[
https://issues.apache.org/jira/browse/MAPREDUCE-6066?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14909840#comment-14909840
]
Chen He commented on MAPREDUCE-6066:
------------------------------------
This problem is interesting. I believe there are already many solutions from
academic publications for this problem. Another corner case that we need to be
careful is the case that if AM only get containers from a single NM, then we
should allow speculative tasks run on the same node.
Categorizing node becomes very important. What is the reason that causes this
task (map or reduce) slow. Then, we can make more reasonable decision.
> Speculative attempts should not run on the same node as their original attempt
> ------------------------------------------------------------------------------
>
> Key: MAPREDUCE-6066
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-6066
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: applicationmaster, scheduler
> Affects Versions: 2.5.0, 2.6.0
> Reporter: Todd Lipcon
> Attachments: conf.xml
>
>
> I'm seeing a behavior on trunk with fair scheduler enabled where a
> speculative reduce attempt is getting run on the same node as its original
> attempt. This doesn't make sense -- the main reason for speculative execution
> is to deal with a slow node, so scheduling a second attempt on the same node
> would just make the problem worse if anything.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
