[
https://issues.apache.org/jira/browse/MAPREDUCE-3692?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13189504#comment-13189504
]
Vinod Kumar Vavilapalli commented on MAPREDUCE-3692:
----------------------------------------------------
Tom/Eli,
This patch changes the log-level for all transitions in MR AM and resource
manager which to me is a serious concern particularly in the context of
debugging. Granted we do log a little too much, but the correct fix should be
to selectively downgrade so that important transitions like node/container/task
failures are not missed. After the current patch, I think debugging on cluster
becomes very painful.
What I think we should do is to remove the logging from all handle methods, and
selectively log in the individual transitions that we think are important.
Either I can reopen this ticket or you can revert this patch. Sorry for coming
in late.
> yarn-resourcemanager out and log files can get big
> --------------------------------------------------
>
> Key: MAPREDUCE-3692
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3692
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Components: mrv2
> Reporter: Eli Collins
> Assignee: Eli Collins
> Fix For: 0.23.1
>
> Attachments: mapreduce-3692.txt
>
>
> I'm seeing 8gb resourcemanager out files and big log files, seeing lots of
> repeated logs (eg every rpc call or event) looks like we're being too verbose
> in a couple of places.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
