[
https://issues.apache.org/jira/browse/HIVE-6658?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13935853#comment-13935853
]
Szehon Ho commented on HIVE-6658:
---------------------------------
Yea we should have two tests and keep testing 0.20.
Sorry for not being clear, my suggestion was : the hadoop1 version can have
--include 0.20,0.20S and the hadoop2 version can have --exclude 0.20,0.20S.
This is how most tests are split right now. In this patch, hadoop1 version
says --exclude hadoop0.23 and hadoop2 version says --include 0.23. But then,
any later version of hadoop beyond 0.23 will test using hadoop1 version, which
would be the 'wrong' result in this case. Just a thought.
> Modify Alter_numbuckets* test to reflect hadoop2 changes
> --------------------------------------------------------
>
> Key: HIVE-6658
> URL: https://issues.apache.org/jira/browse/HIVE-6658
> Project: Hive
> Issue Type: Bug
> Affects Versions: 0.12.0
> Reporter: Laljo John Pullokkaran
> Assignee: Laljo John Pullokkaran
> Attachments: HIVE-6658.patch
>
>
> Hadoop2 now honors number of reducers config while running in local mode.
> This affects bucketing tests as the data gets properly bucketed in Hadoop2
> (In hadoop1 all data ended up in same bucket while in local mode).
--
This message was sent by Atlassian JIRA
(v6.2#6252)
