After upgrading to v13, all our Hyper-V backups stopped working. I understand there was a change in the way SCVMM is used in v13. I found out I was unable to deploy the Veeam components to SCVMM because we are using an HA cluster. Trying to deploy the components on the existing SCVMM record didn't work because it's a cluster name, not a server and the admin$ share doesn't exist. I did open a case (08097273) but support didn't seem to understand my issue. I was able to get it working by adding each individual SCVMM cluster node as a general Windows server.
There doesn't appear to be anything in the documentation on how to add an SCVMM cluster to Veeam in v13, so is this the correct way to add an SCVMM cluster?
The only reference I found to an HA SCVMM install is this note about VSA:
[Microsoft Hyper-V only] Linux-based Veeam Software Appliance does not support the SCVMM High Availability feature.
What exactly is the reason for this limitation? Does this mean I can't migrate to VSA in the future?
I clarified internally - User's Guide is accurate - Veeam Software Appliance does not support adding SCVMM clusters, this is due to architectural limitation as of now.
We can't commit to a timeline for adding it, but it's on our radar and may appear in future releases. I must also let you know, we haven't heard such requests before, so I'd love to hear other customers with similar requirements in this thread.
Veeam Backup & Replication for Windows does support SCVMM HA clusters starting v13.0.2, where you don't need to install any components to the cluster nodes directly, instead the special service is installed on the VBR server itself - so that's the way to go for now. (if you decide to go that way, please ask your support engineer about SCVMMUseLocalPSManager registry key - they have that information)
One thing to avoid: please don't add one or several individual SCVMM cluster nodes to the inventory from the same cluster. That setup isn't supported.
You may also consider adding your HV workloads (HA clusters or standalone nodes) directly to the inventory, bypassing SCVMM entirely. That is supported on Veeam Software Appliance.
In such case the only thing you will be missing is support of SCVMM tags in the jobs.
We have many Hyper-V clusters and we need to use SCVMM to track VMs that we migrate between clusters, so adding the clusters directly isn't an option for us. And we have multiple datacenters so support for SCVMM HA is critical. An SCVMM cluster is very little extra effort to set up compared to a standalone SCVMM host; we've been running SCVMM in a clustered setup since 2012 and been using it with Veeam all that time. It's already possible to add a Hyper-V cluster and an SMB File Server Cluster, so it shouldn't be that difficult to support an SCVMM cluster.
I'll inquire about the SCVMMUseLocalPSManager registry key with support. It is working right now without that key at least. Prior to upgrading to v13, we were having intermittent issues with the PSManager.exe becoming unresponsive and having to kill the process to get backups to continue, so I hope that is not an issue if we enable this key.
Thanks for the extra details. I've shared that with our R&D team.
As I mentioned before, adding individual cluster nodes to the inventory isn't supported, and in that case we can't guarantee data integrity. This feature was built for standalone SCVMM hosts only, not for setups where several nodes report the same information to the system from different sources.
For an HA cluster, the orchestration service needs to run on the Veeam Backup & Replication server instead of installing remote components on each node - Veeam support will guide you, several manual steps are required. (reminder: for this to work out of the box, you need to be at least on 13.0.2 version)