failed, exit status1

A message such as the following is expected as a response to the command and does notindicate an error since this node is already configured into a cluster:Cannot uninstall Serviceguard while a cluster is configured error:%preun(serviceguard-<old_version>) scriptlet failed, exit status 1

After the Serviceguard upgrade, you can remove the previous rpm version. First use rpm -qto show any Serviceguard rpm installed:Example:To query any Serviceguard rpm installed:rpm -q serviceguard

To remove (erase) an old version of Serviceguard from the RPM database:rpm -e --noscripts serviceguard-A.11.19.03-0

If you have only upgraded the kernel to a new errata then the deadman must be reinstalled(sg_pidentd or pidentd do not need to be reinstalled on Red Hat 5, SLES 10, or SLES11). Do this after the errata is installed and the system has been rebooted.The deadman driver sources and associated files are located at $SGROOT/drivers.

Compatibility and Installation Requirements 17

7. Ensure that the sg_pidentd rpm is not installed, if present erase the rpm rpm -e --nodepssg_pidentd-3.0.19-2

8. Ensure that the authd rpm is installed. If not, install authd rpm from distro. For example,#rpm -i <authd rpm>

Follow the instructions in the README file in the directory of each driver. If you have installeda new OS version, you must run a convert program. This will convert the binary file(cmclconfig) to its new release format. To run the program on the upgraded node,enter:$SGGSBIN/convert

a. Reboot the node.b. After the node has rebooted, verify the cluster status using cmviewcl, and also verify all

file systems with fsck.c. Restart Serviceguard on this node using cmrunnode. Check that the node joins the cluster

successfully, and if necessary, move packages back onto the node.d. Edit /usr/local/cmcluster/conf/cmcluster.rc file to include the following

line:AUTOSTART_CMCLD = 1

e. Repeat this process for each node in the cluster.

NOTE: Be sure to plan sufficient system capacity to allow moving the packages fromnode to node during the process without an unacceptable loss of performance. If thecluster fails before the rolling upgrade is complete (because of a catastrophic powerfailure, for example), you can restart it by entering the cmruncl command from a nodewhich has been upgraded to the latest revision of the software.

Keeping Kernels ConsistentIf you change kernel parameters as a part of doing a rolling upgrade, be sure to make the samechanges on all nodes that can run the same packages.

ExampleThe following example shows a simple rolling upgrade on two nodes running one package each.

NOTE: This and the following figures are from a use case in which the starting point of theupgrade was Serviceguard A.11.19 and Red Hat 5.7, and the rolling upgrade was to ServiceguardA.11.20.00.

Figure 1 Running Cluster Before Rolling Upgrade

18

Step 1. Halt First NodeHalt the first node, as follows# cmhaltnode -f node1

This will cause Pkg1 to be halted cleanly and moved to node 2. The Serviceguard daemon onnode 1 is halted, and the result is shown in Figure 2.

Figure 2 Running Cluster with Packages Moved to Node 2

Step 2. Upgrade First NodeUpgrade node1new version of Serviceguard as shown in the figure.

Figure 3 Node 1 Upgraded to Serviceguard A.11.20.00

Step 3. Restart Cluster on First NodeWhen upgrade is finished and the node has been rebooted and all data restored, enter the followingcommand on node1 to restart the cluster on node1.# cmrunnode -n node1

At this point, different versions of the Serviceguard daemon (cmcld) are running on the two nodes,as shown in Figure 4.

Figure 4 Node 1 Rejoining the Cluster

Step 4. Repeat the Process on Next NodeRepeat the process on node2. Halt the node as follows:# cmhaltnode -f node2

This causes both packages to move to node1. Then upgrade node2 to new version of ServiceguardA.11.20.00

Figure 5 Running Cluster with Packages Moved to Node 1

Compatibility and Installation Requirements 19

Step 5. Move Package Back to Original NodeMove pkg2 back to its original node. Use the following commands:# cmhaltpkg pkg2

# cmrunpkg -n node2 pkg2

# cmmodpkg -e pkg2

The cmmodpkg command re-enables switching of the package, which is disabled by the cmhaltpkgcommand. The final running cluster is shown in Figure 1-6.

