Using hooks with multiple replicas and shared volumes

[s-rogers]

Hi - we’re struggling to get a backup and restore working in the following scenario:

• We have a deployment with multiple replicas that share a single RWX volume
• We are using hooks to call some application specific backup and restore scripts that call APIs to zip up data and store it on the deployment volume

Backing up and restoring the deployment we see the following issues:

• On backup, the hook is called ocne per replica, sequentially, which causes the backup script to run 3 times - we only want it to run once
• On restore, the restore script is called once per replica which causes clashes -a gain we only want to run this once.

What options do we have for enforcing that the hooks only run once, regardless of the number of replicas?

We have looked into creating a separate pod with a PV, just for backup/restore with a single replica which uses the hooks and scripts. This seems to work ok but means our users need to run a pod all the time just in case they want to do a backup so we’d like to avoid this if possible.