[
https://issues.apache.org/jira/browse/FLINK-20649?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17253655#comment-17253655
]
Kezhu Wang commented on FLINK-20649:
------------------------------------
I think I get your point. I was kind of mess up the symptom with your
suggestion. Do you mean a misbehaved or buggy {{SourceOperator.close}} could
cause {{StreamTask}} stuck when {{StreamTask}} is about to terminate due to
internal reason(i.e. not cancelled) ? I don't think flink will protect against
this situation, even if {{SourceOperator.close}} is a terminal operation. You
could cancel that job if you run into this situation.
Last, above is all my personal thought, you could ping committer for a try.
> StreamTask closeAllOperators stuck
> ----------------------------------
>
> Key: FLINK-20649
> URL: https://issues.apache.org/jira/browse/FLINK-20649
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Task
> Affects Versions: 1.12.0
> Reporter: weiyunqing
> Priority: Major
> Attachments: 924f9c5510dd45408f41280c3f3ea48bcc284573.png,
> image-2020-12-17-17-57-41-680.png
>
>
> When a StreamTask exits due to an internal exception, the StreamTask does not
> use an asynchronous thread to closeAllOperators or set a timeout. The
> StreamTask will be stuck here (no upstream and downstream scenarios)
> !image-2020-12-17-17-57-41-680.png!
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
