[
https://issues.apache.org/jira/browse/MAPREDUCE-6724?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15368329#comment-15368329
]
Haibo Chen commented on MAPREDUCE-6724:
---------------------------------------
Thanks a lot for your reviews, [~jira.shegalov]! And my apologies for being
late to update the patch. I have updated the patch to accommodate your comments
except for one thing. Instead of creating a method that just takes
targetSingleShuffleLimit as its parameter, I have four parameters for the new
private method, since the setup of testLargeMemoryLimits involves
REDUCE_MEMORY_TOTAL_BYTES, SHUFFLE_INPUT_BUFFER_PERCENT and
REDUCE_INPUT_BUFFER_PERCENT whereas the new test method involves
REDUCE_MEMORY_TOTAL_BYTES, SHUFFLE_INPUT_BUFFER_PERCENT and
SHUFFLE_MEMORY_LIMIT_PERCENT (testLargeMemoryLimits does not care about the
value of targetSingleShuffleLimit). Thanks again for your review and please let
me know any more comments you have.
> Unsafe conversion from long to int in MergeManagerImpl.unconditionalReserve()
> -----------------------------------------------------------------------------
>
> Key: MAPREDUCE-6724
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-6724
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2
> Reporter: Haibo Chen
> Assignee: Haibo Chen
> Attachments: mapreduce6724.001.patch, mapreduce6724.002.patch,
> mapreduce6724.003.patch
>
>
> When shuffle is done in memory, MergeManagerImpl converts the requested size
> to an int to allocate an instance of InMemoryMapOutput. This results in an
> overflow if the requested size is bigger than Integer.MAX_VALUE and
> eventually causes the reducer to fail.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
