search

spotify / flink-on-k8s-operator

Kubernetes operator for managing the lifecycle of Apache Flink and Beam applications.
Apache License 2.0
188 stars 68 forks source link

flink operator not clean up zookeeper nodes #828

Open ZHOUSH41 opened 1 month ago

ZHOUSH41 commented 1 month ago

Hello. We are creating clusters with zookeeper high availability enabled. We set this properties on the cluster:

high-availability.cluster-id: ....
high-availability.jobmanager.port: .....
high-availability.storageDir: ....
high-availability.type: zookeeper
high-availability.zookeeper.quorum: ....

when we want to delete the cluster because we want to recreate it with a new property or new code, not all resources created by the HA (like the ZNode) are being properly cleaned up, which means that when we relaunch the cluster, the old job will continue to exists, competing for resources with the new job being launched