[
https://issues.apache.org/jira/browse/FLINK-18891?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17176090#comment-17176090
]
Till Rohrmann commented on FLINK-18891:
---------------------------------------
Thanks for reporting this issue [~wooplevip]. Could you provide us with the
debug logs of the cluster?
When deploying a Flink per-job mode cluster in attached mode, then the cluster
won't shut down until the final result of the job execution (in your case that
it has FAILED) has been queried by the client. I could imagine that this is
what happened in your case.
> flink job failed but yarn application still running via yarn-per-job mode
> -------------------------------------------------------------------------
>
> Key: FLINK-18891
> URL: https://issues.apache.org/jira/browse/FLINK-18891
> Project: Flink
> Issue Type: Bug
> Components: Deployment / YARN
> Affects Versions: 1.10.1
> Reporter: Peng
> Priority: Critical
>
> Run a flink stream sql job via yarn-per-job mode, flink-conf.yaml like below:
> {code:java}
> execution.target: yarn-per-job
> restart-strategy: fixed-delay
> restart-strategy.fixed-delay.attempts: 3
> restart-strategy.fixed-delay.delay: 2s
> {code}
>
> After restarting job 3 times, the flink job status is FAILED, but yarn
> application state is still Running. Should the yarn application state be
> Finished when flink job FAILED? Thanks.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
