[
https://issues.apache.org/jira/browse/TEZ-3486?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15603378#comment-15603378
]
Rohini Palaniswamy commented on TEZ-3486:
-----------------------------------------
Few comments:
1) Please include a testcase where the combiner implementation does not always
return 1 record for every Key,List<Values> input. For eg: write a combiner that
writes out first or top two elements.
2) Please rename combineInputKeyCounter -> combineInputRecordsCounter and
combineInputValueCounter -> combineOutputRecordsCounter. Old names do not make
sense anymore.
> COMBINE_OUTPUT_RECORDS/COMBINE_INPUT_RECORDS are not correct
> ------------------------------------------------------------
>
> Key: TEZ-3486
> URL: https://issues.apache.org/jira/browse/TEZ-3486
> Project: Apache Tez
> Issue Type: Bug
> Reporter: Eric Badger
> Assignee: Eric Badger
> Attachments: TEZ-3486.001.patch
>
>
> They appear to be swapped, though that is not quite the case. You can
> consistently see {{COMBINE_INPUT_RECORDS}} < {{COMBINE_OUTPUT_RECORDS}},
> which shouldn't be possible.
> This issue relates to TEZ-1344, which was closed due to not being
> reproducible, but it actually is fairly easily reproducible if you run the
> same job on Tez and on MR. For whatever reason, I cannot reopen that Jira, so
> I'm creating this one.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
