hadoop-mapreduce-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jonathan Eagles (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (MAPREDUCE-5027) Shuffle does not limit number of outstanding connections
Date Fri, 01 Mar 2013 22:39:15 GMT

    [ https://issues.apache.org/jira/browse/MAPREDUCE-5027?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13591034#comment-13591034
] 

Jonathan Eagles commented on MAPREDUCE-5027:
--------------------------------------------

Thanks for the patch Rob. There seems to be a race condition in the testMaxConnections test.


On my desktop I get two different results consistently.

First: 
Failed tests:   testMaxConnections(org.apache.hadoop.mapred.TestShuffleHandler): expected:<2>
but was:<1>

Second:
Tests in error: 
  testSerializeMeta(org.apache.hadoop.mapred.TestShuffleHandler): test timed out after 2000
milliseconds
  testShuffleMetrics(org.apache.hadoop.mapred.TestShuffleHandler): test timed out after 2000
milliseconds
  testClientClosesConnection(org.apache.hadoop.mapred.TestShuffleHandler): test timed out
after 2000 milliseconds
  testMaxConnections(org.apache.hadoop.mapred.TestShuffleHandler): test timed out after 2000
milliseconds

                
> Shuffle does not limit number of outstanding connections
> --------------------------------------------------------
>
>                 Key: MAPREDUCE-5027
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-5027
>             Project: Hadoop Map/Reduce
>          Issue Type: Bug
>    Affects Versions: 2.0.3-alpha, 0.23.5
>            Reporter: Jason Lowe
>            Assignee: Robert Parker
>         Attachments: MAPREDUCE-5027-2.patch, MAPREDUCE-5027.patch, MAPREDUCE-5027.patch
>
>
> The ShuffleHandler does not have any configurable limits to the number of outstanding
connections allowed.  Therefore a node with many map outputs and many reducers in the cluster
trying to fetch those outputs can exhaust a nodemanager out of file descriptors.

--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

Mime
View raw message