[
https://issues.apache.org/jira/browse/HADOOP-3616?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12607258#action_12607258
]
Ari Rabkin commented on HADOOP-3616:
------------------------------------
This code looks correct -- the programmer's goal was presumably to ensure that
the split size was at least mapred.min.split.size -- and that means, you need
to take the maximum.
> TextInputFormat taking max of two minima as the minimum
> -------------------------------------------------------
>
> Key: HADOOP-3616
> URL: https://issues.apache.org/jira/browse/HADOOP-3616
> Project: Hadoop Core
> Issue Type: Bug
> Affects Versions: 0.17.0
> Reporter: Josh Myer
> Priority: Minor
>
> When choosing its minimum split size, FileInputFormat is using the larger of
> the two minimum split values, instead of the smaller. I can't find any good
> explanation for why this would be, so it would be helpful to add a comment
> there (or change to Math.min if that's the intent).
> Line 237:
> long minSize = Math.max(job.getLong("mapred.min.split.size", 1),
> minSplitSize);
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
