Recovery after rm -rf /var/lib/postgresql/data/pgdata

[godunko-mikhail]

Hi!
I'm new to cnpg, so this question might be weird.
Let's look at the case of a simple cluster from docs

apiVersion: postgresql.cnpg.io/v1
kind: Cluster
metadata:
  name: cluster-example
spec:
  instances: 3
  storage:
    size: 1Gi

I ran rm -rf/var/lib/postgresql/data/pgdata to see how failover works, and as I saw it works well, but the pod was left in an endless restart
So I'm wondering how can the pod be recovered after that?

[gbartolini]

Hi, this is a known behavior and it should be addressed as part of a broader initiative of when a PVC group (how we define a set of volumes in CNPG) is not available anymore for any reason. Superuser escalation is an example (BTW, by default superuser access is not allowed in CNPG). But another one could be an unrepairable failure on a node when you use local disks.

This issue is somehow related to the work we started with #2570, and it will involve determining a time-based value of an instance not being ready after which we give CNPG the permission to delete a PVC Group. However, this is a feature in the self-healing area that requires a lot of thought and care.

What you need to do now is destroy the PVC. You can also use the cnpg destroy command.