You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@flink.apache.org by "liuzhuo (Jira)" <ji...@apache.org> on 2023/01/28 10:05:00 UTC

[jira] [Created] (FLINK-30813) Residual zk data when using the kubernetes session mode

liuzhuo created FLINK-30813:
-------------------------------

             Summary: Residual zk data when using the kubernetes session mode
                 Key: FLINK-30813
                 URL: https://issues.apache.org/jira/browse/FLINK-30813
             Project: Flink
          Issue Type: Bug
          Components: Deployment / Kubernetes
            Reporter: liuzhuo


    If we use kubernetes session mode and use Zookeeper(ZK) as the HA service, the HA data on ZK is not cleared after the session stops.

    Because  when deleting a session, only call this method:
{code:java}
kubernetesClusterDescriptor.killCluster(clusterId);
{code}
    However, this method only deletes the deployment associated with the clusterId. If ZK is used as the HA service, data on ZK will be left over when the HA stops, resulting in more and more data on zk.

 

Maybe we need to add
{code:java}
ClusterClient#shutDownCluster(){code}
or
{code:java}
HighAvailabilityServices#closeAndCleanupAllData(){code}
When using session mode (native kubernetes)



--
This message was sent by Atlassian Jira
(v8.20.10#820010)