[
https://issues.apache.org/jira/browse/MAPREDUCE-6343?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14516383#comment-14516383
]
Karthik Kambatla commented on MAPREDUCE-6343:
---------------------------------------------
bq. Casting to integer may not be a good idea, because the user might expect
the same heap as he configured but actually not.
The heap size in javaopts as parsed by this method is used ONLY to determine
what the container size should be, if it is not specified. The javaopts string
is used to spawn the tasks as is.
The approach in the current patch seems reasonable to me. Comments on the patch:
# Instead of using the Guava call, let us just use the simple Java type cast.
# We should add unit tests for this method for various reasonable heap sizes
specified in any of GB, MB, KB, B.
> JobConf.parseMaximumHeapSizeMB() fails to parse value greater than 2GB
> expressed in bytes
> -----------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-6343
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-6343
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Affects Versions: 3.0.0
> Reporter: Hao Xia
> Attachments: parse-heap-size.patch
>
>
> It currently tries to parse the value as an integer, which blows up whenever
> the value is greater than 2GB and expressed in bytes.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
