[
https://issues.apache.org/jira/browse/MAPREDUCE-2116?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12921040#action_12921040
]
Joydeep Sen Sarma commented on MAPREDUCE-2116:
----------------------------------------------
@luoli - shouldClose() had an expensive synchronized call that was not
necessary. i have updated the patch with that change - it should help things
(after we made this change - shouldClose() stopped appearing as one of the top
calls in the profile).
beyond this - we are working on restructuring the locks so that the JT lock is
not held for calls that are made at the JIP level. think about it - there's no
reason why we should hold JT lock when trying to find out whether a taskattempt
needs to be closed (it should suffice to hold the JIP lock at the most). we
are tracking it via MAPREDUCE-2118.
> optimize getTasksToKill to reduce JobTracker contention
> -------------------------------------------------------
>
> Key: MAPREDUCE-2116
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-2116
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Components: jobtracker
> Reporter: Joydeep Sen Sarma
> Attachments: 2116.1.patch, 2116.2.patch, getTaskToKill.JPG
>
>
> getTasksToKill shows up as one of the top routines holding the JT lock.
> Specifically, the translation from attemptid to tip is very expensive:
> at java.util.TreeMap.getEntry(TreeMap.java:328)
> at java.util.TreeMap.get(TreeMap.java:255)
> at
> org.apache.hadoop.mapred.TaskInProgress.shouldClose(TaskInProgress.java:500)
> at
> org.apache.hadoop.mapred.JobTracker.getTasksToKill(JobTracker.java:3464)
> locked <0x00002aab6ebb6640> (a org.apache.hadoop.mapred.JobTracker)
> at org.apache.hadoop.mapred.JobTracker.heartbeat(JobTracker.java:3181)
> this seems like an avoidable expense since the tip for a given attempt is
> fixed (and one should not need a map lookup to find the association). on a
> different note - not clear to me why TreeMaps are in use here (i didn't find
> any iteration over these maps). any background info on why things are
> arranged the way they are would be useful.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