Figure 6 Running Cluster After Upgrades

Rolling Upgrade from A.11.19.X to A.11.20.00The following procedure describes how to perform the Rolling Upgrade from A.11.19.X toA.11.20.00:1. Enable global switching for packages running on Node 1.

The cmmodpkg command enables switching of the package.For example, #cmmodpkg -e pkg1

2. Halt First Node.Halt the node you want to upgrade. This results in the node's packages to start up on anadoptive node. The Serviceguard daemon on node 1 is halted.For example, # cmhaltnode -f node1

3. Uninstall sg_pidentd rpm.

CAUTION: Serviceguard commands will not work unless step 5 is complete. You cannotchange the cluster configuration files until the process of upgrade is complete on all nodes.

For example, #rpm -e --nodeps sg_pidentd-3.0.19-2

4. Install authd rpm from the Linux Distribution DVD/RepositoryFor example, #rpm -i <authd rpm>

5. Upgrade node1.The following examples show only the Serviceguard rpm. Generally, you also need to installthe following:

• cmsnmpd

• sgprovidersFor example, Serviceguard rpm for Red Hat 5 x86_64#rpm –Uvh serviceguard-A.11.20.00-0.rhel5.x86_64.rpm

• SGMgrPI6. Restart Cluster on First Node.

For example, # cmrunnode node1Check that the node joins the cluster successfully, and if necessary, move packages back tothe node.

7. Repeat steps from 2 to 6 for all the nodes.

20

Performing Offline Upgrade from A.11.18.X to A.11.20.00The following procedure describes how to perform offline rolling upgrade from A.11.18.X toA.11.20.00:1. Halt the cluster.

For example, #cmhaltcl -f

2. Select a node you want to upgrade and uninstall pidentd rpm

CAUTION: Serviceguard commands will not work unless step 4 is complete. You cannotchange the cluster configuration files until the process of upgrade is complete on all nodes.

#rpm -e --nodeps pidentd-3.0.19-0

3. Install authd rpm from distro.#rpm -i <authd rpm>

4. Upgrade the node.The following examples shows only the Serviceguard rpm. Generally, you also need to installthe following:

• cmsnmpd

• sgproviderFor example, (Serviceguard rpm for Red Hat 5 x86_64):#rpm –Uvh serviceguard-A.11.20.00-0.rhel6.x86_64.rpm

5. Repeat steps 2 to 4 on each node of the cluster.6. After all nodes are upgraded, restart the cluster.

#cmruncl

Rolling Upgrade in Serviceguard ManagerServiceguard Manager does not support rolling upgrade.1. Ensure Java JDK 1.6 or greater is installed, if not download recent Java JDK for your Linux

distribution and install. Make sure /usr/bin/java points to the JAVA 1.6 or later.2. Ensure tomcat 5.x or tomcat6.x is installed before proceeding the installation of Serviceguard

Manager.3. Rolling upgrade from Serviceguard Manager B.02.xx to Serviceguard Manager B.03.31 is

not supported and users need to uninstall the existing version of the following products:• HPSMH

• HPSMH-Tomcat

• SGMgrPI4. Reinstall the following products from Serviceguard Linux A.11.20.00 bundle:

• HPSMH (hpsmh-7.1.0-16.x86_64.rpm)

• SGMgr-Tomcat (serviceguard-manager-tomcat-01.00.0.redhat.noarch.rpm)

• SGMgrB03.31 (serviceguard-manager-B.03.31.00.redhat.noarch.rpm)

Operating System and Version CompatibilityThis release is specific for Red Hat Enterprise Linux 5 and 6.

Hardware RequirementsSupported hardware architecture is X86_64 (64 bit).

