I’ve just gone through this for the first time in production with OpsMgr Failover Clustering and VMM 2008 management in place. It was … interesting to say the least.
First bit of advice … put _everything_ into maintenance mode even if you’re only patching one node:
- Cluster
- All nodes
- VMM server(s)
If you don’t then you will get flooded by alerts from the cluster, cluster resources, agent heartbeat (no avoiding that one I’m afraid – wish MS would sort this old one out) and VMM resources.
Next up, have the Failover Clustering MMC handy. If you patch one node and not the other (you’re prone to do one at a time in production) the your VM’s will refresh and come up with “unsupported cluster configuration”. This means the VM’s become unmanageable in VMM. To patch the nodes they’re on you’ll have to migrate them manually using the Failover Clustering MMC instead of VMM. Once they’re on the “up-level” patched host they’ll go back to green. You can then patch the down-level host. Repeat this for each host in the cluster.
EDIT #1:
Tip: when a VM is sitting in "unsupported cluster configuration" even after all your work is done, you can wait for a VM refresh or force one with a "Repair-Ignore". It should return to green status after the job.
EDIT #2:
I asked MS about VM’s becoming unmanagable when a host node is offline, i.e. "unsupported cluster configuration”. The answer I got from MS was that this issue will be fixed in VMM 2008 R2 and as such, they wouldn’t fix it in the current release.