[
https://issues.apache.org/jira/browse/FLINK-15173?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16992585#comment-16992585
]
Aljoscha Krettek commented on FLINK-15173:
------------------------------------------
The issue is this code:
https://github.com/apache/flink/blob/f65be9b37f0405b1605ecec533614ff2aab3cd52/flink-runtime/src/main/java/org/apache/flink/runtime/dispatcher/MiniDispatcher.java#L138.
When running in _NORMAL_ mode, the cluster will wait for someone to request the
job result before it shuts down.
Side note, from a client perspective we can run in _ATTACHED_ or _DETACHED_
mode, the runtime calls this _NORMAL_ and _DETACHED_ mode.
> perjob on yarn appcluster shutdown depend on CliFrontend
> --------------------------------------------------------
>
> Key: FLINK-15173
> URL: https://issues.apache.org/jira/browse/FLINK-15173
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Affects Versions: 1.9.1
> Reporter: xiaodao
> Priority: Major
>
> i submit a job by command like:
> ./bin/flink run -m yarn-cluster -yjm 1024m -ytm 1024m -ynm test -yqu c -class
> com.xxx.SocketWordCount /data/tmp/test.jar
> after the job run on yarn, i close the client, and then if the job failed
> with some problem,the yarn application not exit with failed status, but it
> still running and the job in the application failed. just like run on a
> session mode. i check the code , the
> org.apache.flink.runtime.dispatcher.Dispatcher#shutDownCluster whick call by
> CliFrontend.
> so when i close submit client, the cluster not exit. i think it may be a bug,
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
