[
https://issues.apache.org/jira/browse/OOZIE-33?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13092581#comment-13092581
]
Alejandro Abdelnur commented on OOZIE-33:
-----------------------------------------
Same rationale as my comment for OOZIE-29, Oozie should just give back the
Hadoop Job IDs. Using those Job IDs the user can fetch the stats directly from
Hadoop.
> Oozie to provide the job(MR/pig) stats through WS API and CLI
> -------------------------------------------------------------
>
> Key: OOZIE-33
> URL: https://issues.apache.org/jira/browse/OOZIE-33
> Project: Apache Oozie (Incubating)
> Issue Type: Improvement
> Reporter: Mohammad Kamrul Islam
>
> Oozie users want to access to their MR and Pig job stats through API and CLI.
> Currently Oozie doesn't supports this functionality.
> However, Oozie stores only the job counter of MR job in its database.
> We want to do it the following way:
> 1. Oozie will not store any additional stats in its table. Rather Oozie will
> act as pass-thru to the downstream (JT).
> 2. When user will ask for any stats of a MR job, Oozie will query the job
> tracker using JobClient/Job and returns the response as-it-is.
> 3. For Pig job, we will need to store the pig stats as Launcher Mapper job
> counter. So when people will ask for stats they will get it from the
> Launcher Mapper job counter.
> Comments?
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
