[FLINK-30004] Cannot resume deployment after suspend with savepoint due to leftover configmaps

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: kubernetes-operator-1.2.0
• Fix Version/s: kubernetes-operator-1.3.0
• Component/s: Kubernetes Operator
• Labels: pull-request-available
• Assignee: Thomas Weise
• Reporter: Thomas Weise
• Created: 12/Nov/22 17:57
• Updated: 16/Nov/22 01:26
• Resolved: 16/Nov/22 01:26

Description

Due to the possibility of incomplete cleanup of HA data in Flink 1.14, the deployment can get into a limbo state that requires manual intervention after suspend with savepoint. If the config maps are not cleaned up the resumed job will be considered finished and the operator recognize the JM deployment as missing. Due to check for HA data which are now cleaned up, the job fails to start and manual redeployment with initial savepoint is necessary.

This can be avoided by removing any leftover HA config maps after the job has successfully stopped with savepoint (upgrade mode savepoint).

Issue Links

• links to: GitHub Pull Request #440