Operating System and Version Compatibility 21

Related InformationThe latest documentation for Serviceguard Linux is available in English at http://www.hp.com/go/linux-serviceguard-docs. Available documents include:

• Managing HP Serviceguard for Linux

• HP Serviceguard for Linux Deployment Guide

• HP Serviceguard for Linux Release Notes

• HP Serviceguard Quorum Server Release Notes available at www.hp.com/go/hpux-serviceguard-docs —> HP Serviceguard Quorum Server Software

• Using Serviceguard for Linux with VMware Virtual Machines White Paper

• Optimizing Failover Time in a Serviceguard Environment White Paper

• HP Serviceguard Toolkit for NFS User Guide

• HP Serviceguard Toolkit for Oracle for Linux User Guide

• HP Serviceguard Contributed Toolkit Suite User Guide

• Distributed Systems Administration Utilities User’s Guide and Release Notes for Linux, underNetwork and Systems Management

• Clustering Linux Servers with the Concurrent Deployment of HP Serviceguard for Linux andRed Hat Global File Systems for RHEL5

• The Serviceguard/Serviceguard Manager Plug-in Compatibility and Feature Matrix

• Securing Serviceguard and other Serviceguard white papers available at www.hp.com/go/hpux-serviceguard-docs HP Serviceguard —> White Papers

• HP Serviceguard for Linux Certification Matrix at www.hp.com/info/sglx. This contains thelatest information about supported hardware and Linux distributions, and other recent changes.

Further InformationSupport tools and information are available from the Hewlett-Packard Support Centre:www.hp.com/go/hpsc

Legacy PackagesIn A.11.20, you can still create new packages using the legacy method (the method used beforeServiceguard A.11.18; see Chapter 7: Configuring a Legacy Package of Managing HPServiceguard for Linux). In some cases new capabilities provided in modular packages (thosecreated by the method introduced in A.11.18) have been added to legacy packages as well.Use the modular method to create new packages whenever possible. See Chapter 6 of ManagingHP Serviceguard for Linux for more information..rhosts deprecated using the .rhosts file as a means of allowing root access to an unconfigurednode is deprecated as of Serviceguard A.11.19. You should use $SGCONF/cmclnodelistinstead. See Chapter 5: Allowing Root Access to an Unconfigured Node of Managing HPServiceguard for Linux for more information.

Quorum Server Upgrade Required if You Are Using an Alternate AddressIf you are using an Alternate Quorum Server Subnet, you must upgrade the Quorum Server toversion A.04.00 before you upgrade the cluster to Serviceguard A.11.20.00

CAUTION: If you fail to do this, the upgraded cluster will be running without a cluster lock untilyou have upgraded the Quorum Server

22

ipnodes Entries Needed in /etc/nsswitch.confipnodes entries are needed in /etc/nsswitch.conf beginning with version A.11.19.Serviceguard uses calls that require ipnodes entries to be configured in /etc/nsswitch.conf.For more information, see Chapter 5:Safeguarding against Loss of Name Resolution Servicessection in the Managing HP Serviceguard for Linux manual available at www.hp.com/go/linux-serviceguard-docs.

High Availability Consulting ServicesBecause Serviceguard configurations can be complex to configure and maintain, HP recommendsthat you use its high availability consulting services to ensure a smooth installation and rollout;contact your HP representative for more information. You should also work with your HPrepresentative to ensure that you have the latest firmware revisions for disk drives, disk controllers,LAN controllers, and other hardware.

Release Notes RevisionsOccasionally, important new information warrants revising the Release Notes after they have goneto press. In such cases HP updates the Release Notes on www.hp.com/go/linux-serviceguard-docs—> Release Notes. Versions with the same part number aredifferentiated by the publication date.

Software Availability in Native LanguagesDocumentation for major releases of Serviceguard and Serviceguard Manager is translated intothe following languages:• Japanese

• Simplified Chinese

ipnodes Entries Needed in /etc/nsswitch.conf 23