You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@aurora.apache.org by "Benjamin Mahler (JIRA)" <ji...@apache.org> on 2015/05/02 00:56:09 UTC
[jira] [Commented] (AURORA-1228) Use explicit status update
acknowledgements to improve driver throughput.
[ https://issues.apache.org/jira/browse/AURORA-1228?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14524228#comment-14524228 ]
Benjamin Mahler commented on AURORA-1228:
-----------------------------------------
Initial review here: https://reviews.apache.org/r/33689/
Was unable to get the end to end tests running, seems to be broken on master.
> Use explicit status update acknowledgements to improve driver throughput.
> -------------------------------------------------------------------------
>
> Key: AURORA-1228
> URL: https://issues.apache.org/jira/browse/AURORA-1228
> Project: Aurora
> Issue Type: Task
> Components: Scheduler
> Reporter: Benjamin Mahler
> Assignee: Benjamin Mahler
>
> Mesos 0.22.0 is released and includes the ability for schedulers to perform explicit acknowledgments of status updates.
> This allows schedulers to process status updates asynchronously from the driver thread. Updates can then be processed in batches to reduce the per-update processing overhead. These should help provide substantial improvements to the schedulers update processing throughput.
> I've placed this under the GC executor retirement effort, per AURORA-1047.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)