CIT 470: Advanced Network and System Administration

CIT 470: Advanced Network and System Administration Slide #1

CIT 470: Advanced Network and System Administration

Change and Configuration Management

CIT 470: Advanced Network and System Administration Slide #2

Topics

1. Change Management

2. Change Processes

3. Revision Control

4. Configuration Management

5. cfengine

Images from Pro Git

CIT 470: Advanced Network and System Administration Slide #3

Change Management

Effective planning and implementation of changes to systems.

Changes should be1. Well documented.

2. Have a backout plan.

3. Reproducible.

CIT 470: Advanced Network and System Administration Slide #4

Why do we need Change Management?

March 26-29, 2006: BART trains halted to avoid running into each other when computer systems crashed.• Crashes on Monday/Tuesday resulted from

software maintenance upgrades.• Crash on Wednesday resulted from installing a

backup system to avoid future crashes.• Thousands of passengers stranded for several

hours each time.

CIT 470: Advanced Network and System Administration Slide #5

Change Management

1. Plan change.

2. Test change on single system.

3. Test change on multiple systems.

4. File a change request.

5. Change committee approves request.

6. Schedule change.

7. Communication with users/admins.

8. Change systems at scheduled time.

9. Post-event analysis.

CIT 470: Advanced Network and System Administration Slide #6

Testing Changes

• Automated checks.– Sanity checks like Samba testparm.– Reboot system.

• Test on one system first.

• Then test on set of systems.– Dedicated test systems.– System admin workstations.– Virtual machines.

CIT 470: Advanced Network and System Administration Slide #7

When do you need a Change Proposal?

Does the change impact critical services?

Critical machines/services– Business critical: e-commerce server, etc.– Essential services: routers, DNS, NFS, auth.

Non-critical machines/services– Individual desktops– Internal news web server

CIT 470: Advanced Network and System Administration Slide #8

Change Proposal

1. Description of the change.

2. Systems impacted by change.

3. Why the change is being made.

4. Risks presented by the change.

5. Test procedure.

6. Backout plans.

7. How long the change will require.

CIT 470: Advanced Network and System Administration Slide #9

CommunicationCommunicate change to impacted people.

– What change is being made (nontechnical.)

– Which services will be unavailable.

– When and how long will they be unavailable.

– What actions do they need to task (if any.)

Communication issues– If you send too many notes, they’ll be ignored.

– Send notices only to those impacted.

– Push critical notices; use pull for non-critical.

CIT 470: Advanced Network and System Administration Slide #10

Scheduling

Scope When Notification Type

Routine Single host or user.

Anytime. Personal.

Major Many hosts or users.

Off-peak Push.

Sensitive None but major impact on failure.

Off-peak. Pull.

CIT 470: Advanced Network and System Administration Slide #11

Change Freezes

Time when only minor updates can be done.– End of quarter or year.– “Crunch time” for projects.

CIT 470: Advanced Network and System Administration Slide #12

Backing Out

Decide back-out conditions before downtime– Avoid the “just 5 more minutes” problem.

– Be sure that someone is keeping track of time.

Questions:– How much time is required for back out?

– When is the latest time you can successfully back out?

– Will backing out this change prevent other changes from being committed?

CIT 470: Advanced Network and System Administration Slide #13

Backing Out: How to do it?

Service-level changesUse revision control system to revert config.

Restart service.

Machine-level changesSoft cutover: Old service is still running.

Hard cutover: Power up old server or restore from backups.

IssuesData migration.

Compatibility.

CIT 470: Advanced Network and System Administration Slide #14

Automatic Checks

Check integrity of critical files before use.– Some services provide checks: LDAP, SMB.– Check startup files by rebooting machine.– Write your own checks for other files.

• Most people only do this after they have a problem.

CIT 470: Advanced Network and System Administration Slide #15

Revision ControlRevision control systems provide

Conflict management: prevents multiple people from modifying file at once and corrupting it.

Change history: records who modified the file when and why the change was made.

Revision control paradigmsLock-Modify-Unlock: rcs

Copy-Modify-Merge: cvs, subversion, etc.

Distributed: darcs, git, mercurial

Local Version Control

CIT 470: Advanced Network and System Administration Slide #16

Centralized Version Control

CIT 470: Advanced Network and System Administration Slide #17

Distributed Version Control

CIT 470: Advanced Network and System Administration Slide #18

Local Git Operations

CIT 470: Advanced Network and System Administration Slide #19

Git File Lifecycle

CIT 470: Advanced Network and System Administration Slide #20

Gitk history visualizer

CIT 470: Advanced Network and System Administration Slide #21

CIT 470: Advanced Network and System Administration Slide #22

CIT 470: Advanced Network and System Administration Slide #23

References1. Mark Burgess, Principles of Network and System

Administration, 2nd edition, Wiley, 2004.2. Aeleen Frisch, Essential System Administration, 3rd

edition, O’Reilly, 2002.3. Thomas A. Limoncelli and Christine Hogan, The Practice

of System and Network Administration, Addison-Wesley, 2002.

4. Evi Nemeth et al, UNIX System Administration Handbook, 3rd edition, Prentice Hall, 2001.

5. Todd R. Weiss, “IT upgrades slow BART trains in San Francisco,” http://www.computerworld.com/printthis/2006/0,4814,110107,00.html, ComputerWorld, March 31, 2006.