[
https://issues.apache.org/jira/browse/MAPREDUCE-5006?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13584183#comment-13584183
]
Tom White commented on MAPREDUCE-5006:
--------------------------------------
bq. MAPREDUCE-4994 made it so that LocalClientProtocolProvider doesn't always
set the number of map tasks to 1.
Changing the number of map tasks for the local job runner doesn't sound right
to me. It makes the behaviour different to Hadoop 1, which is bad because
users' streaming jobs may behave differently. Also, setting the number of maps
is just a hint. Even with mapreduce.job.maps set to 1, then there will still be
one map per block (see FileInputFormat#getSplits), which is why applications
don't normally set mapreduce.job.maps. So I think we should revert the change
that MAPREDUCE-4994 made here.
> streaming tests failing
> -----------------------
>
> Key: MAPREDUCE-5006
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-5006
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: contrib/streaming
> Affects Versions: 2.0.4-beta
> Reporter: Alejandro Abdelnur
> Assignee: Sandy Ryza
> Attachments: MAPREDUCE-5006.patch
>
>
> The following 2 tests are failing in trunk
> * org.apache.hadoop.streaming.TestStreamReduceNone
> * org.apache.hadoop.streaming.TestStreamXmlRecordReader
--
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
