[
https://issues.apache.org/jira/browse/PIG-3915?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14189388#comment-14189388
]
Suhas Satish commented on PIG-3915:
-----------------------------------
Thanks for the reply guys. The issue turned out to be that the jar containing
the pigmix classes was compiled against hadoop-2 and was being run against
hadoop-1. After building 2 separate jars 1 against hadoop-1 and another against
hadoop-2 and using the appropriate one solved the issue.
> MapReduce queries in Pigmix outputs different results than Pig's
> ----------------------------------------------------------------
>
> Key: PIG-3915
> URL: https://issues.apache.org/jira/browse/PIG-3915
> Project: Pig
> Issue Type: Bug
> Components: tools
> Affects Versions: 0.12.0
> Reporter: Keren Ouaknine
> Assignee: Keren Ouaknine
> Fix For: 0.13.0
>
> Attachments: PIG-3915.2.patch
>
>
> Hello,
> The Pigmix benchmark has 17 queries comparing Pig to MapReduce Java. Some of
> these queries are not outputting the same results in Pig and MapReduce.
> Looking into the outputs, it seems the errors reside in the MapReduce code.
> For example, L6 has no output because the output of the map function sends
> the wrong key to the reducer: "query_term" (field 3) instead of "timespent"
> (field 2). Hence an exception is thrown, and there is no output to the query
> in MapReduce. I am planning to submit a patch once I fixed all the queries in
> MapReduce :)
> Thanks,
> Keren
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
