Hello,

We use Sanoid to perform ZFS snapshot with the help of pre and post snapshot scripts. The pre snapshot script looks like this (/opt/sanoid-scripts/start-backup.sh):

Then Sanoid creates a snapshot.

The post-snapshot script is (/opt/sanoid-scripts/stop-backup.sh):

The snapshot has been created:

But the backup itself is marked as still running:

It is now impossible to create a new backup:

Nor delete the current one:

Here is the backup description:

PBM version:

Installed via Percona Debian repositories.

The error can be reproduced by creating a script that runs both pre and post script sequentially:

The fix is to follow the optional step (3) described in the documentation and perform a describe backup, then search for statuses and wait if they are not “copyReady” (at PBM and replica set levels) before ending the pre snapshot script (/opt/sanoid-scripts/start-backup.sh):

Is the describe operation really optional? What could explain this broken backup state when all the operations are performed very quickly?

Kind regards,