hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sean Busbey (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-16383) TableMapReduceUtil should allow using JobConf, not Job
Date Thu, 11 Aug 2016 16:13:20 GMT

    [ https://issues.apache.org/jira/browse/HBASE-16383?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15417506#comment-15417506

Sean Busbey commented on HBASE-16383:

Both the {{mapred}} and {{mapreduce}} APIs are supported upstream in Hadoop last I checked.
We provide implementations for both, but in each we stick to the conventions of that API.
In the {{mapreduce}} API the {{Job}} object is the submitters' view of the submittal / configuration.

If Oozie is only providing the mechanisms from the {{mapred}} API, why not just stick with
that? Or ask them to provide a {{mapreduce}} oriented version?

> TableMapReduceUtil should allow using JobConf, not Job
> ------------------------------------------------------
>                 Key: HBASE-16383
>                 URL: https://issues.apache.org/jira/browse/HBASE-16383
>             Project: HBase
>          Issue Type: Improvement
>          Components: util
>            Reporter: Nikolai Grigoriev
>            Priority: Minor
> I was recently trying to set up an MR job initialized with TableMapReduceUtil to be ran
by Oozie. Oozie offers a way to configure the job by providing an implementation of org.apache.oozie.action.hadoop.OozieActionConfigurator.
Unfortunately, this interface is using org.apache.hadoop.mapred.JobConf, not Job. I may be
wrong but I believe that everything TableMapReduceUtil does actually maps to the job configuration
so it does not really need the Job itself. If this is true, probably it would be more appropriate
to use JobConf in TableMapReduceUtil instead of Job? Or provide the methods for both.

This message was sent by Atlassian JIRA

View raw message