[ 
https://issues.apache.org/jira/browse/KAFKA-2713?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14993955#comment-14993955
 ] 

ASF GitHub Bot commented on KAFKA-2713:
---------------------------------------

GitHub user ewencp opened a pull request:

    https://github.com/apache/kafka/pull/443

    KAFKA-2713: Run task start and stop methods in worker threads so they 
execute in parallel and cannot block the herder thread.

    

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

    $ git pull https://github.com/ewencp/kafka 
kafka-2713-task-start-stop-threaded

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

    https://github.com/apache/kafka/pull/443.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 #443
    
----
commit b0aa81f0d32019a4d588daf1e12bd47dd6c95930
Author: Ewen Cheslack-Postava <m...@ewencp.org>
Date:   2015-11-06T16:45:21Z

    KAFKA-2713: Run task start and stop methods in worker threads so they 
execute in parallel and cannot block the herder thread.

----


> Copycat worker should not call connector's/task's start methods in the 
> control thread
> -------------------------------------------------------------------------------------
>
>                 Key: KAFKA-2713
>                 URL: https://issues.apache.org/jira/browse/KAFKA-2713
>             Project: Kafka
>          Issue Type: Sub-task
>          Components: copycat
>            Reporter: Ewen Cheslack-Postava
>            Assignee: Ewen Cheslack-Postava
>             Fix For: 0.9.0.0
>
>
> Currently the DistributedHerder calls start() methods in the same thread as 
> the group membership is handled. This is simple and makes lifecycles easier 
> to reason about, but also means that user code (and even code that can simply 
> block for a long time, like sink task's connectors) can potentially block 
> group membership, which in turn causes the worker to fall out of the group.
> To avoid this, we should run these methods in the worker thread for each 
> connector/task.



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

Reply via email to