[
https://issues.apache.org/jira/browse/MAPREDUCE-3385?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Chris Nauroth updated MAPREDUCE-3385:
-------------------------------------
Labels: BB2015-05-TBR (was: BB2015-05-RFC)
Thank you for providing the patch. Here are a few comments.
# Please make the logger private.
# I suggest using the message "Arithmetic underflow" for the second condition.
# There are several other reducers that are also subject to overflow/underflow.
To make this patch comprehensive, we could update all of them. Here are the
relevant ones that I spotted in current trunk.
## {{org.apache.hadoop.mapred.lib.LongSumReducer}}
## {{org.apache.hadoop.mapreduce.lib.reduce.IntSumReducer}}
## {{org.apache.hadoop.mapreduce.lib.reduce.LongSumReducer}}
## {{org.apache.hadoop.mapred.nativetask.kvtest.HashSumReducer}}
## {{org.apache.hadoop.mapreduce.lib.aggregate.LongValueSum}}
## {{org.apache.hadoop.mapreduce.lib.aggregate.DoubleValueSum}}
## {{org.apache.hadoop.fs.AccumulatingReducer}}
> Add warning message for the overflow in reduce() of
> org.apache.hadoop.mapreduce.lib.reduce.IntSumReducer
> --------------------------------------------------------------------------------------------------------
>
> Key: MAPREDUCE-3385
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-3385
> Project: Hadoop Map/Reduce
> Issue Type: Improvement
> Reporter: JiangKai
> Priority: Minor
> Labels: BB2015-05-TBR
> Attachments: MAPREDUCE-3385.patch
>
>
> When we call the function reduce() of IntSumReducer,the result may overflow.
> We should send a warning message to users if overflow occurs.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
