[
https://issues.apache.org/jira/browse/HADOOP-3384?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12596812#action_12596812
]
Rick Cox commented on HADOOP-3384:
----------------------------------
This appears to be a duplicate of HADOOP-3089, which will be in 0.17.
> streaming process hang when no input to task + a large stderr debug output
> --------------------------------------------------------------------------
>
> Key: HADOOP-3384
> URL: https://issues.apache.org/jira/browse/HADOOP-3384
> Project: Hadoop Core
> Issue Type: Bug
> Components: contrib/streaming
> Affects Versions: 0.16.4
> Reporter: Koji Noguchi
>
> We've seen streaming task hang forever, (thus hang the job), when
> 1) There's no input to a mapper or a reducer
> 2) Streaming process writes some debug statements.
> (Either flush() or a large output to cause a flush)
> PipeMapper/Reducer waits for the streaming process to finish.
> Streaming process stuck on writing to stdout/stderr.
> This is due to no MROutputThread/MRErrThread being created until first input
> record is passed to map() or reduce().
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
