Cluster Rolling Upgrade from Windows 2012 R2 to Windows 2016

[AlanM]

Case ID: 02366801

Hi

I'm currently undertaking a MS Cluster rolling upgrade from Windows 2012 R2 to Windows 2016
I've just done the first node but can't get Veeam to work

I removed the old node using Remove-VBServer and this appeared to work OK
I did a rescan of the new node in Veeam once the node was back in the MS Cluster and working OK
Veeam found the node and installed the services etc on it

Now though, the new server is cycling between being available and being listed as unavailable. Re-scanning it doesn't appear to help.
When a backup runs now I receive the following error for any VM on the new host:

Unable to allocate processing resources. Error: On-host proxy [DC02-HV04] requires upgrade before it can be used.

What does this mean? Does Veeam support the rolling upgrade of a cluster? Will this occur until all the nodes are 2016 and I've raised the cluster functional level?
Any ideas on how to fix this?

[ptoro]

Was this an upgrade of OS or clean install?

If OS was "upgraded" meaning Veeam agent was left on server that is now 2016, then I would uninstall Veeam software on the cluster node and then from the console just goto properties of the cluster in question and let veeam detect the nodes again. veeam will notice it is missing software and automatically reinstall it on the upgraded node. But make sure to uninstall the software beforehand.

Again, I am assuming you did an OS "upgrade" vs a clean install on that node.

This happened to me when I did a rolling upgrade (no clean OS install) and didn't uninstall the Veeam software on the node. Also don't forget to upgrade the VM configs after you upgrade all nodes and raise the cluster level. The new checkpoints during backups are great. I used to get a lot of contention on the backend because the whole CSV would pause for VSS snapshots; no more. The new RCT is also a plus (will not occur until you update the VM config and is an offline operation on the VM).

Good luck!

[AlanM]

All good now. Worked through with Veeam Support. Essentially my lack of familiarity with Veeam.

It was a clean install on all the hosts.

In short, the process was - rolling upgrade all hosts to Win 2016 and update the cluster functional level. In Veeam, Backup Infrastructure, right click the cluster/properties and click through the wizard so the cluster is rescanned. This installed the Veeam services on the new nodes. Then go into backup jobs - remove VMs then add them back in - this allows Veeam to associate the VMs with the new nodes. The old nodes disappeared from the cluster after this step. Tested backups - all OK.

[ptoro]

Don't forget to upgrade each VM as well to version 8 so you can take advantage of RCT for backups.