Backups not working with fleecing and snapshot-as-volume-chain

[joeplaa]

I have an iSCSI LUN with LVM storage layered on top. I'm not sure if this is the best configuration, but I enabled fleecing to the same storage and snapshot-as-volume-chain. Fleecing used to work fine before I turned on the volume chain function and converted the .raw disks to .qcow2. Now it gives an error, see below.

Am I running into a bug, am I missing some specific config option or is my setup plain wrong? Does fleecing have any merit with snapshot-as-volume-chain?

INFO: starting new backup job: vzdump 1002 --notification-mode notification-system --node jpl-proxmox1 --notes-template '{{guestname}}' --storage jpl-pbs-nfs --mode snapshot --remove 0
INFO: Starting Backup of VM 1002 (qemu)
INFO: Backup started at 2025-11-24 16:02:07
INFO: status = running
INFO: VM Name: jpl-technitium1
INFO: include disk 'scsi0' 'starwind-lvm-vm:vm-1002-disk-0.qcow2' 8G
INFO: backup mode: snapshot
INFO: ionice priority: 7
INFO: special config section found (not included into backup)
INFO: creating Proxmox Backup Server archive 'vm/1002/2025-11-24T16:02:07Z'
INFO: removing (old) fleecing image 'starwind-lvm-vm:vm-1002-fleece-0'
WARN: error removing fleecing image 'starwind-lvm-vm:vm-1002-fleece-0' - lvremove 'sw-vg/vm-1002-fleece-0' error:   Failed to find logical volume "sw-vg/vm-1002-fleece-0"
  Rounding up size to full physical extent 8.00 GiB
  Logical volume "vm-1002-fleece-0" created.
Formatting '/dev/sw-vg/vm-1002-fleece-0', fmt=qcow2 cluster_size=65536 extended_l2=off preallocation=metadata compression_type=zlib size=8589934592 lazy_refcounts=off refcount_bits=16
INFO: drive-scsi0: attaching fleecing image starwind-lvm-vm:vm-1002-fleece-0 to QEMU
ERROR: drive 'scsi0' - volume 'starwind-lvm-vm:vm-1002-fleece-0' - 'format=qcow2' option different from storage format 'raw'
INFO: aborting backup job
INFO: resuming VM again
ERROR: Backup of VM 1002 failed - drive 'scsi0' - volume 'starwind-lvm-vm:vm-1002-fleece-0' - 'format=qcow2' option different from storage format 'raw'
INFO: Failed at 2025-11-24 16:02:09
INFO: Backup job finished with errors
INFO: notified via target `mail-to-root`
TASK ERROR: job errors

/etc/pve/storage.cfg

lvm: starwind-lvm-vm
        vgname sw-vg
        content images,rootdir
        saferemove 0
        shared 1
        snapshot-as-volume-chain 1

 

[bbgeek17]

It appears this may not be a permissible combination of options.

In general, you’d want to place your fleecing blocks on a different storage system than the one being backed up, though the ideal setup depends on the capabilities of your storage.

In this scenario, any new disks created on a “snapshot-as-volume-chain” storage will be in QCOW2 format, but the fleecing mechanism currently expects RAW.

It could simply be a combination that hasn’t been addressed and tested yet, so you may need to wait for the developers to review and implement it.

---

Blockbridge : Ultra low latency all-NVME shared storage for Proxmox - https://www.blockbridge.com/proxmox

 

[joeplaa]

@bbgeek17 Yeah, I was thinking the same.

I'm testing new storage (nvme backed), so it's fast enough to either have fleecing on the same storage or disable it altogether. I disabled fleecing and now backups work fine.

Would love to learn more about when and how to use these two features though. The current documentation is very limited.