[
https://issues.apache.org/jira/browse/FLINK-24178?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17413921#comment-17413921
]
frey commented on FLINK-24178:
------------------------------
hi,I recreated the situation,and get the JM and TM logs .from the TM log we can
see the TM is created ,but it can't connect to the RM.
my job will be CREATING status always , and the JM will not apply a new TM for
the job.
and when i delete the TM it call the pod not found.
I had to take the job offline, and increase the parallelism and then bring it
back to its original value to start the job
> Flink on Kubernetes TaskManager
> --------------------------------
>
> Key: FLINK-24178
> URL: https://issues.apache.org/jira/browse/FLINK-24178
> Project: Flink
> Issue Type: Bug
> Components: Client / Job Submission
> Affects Versions: 1.13.2
> Environment: flink version :1.13.2
> kubernetes version : 1.19.3
> Reporter: frey
> Priority: Blocker
> Attachments: image-2021-09-07-13-31-10-077.png,
> image-2021-09-07-13-31-40-796.png, image-2021-09-07-13-31-51-206.png,
> image-2021-09-13-14-05-54-681.png, image-2021-09-13-14-06-05-433.png,
> image-2021-09-13-14-14-11-384.png, jobmanager.log,
> k8s-flink-session-message-01-taskmanager-1-2.log
>
>
>
> when submit a job on kubernetes in native session mode,
> sometimes the TaskManager is created,but we can't find the TaskManager at all.
> eg:
> kubernetes is already created the TaskManager pod,and it's running
> but flink can't find it
> !image-2021-09-13-14-05-54-681.png!
>
> !image-2021-09-13-14-06-05-433.png!
> !image-2021-09-13-14-14-11-384.png!
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
