Backup fails to create RBD snapshot

[aarononeal]

Using the "Backup Now" button on an unconfined LXC container with RBD root disk, the task hangs forever waiting on a snapshot to be created.

INFO: starting new backup job: vzdump 753 --mode snapshot --node node01 --compress zstd --storage cephfs_storage --remove 0
INFO: filesystem type on dumpdir is 'ceph' -using /var/tmp/vzdumptmp3365968_753 for temporary files
INFO: Starting Backup of VM 753 (lxc)
INFO: Backup started at 2022-01-25 08:54:41
INFO: status = running
INFO: CT Name: container753
INFO: including mount point rootfs ('/') in backup
INFO: excluding bind mount point mp0 ('/usr/lib/modules') from backup (not a volume)
INFO: backup mode: snapshot
INFO: ionice priority: 7
INFO: create storage snapshot 'vzdump'

Stopping the task results in the following.

fsthaw '/proc/3338000/root//' failed - Invalid argument
interrupted by signal
snapshot create failed: starting cleanup
trying to acquire lock...
OK
no lock found trying to remove 'backup' lock
ERROR: Backup of VM 753 failed - interrupted by signal
INFO: Failed at 2022-01-25 08:55:11
INFO: Backup job finished with errors
TASK ERROR: job errors

I can use `rbd snap create` from the terminal and the snap of the image is taken in seconds.

Shutting down the VM takes a successful backup, but that's because it's not using the snapshot feature.

How can I troubleshoot?

 

[aarononeal]

I have determined backup works until I install snapd inside the container.

apt install libsquashfuse0 squashfuse fuse snapd

I don't know yet why it hangs the backup process.

 

[fiona]

Hi,
please see here.

 

[aarononeal]

Thanks @Fabian_E. So, many Linux installs these days rely on snap, which relies on fuse. Sounds like you're saying Proxmox just won't be able to backup Canonical containers.

 

[fiona]

Using stop mode backups should be a workaround, but yes, the real solution would be to get fsfreeze and fuse to interact nicely.