hadoop-mapreduce-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "luoli (JIRA)" <j...@apache.org>
Subject [jira] Commented: (MAPREDUCE-2116) optimize getTasksToKill to reduce JobTracker contention
Date Thu, 14 Oct 2010 09:45:34 GMT

    [ https://issues.apache.org/jira/browse/MAPREDUCE-2116?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12920905#action_12920905
] 

luoli commented on MAPREDUCE-2116:
----------------------------------

bq. Joydeep has similar idea. But he created another class that keeps a pair. We are testing
this internally.
en~, that can work too.

And now we found the reason why the taskStatuses.get in shouldClose() consume so much time.
That's because in getTasksToKill(), the shouldClose() got called so many times. This is because
for every TaskAttemptID in taskIds which got from 
Set<TaskAttemptID> taskIds = trackerToTaskMap.get(taskTracker);
call, the taskIds will contains hundreds of entry if the tasktracker have finished lots of
task attempts and those task's job all have not finished finally. So all those task attempts
are contained in trackerToTaskMap's value set. In this case, even lots of tasks attempt have
finished in the tasktracker, it gets iterated in every heartbeat of the tasktracker before
the jobs which they belong to have finished. This is not good.

> 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, 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.


Mime
View raw message