[
https://issues.apache.org/jira/browse/FLINK-21264?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17278692#comment-17278692
]
Yang Wang commented on FLINK-21264:
-----------------------------------
This is not a bug. Please note that you are submitting the Flink per-job
cluster in attached mode, only when the result is retrieved by client. Then the
application will be deregistered from Yarn.
You could use the following code to start a detached per-job cluster.
{code:java}
./bin/flink run -d -m yarn-cluster -yjm 2048 -ytm 8086
./examples/streaming/SocketWindowWordCount.jar --hostname ip --port 9998
{code}
> at yarn-per-job mode, tm stopped ,the jm dont relase resource,its also running
> ------------------------------------------------------------------------------
>
> Key: FLINK-21264
> URL: https://issues.apache.org/jira/browse/FLINK-21264
> Project: Flink
> Issue Type: Bug
> Components: Deployment / YARN
> Affects Versions: 1.12.0
> Reporter: tonychan
> Priority: Blocker
> Attachments: Flink任务结束后Yarn上仍然显示RUNNING.docx,
> image-2021-02-03-17-42-53-269.png, image-2021-02-03-17-43-28-017.png
>
>
> i make a socket (nc -lk 9999) for flink test, and at first the flink running
> is ok,when the source server stopped, the flink tm is stopped and the status
> is finish ,but the jm is also running, !image-2021-02-03-17-42-53-269.png!
>
> !image-2021-02-03-17-43-28-017.png!
>
>
>
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
