[
https://issues.apache.org/jira/browse/MAPREDUCE-4730?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jason Lowe updated MAPREDUCE-4730:
----------------------------------
Attachment: MAPREDUCE-4730.patch
New patch that attempts to scale the maximum number of events a reducer will
ask for per RPC call based on some fuzzy numbers. It still keeps
maxEventsToFetch between 100 and 10000 to avoid extremes.
Since events appear to be just a little under 100 bytes each, the patch
currently targets around 300MB of memory on the AM for RPC response processing.
This can still be exceeded given enough reducers, but the user should be able
to bump up the AM memory size at that point and buy quite a bit more reducers.
This patch also implements the do-not-wait-if-we-got-a-full-response logic to
avoid wasting time while trying to fetch all the completion events.
Still need to do some testing at scale, but quick touch-testing on a
single-node cluster seems to work so putting it out there for comment and
Jenkins.
> AM crashes due to OOM while serving up map task completion events
> -----------------------------------------------------------------
>
> Key: MAPREDUCE-4730
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-4730
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: applicationmaster, mrv2
> Affects Versions: 0.23.3
> Reporter: Jason Lowe
> Assignee: Jason Lowe
> Priority: Blocker
> Attachments: MAPREDUCE-4730.patch, MAPREDUCE-4730.patch
>
>
> We're seeing a repeatable OOM crash in the AM for a task with around 30000
> maps and 3000 reducers. Details to follow.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
