[
https://issues.apache.org/jira/browse/MAPREDUCE-3729?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13225334#comment-13225334
]
Robert Joseph Evans commented on MAPREDUCE-3729:
------------------------------------------------
Looking at the code there appear to be a couple of things happening. The code
to create the submitJobDir calls FileSystem.mkdirs, which calls fs.mkdirs, and
then tries to call fs.setPermission. If the first mkdirs fails the result is
ignored and the chmod call throws the exception that we are seeing. There are
lots of reasons why the mkdirs could fail. My guess is that either /tmp is
full or /tmp/hadoop-jenkins/mapred/staging has over 32000 entires in it, which
is the limit for ext3. Why is that my guess, well because when I looked at my
directory I have close to 4000 in mine, and I have not been running nearly as
long or under as much of a load as this one has.
I need to log onto the box to be sure of any of that though.
> Commit build failing TestJobClientGetJob, TestMRWithDistributedCache,
> TestLocalModeWithNewApis
> ----------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3729
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3729
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2, test
> Affects Versions: 0.23.0
> Reporter: Vinod Kumar Vavilapalli
> Assignee: Robert Joseph Evans
> Priority: Blocker
> Fix For: 0.23.1
>
>
> See https://builds.apache.org/job/PreCommit-MAPREDUCE-Build/1671//testReport/.
--
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
