[ 
https://issues.apache.org/jira/browse/HADOOP-12310?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14662024#comment-14662024
 ] 

Allen Wittenauer edited comment on HADOOP-12310 at 8/7/15 4:06 PM:
-------------------------------------------------------------------

OK, this looks like an easy one.  This is the "head and tail will exit when it 
can fulfill the request" feature.  The program writing to the pipe still has 
content to send but head/tail has already exited, generating the error.  We've 
hit this before in compute_gitdiff where the fix was just to throw the error 
away.  So in this case, sort's stderr needs to get ignored.


was (Author: aw):
OK, this is an easy one.  This is the "head and tail will exit when it can 
fulfill the request" feature.  The program writing to the pipe still has 
content to send but head/tail has already exited, generating the error.  We've 
hit this before in compute_gitdiff where the fix was just to throw the error 
away.  So in this case, sort's stderr needs to get ignored.

> final memory report sometimes generates spurious errors
> -------------------------------------------------------
>
>                 Key: HADOOP-12310
>                 URL: https://issues.apache.org/jira/browse/HADOOP-12310
>             Project: Hadoop Common
>          Issue Type: Sub-task
>          Components: yetus
>    Affects Versions: HADOOP-12111
>            Reporter: Allen Wittenauer
>
> There are spurious sort write pipeline failures coming from the maven memory 
> check on Jenkins.
> https://builds.apache.org/job/PreCommit-HADOOP-Build/7422/console
> with bash debug turned on:
> https://builds.apache.org/job/PreCommit-HADOOP-Build/7423/console



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to