vzdump: rbd: sysfs write failed ?

[fips]

Hi,

in the last weeks, I would say since update to 5.1, I receive this behaviour:
Backups from ceph are made to a NFS Storage.
I receive this error, almost daily:

rbd snapshot 'vm-208-disk-2' error: rbd: failed to create snapshot: (17) File exists

So I delete the snapshopt, next time vzdump want to run a backup I receive this error:

INFO: starting new backup job: vzdump 208 --storage Backup --node ceph6 --mode snapshot --compress lzo --remove 0
INFO: Starting Backup of VM 208 (lxc)
INFO: status = running
INFO: CT Name: www.example.at
INFO: found old vzdump snapshot (force removal)
rbd: sysfs write failed
can't unmap rbd volume vm-208-disk-2: rbd: sysfs write failed
INFO: backup mode: snapshot
INFO: ionice priority: 7
INFO: create storage snapshot 'vzdump'
mount: /dev/rbd11 is already mounted or /mnt/vzsnap0 busy
umount: /mnt/vzsnap0/: not mounted
command 'umount -l -d /mnt/vzsnap0/' failed: exit code 32
ERROR: Backup of VM 208 failed - command 'mount -o ro,noload /dev/rbd11 /mnt/vzsnap0//' failed: exit code 32
INFO: Backup job finished with errors
TASK ERROR: job errors

When I move the CT to another node, backup of it can be made. If I move the CT back to the previous node, I get the same error like above.

How can I get rid of this?

 

[fips]

It's a quit big issue, because if the vm which can't be backuped in the beginning, the whole vzdump job on that node stucks.

 

[Eddy Buhler]

Sorry for the necro, but has this issue ever been resolved?

 

[fips]

No.
Workaround: root disks of those VMs are stored on local-zfs
Since this workaround (about Oct 2018), backup to the same NFS store works without any issues.

 

[Eddy Buhler]

Yeah, local storage is not an option for me, unfortunately.