[FLINK-26750] HA cluster cleanup not complete

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 1.14.4, 1.15.0
• Component/s: Runtime / Coordination
• Assignee: Unassigned
• Reporter: Matthias Pohl
• Created: 20/Mar/22 14:57
• Updated: 21/Mar/22 07:51

Description

When starting a Flink cluster with HA (tested with ZooKeeper) in standalone mode and stopping it right away (i.e. ./bin/start-cluster.sh; ./bin/stop-cluster.sh), the paths are not properly cleaned:

$ ls -R /
/
/flink
/zookeeper
/flink/default
/flink/default/jobgraphs
/flink/default/leader
/zookeeper/config
/zookeeper/quota

The /flink path should not be present when stopping the cluster.

This is most likely also a problem in k8s HA. That should be checked as part of this ticket.

Issue Links

• is related to: FLINK-20219 Rethink the HA related ZNodes/ConfigMap clean up for session cluster