When Changing to allowUnsafeConfigurations: true cluster goes to failuresand ongos doesnot get to Ready state

MongoDB Percona Operator for MongoDB
,
1

Using the Helm Chart for PSMDB
Chart Version: 1.12.4
Operator Version: 1.12.0
When changing to (was false) allowUnsafeConfigurations: true
The cluster starts to reinitialize the replicasets pods and eventually the mongos cannot get to Ready State.
Moreover, it seems that the replicaset could not get one of the nodes as Primary.

To simulate:
Just run any CR with allowUnsafeConfigurations: false and then change to allowUnsafeConfigurations: true.

btw, I think pertains to another issue with pause: true which also means changing to allowUnsafeConfigurations: true and then the cluster fails when coming back with pause: false

2 birds in one bug

1 Like
2

Hello @Yossi_Cohn ,

sorry for not responding sooner - the topic is created in a non-operator thread and was ignored.
Yeah, I was able to reproduce it unfortunately. Also I see you created the ticket for it here [K8SPSMDB-786] When Changing to allowUnsafeConfigurations: true cluster goes to failuresand ongos doesnot get to Ready state - Percona JIRA

The dirty fix here would be to delete this mongos pod. It will then get into running state automatically.
We will check with the team what might be the issue and update the JIRA ticket.

1 Like