Wednesday, February 27, 2008

SCOM 2007 SP1 Upgrade Notes


I upgraded a client's SCOM 2007 infrastructure today from SCOM SP1 RC (build 6246) to SP1 RTM (build 6278).

No real problems encountered, except I should have followed my own #1 rule: Always restart your server before installing a major update. The only issue I ran up against was that the upgrade hung when installing the Management Packs on the Root Management Server (RMS). I reviewed the event logs during the install and found three of these events:
The OpsMgr Config Service service terminated unexpectedly. It has done this 1 time(s). The following corrective action will be taken in 60000 milliseconds: Restart the service.
Followed one minute later with:
The Service Control Manager tried to take a corrective action (Restart the service) after the unexpected termination of the OpsMgr Config Service service, but this action failed with the following error:
An instance of the service is already running.
I'm not sure that these caused the hang, but after I canceled the installation, restarted the RMS server and reinstalled SP1 again, it worked fine with no errors.


My biggest recommendation is to thoroughly read the online version of the SCOM SP1 Upgrade Guide before beginning your upgrade. The online version includes notes that didn't make it into the release notes included in the SP1 package itself. Particularly important are the notes about having to repair all agent installations if you are upgrading from SP1 RC, like I was.

The upgrade path for SP1 is very strict and must be performed in this order:
  1. Prerequisite work (expanding the database and logs), disabling notification subscriptions (why, oh why, can't we do this against multiple subscriptions at once!), and removing pending agent installations.


  2. Upgrade the RMS


  3. Upgrade the Reporting Server


  4. Upgrade stand-alone Management Consoles


  5. Upgrade Management Servers


  6. Upgrade Gateway Servers


  7. Upgrade (or in my case, repair) Agents on managed computers


  8. Upgrade the Audit Collection Service (ACS) server


  9. Reboot the SCOM servers (my suggestion, not required) and re-enable the subscriptions
The entire upgrade took about 3 hours to upgrade nine SCOM servers and 289 managed computers.


1:33 PM