[
https://issues.apache.org/jira/browse/MAPREDUCE-6697?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Nathan Roberts updated MAPREDUCE-6697:
--------------------------------------
Attachment: MAPREDUCE-6697-v1.patch
Attaching patch. Ideally it wouldn't call getTotalMaps/Reduces() on each
heartbeat but it keeps the patch small and it looks to be a pretty cheap call.
> Concurrent task limits should only be applied when necessary
> ------------------------------------------------------------
>
> Key: MAPREDUCE-6697
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-6697
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2
> Affects Versions: 2.7.0
> Reporter: Jason Lowe
> Assignee: Nathan Roberts
> Attachments: MAPREDUCE-6697-v1.patch
>
>
> The concurrent task limit feature should only adjust the ANY portion of the
> AM heartbeat ask when a limit is truly necessary, otherwise extraneous
> containers could be allocated by the RM to the AM adding some overhead to
> both. Specifying a concurrent task limit that is beyond the total number of
> tasks in the job should be the same as asking for no limit.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
