Where are we at with in-guest clusters and backup?

[JamesNT]

I have a Windows Server 2022 Hyper-V cluster. On top of that cluster, I have an in-guest SQL Server cluster also running Windows Server 2022. Veeam seems to have a tough time backing up the cluster. Further, checkpoints aren't always deleted after backup so we have the occasional avhdx file laying around.

Is having an in-guest cluster just not a good idea or do we have any future outcomes to look forward to?

JamesNT

[david.domask]

Hi JamesNT,

Sorry to hear about the troubles; just to confirm, these are configured as a VHDSet ( VHDS ), correct? These are supported, but there can be some challenges with such configurations

VHDSets can have some challenges when doing image level backups as described in the KB article linked; have you considered using our Plugin for Microsoft SQL? This works without the need for Hyper-V snapshots, preventing the issues mentioned.

You can similarly use Veeam Agent for Windows to protect SQL Clusters, but there are some considerations that need to be checked.

All in all, try the MSSQL Plugin; I think it will likely be the best experience for protecting the actual SQL data.

[JamesNT]

@david.domask,

Yes, we are using VHDS. The error we seem to be getting all the time is:

Error: Failed to collect VM (ID: 9e940354-0b12-452d-88bf-4fffcf84e543) information. Failed to collect VM disks. Failed to get shared disk extents. Failed to get virtual disk (path: 'C:\ClusterStorage\Volume1\SQLCLUSTERFILES\SQLDatabase.vhds') setting data. Job failed ('Failed to get the disk information. Failed to open attachment 'C:\ClusterStorage\Volume1\SQLCLUSTERFILES\SQLDatabase.vhds'. Error: 'The process cannot access the file because it is being used by another process.'.'). Error code: '32774'.
Processing finished with errors at 7/22/2025 8:42:29 PM

Rebooting the cluster members does make that error go away only for it to come back within a few short days.

I have opened a support ticket (07775301) and sent logs; however, the initial response from support clearly shows the person has no idea what an in-guest cluster is. I'll check out the SQL plug-in. That may be the best path forward if the current path proves untenable.

James

[PetrM]

Hello,

@JamesNT It would be really interesting to know your opinion about the plug-in. Please try it and let us know what you think. Also, we are working on centralized management for the SQL plug-in (no ETA), which will allow you to control VDI-based backups of SQL workloads from the VBR console.

Thanks!