[
https://issues.apache.org/jira/browse/MAPREDUCE-5848?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13987047#comment-13987047
]
Subramaniam Krishnan commented on MAPREDUCE-5848:
-------------------------------------------------
[~jlowe], thanks for pointing out that the patch is necessary but not
sufficient for the AM to make the decision about tasks being KILLED rather than
FAILED.
We have been running under heavy loads with pre-emption enabled using GridMix
and the patch has significantly reduced the number of tasks reported
incorrectly as FAILED, in fact we haven't seen the SIGTERM issue at all. So we
were wondering if we can have a separate JIRA to track the handling of SIGTERM
scenario while this patch gets pushed as this is definitely required?
> MapReduce counts forcibly preempted containers as FAILED
> --------------------------------------------------------
>
> Key: MAPREDUCE-5848
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-5848
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Affects Versions: 2.1.0-beta
> Reporter: Carlo Curino
> Assignee: Subramaniam Krishnan
> Attachments: YARN-1958.patch
>
>
> The MapReduce AM is considering a forcibly preempted container as FAILED,
> while I think it should be considered as KILLED (i.e., not count against the
> maximum number of failures).
--
This message was sent by Atlassian JIRA
(v6.2#6252)
