[
https://issues.apache.org/jira/browse/KAFKA-8061?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Manikumar updated KAFKA-8061:
-
Description:
In KAFKA-5503, we have added a check
(https://github.com/apache/kafka/pull/5881) for `running` flag in the loop
inside maybeWaitForProducerId. This is to handle concurrent call to Sender
close(), while we attempt to get the ProducerId.
This avoids blocking indefinitely when the producer is shutting down.
This created a corner case, where Sender thread gets blocked, if we had
concurrent producerId reset and call to Sender thread close. The proposed fix
is to check the forceClose flag in the loop inside maybeWaitForProducerId.
was:
In KAFKA-5503, we have added a check
(https://github.com/apache/kafka/pull/5881) for `running` flag in the loop
inside maybeWaitForProducerId. This is to handle concurrent call to Sender
close(), while we attempt to get the ProducerId.
This created a corner case where sender thread gets blocked, if we had
concurrent producerId reset and shutdown call. The proposed fix is to check the
forceClose flag in the loop inside maybeWaitForProducerId.
> Handle ProducerId reset while closing Sender thread.
>
>
> Key: KAFKA-8061
> URL: https://issues.apache.org/jira/browse/KAFKA-8061
> Project: Kafka
> Issue Type: Bug
>Affects Versions: 2.1.1
>Reporter: Manikumar
>Assignee: Manikumar
>Priority: Major
>
> In KAFKA-5503, we have added a check
> (https://github.com/apache/kafka/pull/5881) for `running` flag in the loop
> inside maybeWaitForProducerId. This is to handle concurrent call to Sender
> close(), while we attempt to get the ProducerId.
> This avoids blocking indefinitely when the producer is shutting down.
> This created a corner case, where Sender thread gets blocked, if we had
> concurrent producerId reset and call to Sender thread close. The proposed fix
> is to check the forceClose flag in the loop inside maybeWaitForProducerId.
>
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
