hadoop-mapreduce-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Jason Lowe (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (MAPREDUCE-6622) Add capability to set JHS job cache to a task-based limit
Date Tue, 23 Feb 2016 14:43:18 GMT

    [ https://issues.apache.org/jira/browse/MAPREDUCE-6622?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15158975#comment-15158975
] 

Jason Lowe commented on MAPREDUCE-6622:
---------------------------------------

Sorry for the delay in responding, as I was out on vacation.

bq.  Jason Lowe, Ray Chiang - did you guys have any other specific concerns with using guava
that I am discounting?

My main concern was that we were adding a separate thread and extra overhead for something
that didn't seem worth the trouble.  Now that the cache code has been significantly cleaned
up, I'm more comfortable with going the guava route.

bq. Jason Lowe - do you remember if the number of jobs was being used as a proxy for the memory
usage?

Yes, it was essentially a proxy for capping memory usage.  It's also a performance tunable,
since caching more jobs can lead to improved performance depending upon client access patterns.
 However I think it is OK to override the old value with the new value as the behavior is
clearly documented and we're not providing any default value that would override the user's
old settings implicitly when they upgrade.  Users would have to go out of their way to set
this value and in doing so should encounter the documentation explaining the semantics when
they set it.


> Add capability to set JHS job cache to a task-based limit
> ---------------------------------------------------------
>
>                 Key: MAPREDUCE-6622
>                 URL: https://issues.apache.org/jira/browse/MAPREDUCE-6622
>             Project: Hadoop Map/Reduce
>          Issue Type: Improvement
>          Components: jobhistoryserver
>    Affects Versions: 2.7.2
>            Reporter: Ray Chiang
>            Assignee: Ray Chiang
>              Labels: supportability
>         Attachments: MAPREDUCE-6622.001.patch, MAPREDUCE-6622.002.patch, MAPREDUCE-6622.003.patch,
MAPREDUCE-6622.004.patch
>
>
> When setting the property mapreduce.jobhistory.loadedjobs.cache.size the jobs can be
of varying size.  This is generally not a problem when the jobs sizes are uniform or small,
but when the job sizes can be very large (say greater than 250k tasks), then the JHS heap
size can grow tremendously.
> In cases, where multiple jobs are very large, then the JHS can lock up and spend all
its time in GC.  However, since the cache is holding on to all the jobs, not much heap space
can be freed up.
> By setting a property that sets a cap on the number of tasks allowed in the cache and
since the total number of tasks loaded is directly proportional to the amount of heap used,
this should help prevent the JHS from locking up.



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

Mime
View raw message