[
https://issues.apache.org/jira/browse/MAPREDUCE-486?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13007630#comment-13007630
]
Todd Lipcon commented on MAPREDUCE-486:
---------------------------------------
Hm, I haven't seen this behavior in a while actually. Maybe we now properly
consider COMMIT_PENDING tasks as taking up a slot, which means that it's
appropriate to count them as Running. Has anyone else ever seen more "running"
tasks than the total capacity?
Should be easy to reproduce by writing an outputcommitter that sleeps for a
minute or something.
> JobTracker web UI counts COMMIT_PENDING tasks as Running
> --------------------------------------------------------
>
> Key: MAPREDUCE-486
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-486
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Reporter: Todd Lipcon
> Assignee: Harsh J Chouraria
> Priority: Minor
>
> In jobdetails.jsp, tasks in COMMIT_PENDING state are listed as "Running". I
> propose creating another column in this table for COMMIT_PENDING tasks, since
> users find it confusing that a given job can have more tasks "Running" than
> their total cluster capacity.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
