You are viewing a plain text version of this content. The canonical link for it is here.
Posted to jira@kafka.apache.org by "Will Perlichek (Jira)" <ji...@apache.org> on 2024/11/19 18:47:00 UTC

[jira] [Commented] (KAFKA-15826) WorkerSinkTask leaks Consumer if plugin start or stop blocks indefinitely

    [ https://issues.apache.org/jira/browse/KAFKA-15826?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17899564#comment-17899564 ] 

Will Perlichek commented on KAFKA-15826:
----------------------------------------

Hi [~gharris1727] do you know if this ticket is still relevant? 

I was looking into flaky test failures in https://issues.apache.org/jira/browse/KAFKA-15891 (I have notes in the comments) and now I am wondering if this could be the root cause of the zombie sink tasks which cause the flaky tests, and should be addressed rather than trying to band-aid the integration tests. 

> WorkerSinkTask leaks Consumer if plugin start or stop blocks indefinitely
> -------------------------------------------------------------------------
>
>                 Key: KAFKA-15826
>                 URL: https://issues.apache.org/jira/browse/KAFKA-15826
>             Project: Kafka
>          Issue Type: Bug
>          Components: connect
>    Affects Versions: 3.6.0
>            Reporter: Greg Harris
>            Priority: Minor
>
> The WorkerSourceTask cancel() method closes the Producer, releasing it's resources. The WorkerSInkTask does not do the same for the Consumer, as it does not override the cancel() method.
> WorkerSinkTask should close the consumer if the task is cancelled, as progress for a cancelled task will be discarded anyway. 



--
This message was sent by Atlassian Jira
(v8.20.10#820010)