The process for upgrading a clustered RMS from OpsMgr SP1 to (the release candidate of) R2 is a simple and straightforward procedure, so long as you have thoroughly read through the upgrade documentation provided by Microsoft (OpsMgr2007_Upgrade.doc). However, there are a few areas in the Release Candidate (RC) documentation which are a little unclear.
The purpose of this post is to give some guidance and provide additional information to augment the Microsoft documentation and assist you with upgrading a clustered RMS to OpsMgr R2.
Firstly, I must point out that OpsMgr still only supports an Active/Passive cluster configuration. The only exception to this is where you have the RMS running on NODE1, and SQL along with the OperationsManager database running on NODE2. But I digress…
At the time of writing this post, some of the wording and terminology used in the documentation supplied with OpsMgr R2 RC is slightly erroneous. For instance…
a. In the Cluster Administrator pane, right-click the service you want to change (for example, OpsMgr Health Service), and then click Properties.
b. In the Properties dialog box, click Modify.
c. In the Modify Preferred Owner dialog box, under Available nodes, select the node you are performing the upgrade on, make sure it is the only node listed under Preferred owners, and then click OK.
Step 6 on Page 16 of OpsMgr2007_Upgrade.doc
OK. I realize I’m being pedantic and the error isn’t obvious but the term “Modify Preferred Owner” is misleading. If you gloss over the instructions, you could be forgiven for setting the Preferred Owners on the cluster group rather than setting the Possible Owners for the cluster resource. This seems innocuous enough, however the ramifications of setting this on the Cluster Group are potentially catastrophic (namely, ALL clustered RMS resources will failover to the passive node during the upgrade which can cause irreversible damage to the RMS).
As an amendment to the upgrade documentation provided by Microsoft, I recommend, instead of following step 6 and its subsequent sub-steps (detailed above), you simply open Cluster Administrator, right click the Passive cluster node and select Pause Node. This will temporarily prevent the Passive node from participating in a cluster failover, ensuring that the upgrade is performed correctly and (hopefully) without incident.
Another noteworthy point is that when you upgrade the first node in the cluster, make sure to select the ‘Database Upgrade’ checkbox. This will not only upgrade the RMS and the relevant services and configuration, but also upgrade the database.
As I stated at the beginning of this article, Microsoft have made the upgrade path from OpsMgr SP1 to R2 a very simple and clear-cut procedure. Just be sure to read the manuals thoroughly and familiarize yourself with the process.
Read the complete post at http://www.1e.com/1EBlogs/post/2009/04/23/Upgrading-a-Clustered-RMS-from-OpsMgr-SP1-to-R2.aspx