You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@kafka.apache.org by "ASF GitHub Bot (JIRA)" <ji...@apache.org> on 2016/04/08 02:43:25 UTC

[jira] [Commented] (KAFKA-3528) AbstractCoordinator wakeup during rebalance can cause spurious rebalances

    [ https://issues.apache.org/jira/browse/KAFKA-3528?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15231432#comment-15231432 ] 

ASF GitHub Bot commented on KAFKA-3528:
---------------------------------------

GitHub user hachikuji opened a pull request:

    https://github.com/apache/kafka/pull/1201

    KAFKA-3528: handle wakeups while rebalancing more gracefully

    

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/hachikuji/kafka KAFKA-3528

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/kafka/pull/1201.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #1201
    
----

----


> AbstractCoordinator wakeup during rebalance can cause spurious rebalances
> -------------------------------------------------------------------------
>
>                 Key: KAFKA-3528
>                 URL: https://issues.apache.org/jira/browse/KAFKA-3528
>             Project: Kafka
>          Issue Type: Bug
>          Components: consumer, KafkaConnect
>    Affects Versions: 0.9.0.0, 0.9.0.1
>            Reporter: Jason Gustafson
>            Assignee: Jason Gustafson
>
> If the user calls wakeup() while a rebalance is pending, the join/sync will be left on the wire. Although the requests will eventually complete on a later call to poll(), since we have aborted the call to ensureActiveGroup(), the onJoinComplete() callback will never be invoked. This causes a spurious rebalance on the next poll() since we never received the assignment for the joined generation. This also affects Kafka Connect, which depends internally on the group management protocol and uses wakeup to break from poll() in order to handle requests.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)