[
https://issues.apache.org/jira/browse/PIG-2652?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Dmitriy V. Ryaboy updated PIG-2652:
-----------------------------------
Attachment: PIG-2652_6.patch
Ok, attaching latest.
This always adds the limiting MR job, even though it's not always strictly
necessary. I feel that getting number of reducers estimated well for big jobs
is more important than saving a tiny MR job (and this job will be tiny -- if
it's unnecessary, by definition all that it does is read a LIMIT-ed number of
rows, and output them back out).
We will create a separate ticket to apply an optimization that eliminates this
extra limit job when possible.
> Skew join and order by don't trigger reducer estimation
> -------------------------------------------------------
>
> Key: PIG-2652
> URL: https://issues.apache.org/jira/browse/PIG-2652
> Project: Pig
> Issue Type: Bug
> Reporter: Bill Graham
> Assignee: Bill Graham
> Fix For: 0.10.0, 0.9.3, 0.11
>
> Attachments: PIG-2652_1.patch, PIG-2652_2.patch, PIG-2652_3.patch,
> PIG-2652_3_10.patch, PIG-2652_4.patch, PIG-2652_5.patch, PIG-2652_6.patch
>
>
> If neither PARALLEL, default parallel or {{mapred.reduce.tasks}} are set, the
> number of reducers is not estimated based on input size for skew joins or
> order by. Instead, these jobs get only 1 reducer.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
