[
https://issues.apache.org/jira/browse/HADOOP-9426?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13622546#comment-13622546
]
Nick Dimiduk commented on HADOOP-9426:
--------------------------------------
The behavior I see indicates that jar classes are available on the classpath at
the point at which the client creates a job. The desire was to use JarFinder to
package those classes back into a jar, so that they can be re-submitted via
-tmpjars along with the job.
It was suggested that this re-packaging and re-submission is not necessary as
the jars should already be available in the distributed cache. If that is in
face the case, I'd prefer to use those existing jars, and request additional
guidance in such an implementation.
My understanding was that if the jars were already available, the existing code
would have found them.
> Hadoop should expose Jar location utilities on its public API
> -------------------------------------------------------------
>
> Key: HADOOP-9426
> URL: https://issues.apache.org/jira/browse/HADOOP-9426
> Project: Hadoop Common
> Issue Type: Bug
> Affects Versions: 1.0.0, 1.1.0, 2.0.0-alpha
> Reporter: Nick Dimiduk
> Attachments:
> 0001-HADOOP-9426-Promote-JarFinder-out-of-test-jar.patch,
> 0001-HADOOP-9426-Promote-JarFinder-out-of-test-jar.patch
>
>
> The facilities behind JobConf#setJarByClass and the JarFinder utility in test
> are both generally useful. As the core platform, these should be published as
> part of the public API. In addition to HBase, they are probably useful for
> Pig and Hive as well. See also HBASE-2588, HBASE-5317, HBASE-8140.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
