Backing up guest fileserver clusters (with VHD Sets) on Server 2016

[stevemciow]

I know there have been threads in the past about this (including microsoft-hyper-v-f25/backing-up-a-wind ... 83-90.html, but I've not yet seen a working solution.

Our set up is as follows:

• Windows Server 2016 Storage Spaces Direct Hyper-converged Cluster
• 2 x Guests, running Windows Server 2016, with the following setup:
  • 2 x shared VHD Sets, one as guest-cluster witness, the other as a data drive
  • Running as a fail-over file server cluster (with the 'general purpose file server' configuration)
• Veeam Backup and Replication 9.5u4 installed

Having followed all the guidelines in https://www.veeam.com/kb2709 I'm down to one remaining error - "Error code: '32775'. More than one VM claimed to be the owner of shared VHDX in VM group 'Hyper-V Collection'". This occurs because, in the configuration above, only the active file server VM will ever have the VHDs online.

I have found an method that can (manually) successfully create checkpoints/snapshots of the cluster, using PowerShell as follows:

1. First create a VM Group, and place the guest VMs into it:
   

   Code: Select all

   New-VMGroup - Name TestClusterGroup -GroupType VMCollectionType
   $vm1 = Get-VM -Name "Test-FS01"
   $vm2 = Get-VM - Name "Test-FS02"
   Add-VMGroupMember -VMGroup TestClusterGroup -VM $vm1,$vm2
   

2. Find the CollectionID of the new VMGroup:

   Code: Select all

   $collectionID = get-wmiobject -Namespace Root\Virtualization\v2 Msvm_VirtualSystemCollection | where {$_.ElementName -eq "TestClusterGroup"} | select CollectionID

3. Create an instance of the WMI object for managing Collection Snapshots:
   

   Code: Select all

   $collSnapSvc = Get-WmiObject -Namespace Root\Virtualization\v2 Msvm_CollectionSnapshotService
   
   [*]Get the WMI object that represents the VMGroup:
   [code]$VmColl = Get-WmiObject -Namespace Root\Virtualization\v2 Msvm_VirtualSystemCollection|? "CollectionID" -eq $collectionID.CollectionID

4. Create a checkpoint/snapshot:

   Code: Select all

   $collSnapSvc.CreateSnapshot($vmColl,$null,32768)

To remove the checkpoint/snapshot, I ran the following:

• Obtain a list of the checkpoint collections that exist (not sure how you're meant to identify which one is the one you want - in my case there was only one):

  Code: Select all

  get-wmiobject -namespace Root\Virtualization\v2 Msvm_SnapshotCollection]
  [*]Get the WMI object that represents the snapshot/checkpoint collection (use the "CollectionID from the previous step):
  [code]$vmColl = get-wmiobject -namespace Root\Virtualization\v2 Msvm_SnapshotCollection|? "CollectionID" -eq "821A8C49-BED5-400F-90D0-F588003D3691"

• Remove the checkpoint/snapshot (this does seem to merge successfully, even with both VMs running - removing the files from the location of the VHD Set):

  Code: Select all

  $collSnapSvc.DestroySnapshot($vmColl)

I've got a case open with support at the moment (#04551335), but was wondering if anyone else has actually got host-level backups of guest cluster working? (specifically with general-purpose fail-over file servers).

Thanks!

[HannesK]

Hello,
a little bit off-topic, but did you think about treating these pseudo-virtual machines like physical machines and simply use the Veeam Agent for Windows?

As for the virtual backup... yes, please continue with support. Guessing for reasons in that kind of setup is complicated.

Best regards,
Hannes

[maxwell.correia]

hello stevemciow. we also have exactly the same problem here in the company (and our two VMs also run a fileserver cluster, like you). we opened case # 04568418 with support and we are waiting for a solution.