[ 
https://issues.apache.org/jira/browse/FLINK-34007?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17807612#comment-17807612
 ] 

Matthias Pohl commented on FLINK-34007:
---------------------------------------

good point, a the holder identity would have to be reset as well.

I'm also wondering whether we should enable 
[isReleaseOnCancel()|https://github.com/fabric8io/kubernetes-client/blob/0f6c696509935a6a86fdb4620caea023d8e680f1/kubernetes-client-api/src/main/java/io/fabric8/kubernetes/client/extended/leaderelection/LeaderElector.java#L124].
 That config property was never used in the kubernetes-client v5.12.4 codebase 
and seems to be disabled by default. We're not enabling it in 
[KubernetesLeaderElector:69|https://github.com/apache/flink/blob/c5808b04fdce9ca0b705b6cc7a64666ab6426875/flink-kubernetes/src/main/java/org/apache/flink/kubernetes/kubeclient/resources/KubernetesLeaderElector.java#L69]
 on the Flink side. Releasing the lock identify might help in certain situation 
where we don't want to wait for the lease to timeout. But that's just a minor 
improvement that I came across while looking through the v6.6.2 code.

[~ZhenqiuHuang] But that finding still doesn't explain that you're seeing the 
same/similar issues with Flink 1.17- deployments. Because the update of the 
kubernetes-client to v6.6.2 only happened in 1.18.0.

> Flink Job stuck in suspend state after losing leadership in HA Mode
> -------------------------------------------------------------------
>
>                 Key: FLINK-34007
>                 URL: https://issues.apache.org/jira/browse/FLINK-34007
>             Project: Flink
>          Issue Type: Bug
>          Components: Runtime / Coordination
>    Affects Versions: 1.16.3, 1.17.2, 1.18.1, 1.18.2
>            Reporter: Zhenqiu Huang
>            Priority: Major
>         Attachments: Debug.log, LeaderElector-Debug.json, job-manager.log
>
>
> The observation is that Job manager goes to suspend state with a failed 
> container not able to register itself to resource manager after timeout.
> JM Log, see attached
>  



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

Reply via email to