Postgres-operator: Cluster stuck in no leader situation

[thepotatocannon]

Hi, we met some issues earlier this week. We have a test cluster made of 3 pods. After the weekend all of pods were demoted to replicas. Unfortunately, we don't have logs saved, but all nodes claimed that they failed acquiring lock and follow another leader. We cannot tell what caused the issue.
Restarting pods resolved the problem and cluster started working fine.
Does anybody know what could be the reason and also if it is healthy state that cluster cannot elect the master and demotes every node to replica?

[CyberDem0n]

Without logs it is not possible to tell what was wrong.

[CyberDem0n]

At least it would be nice if you specify what spilo version you run

[thepotatocannon]

At least it would be nice if you specify what spilo version you run

It is:

spec:
  dockerImage: registry.opensource.zalan.do/acid/spilo-cdp-12:1.6-p16

[CyberDem0n]

I see...
There is one suspect: https://github.com/zalando/patroni/pull/1366/files, but no new image yet released

[thepotatocannon]

Thanks a lot for the answer.
Any insights when the new image is being released on zalando registry?

[dejwsz]

maybe this is related somehow? https://github.com/zalando/postgres-operator/issues/855