Proxmox Snapshot backup generates error when vm contains more than one disk

[whiggs]

So I have recently run into this issue while attempting to backup all of the vms across my different proxmox hosts: I wrote a simple script that would backup all of my vms by running the below command:
vzdump $vmid --storage backup --compress zstd
However, whenever vzdump attempts to backup a vm that has more than one hard disk attached to it, I see this error appear in the console every single time:

INFO: starting new backup job: vzdump 104 --storage backup --node proxslimvirtuallab --compress zstd
INFO: Starting Backup of VM 104 (qemu)
INFO: Backup started at 2025-11-23 15:39:25
INFO: status = running
INFO: VM Name: PB99976-P03
INFO: include disk 'sata0' 'store:vm-104-disk-1' 60G
INFO: include disk 'sata1' 'store:vm-104-disk-2' 59136M
INFO: include disk 'efidisk0' 'store:vm-104-disk-0' 1M
INFO: backup mode: snapshot
INFO: ionice priority: 7
INFO: creating Proxmox Backup Server archive 'vm/104/2025-11-23T20:39:25Z'
INFO: issuing guest-agent 'fs-freeze' command
ERROR: unable to freeze guest fs - failed to add \\?\Volume{fb70bd6c-3a91-4111-a29d-778121d2a4a9}\ (display name: \\?\Volume{fb70bd6c-3a91-4111-a29d-778121d2a4a9}\) to snapshot set: unknown Windows error 0x8004230e
INFO: issuing guest-agent 'fs-thaw' command
INFO: started backup task 'd942eaea-f7e2-489f-9915-8485b72f95de'
INFO: resuming VM again
INFO: efidisk0: dirty-bitmap status: created new
INFO: sata0: dirty-bitmap status: created new
INFO: sata1: dirty-bitmap status: created new

This error does not occur when the vm only has a single hard disk attached to it.

INFO: starting new backup job: vzdump 104 --node proxslimvirtuallab --storage backup --compress zstd
INFO: Starting Backup of VM 104 (qemu)
INFO: Backup started at 2025-11-23 15:55:26
INFO: status = running
INFO: VM Name: PB99976-P03
INFO: include disk 'sata0' 'store:vm-104-disk-1' 60G
INFO: include disk 'efidisk0' 'store:vm-104-disk-0' 1M
INFO: backup mode: snapshot
INFO: ionice priority: 7
INFO: creating Proxmox Backup Server archive 'vm/104/2025-11-23T20:55:26Z'
INFO: issuing guest-agent 'fs-freeze' command
INFO: issuing guest-agent 'fs-thaw' command
INFO: started backup task '87b64e04-da9f-49bc-aeaf-8cb8f5462441'
INFO: resuming VM again
INFO: efidisk0: dirty-bitmap status: created new
INFO: sata0: dirty-bitmap status: created new

 

[fabian]

the error comes from within the guest, probably from some Windows VSS intregration/hook..

 

[fabian]

having two disks is also a difference inside the guest.. in particular, having multiple disks vs having a single disk often means different code paths taken for ensuring consistency. you can see that in both cases the freeze action is triggered, but only with two disks does it fail -> you need to look inside the guest for the reason why, as that is where the error is happening.

 

[whiggs]

the error comes from within the guest, probably from some Windows VSS intregration/hook..

Ok.... So, is this an acknowledgement that there is an issue with proxmox, or are you trying to claim the issue is entirely within the guest os? I really hope it is not the latter, as I am pretty sure the fact that the error and successful outputs above are of the same vm. The only difference between the two is that the error output has two disks attached:

INFO: VM Name: PB99976-P03
INFO: include disk 'sata0' 'store:vm-104-disk-1' 60G
INFO: include disk 'sata1' 'store:vm-104-disk-2' 59136M
INFO: include disk 'efidisk0' 'store:vm-104-disk-0' 1M

And the successful output is when I ran the exact same command on the exact same vm after removing one of the disks:

INFO: VM Name: PB99976-P03
INFO: include disk 'sata0' 'store:vm-104-disk-1' 60G
INFO: include disk 'efidisk0' 'store:vm-104-disk-0' 1M

 

[fabian]

I will try to provide more background information, I hope you understand my answer better that way..

what PVE does is to instruct the guest OS (via the guest agent, which is a piece of software running inside the guest that is not managed by PVE or written by us!) to freeze the guest file systems. under Windows, this uses a Windows mechanism that is extensible by all kinds of software (e.g., Database solutions like to hook into it, but also storage or backup solutions tend to). this mechanism is choking when your guest has more than a single disk. the next step is finding out why, but that is something you mostly need to do inside the guest (e.g., see if there are more logs there, which services are registered with VSS, how the disks are used/configured inside the guest, ...). it might be that that leads to you finding out that there is something that PVE could do better (e.g., something in how the disks are set up that triggers this broken behaviour inside the guest), but very likely it's a bug in some component inside your guest, and not PVE. in any case there is nothing we can do until you find out more, because so far all we have is symptoms without any actionable data - basically "some Windows VM doesn't support freezing multiple disks via the agent". the error message/code comes from VSS, so I would start there..