[
https://issues.apache.org/jira/browse/MAPREDUCE-2026?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12920107#action_12920107
]
luoli commented on MAPREDUCE-2026:
----------------------------------
We use our own hadoop release and we found the same problem because our
cluster is using by lots of application developers and most of them like to
refresh the jobtracker webui to get the real-time situation of their
jobs,specially the jobdetail.jsp. This page want to getCounters of a JIP and
will lock the JIP object, the more jobdetail page accessed by people ,
jobtracker get more inefficient. From our cluster, abount 26% of jobtracker's
lock taken by the jsp access.
> JobTracker.getJobCounters() should not hold JobTracker lock while calling
> JobInProgress.getCounters()
> -----------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-2026
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-2026
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Reporter: Scott Chen
> Assignee: Joydeep Sen Sarma
> Fix For: 0.22.0
>
> Attachments: MAPREDUCE-2026.txt
>
>
> JobTracker.getJobCounter() will lock JobTracker and call
> JobInProgress.getCounters().
> JobInProgress.getCounters() can be very expensive because it aggregates all
> the task counters.
> We found that from the JobTracker jstacks that this method is one of the
> bottleneck of the JobTracker performance.
> JobInProgress.getCounters() should be able to be called out side the
> JobTracker lock because it already has JobInProgress lock.
> For example, it is used by jobdetails.jsp without a JobTracker lock.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
