[
https://issues.apache.org/jira/browse/HIVE-10959?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14578231#comment-14578231
]
Ivan Mitic commented on HIVE-10959:
-----------------------------------
Thanks for reviewing Thejas!
Child jobs are tagged with parent's job Id. So even if there is more then one
job, we should be able to find them when we query for all child jobs (I know
this works for hive/pig jobs which spawn more then one mr job - I tested this).
I assume user can do the wrong thing here by not carrying the tag explicitly,
but I would argue this is not supported.
In this patch I log a warning if we detect more then one child job in case of
MR. Another possibly better way to handle this is to say that reconnect is not
supported in this case, and let the regular code path handle this (kill all
child jobs and relaunch). Let me know what you think.
> Templeton launcher job should reconnect to the running child job on task retry
> ------------------------------------------------------------------------------
>
> Key: HIVE-10959
> URL: https://issues.apache.org/jira/browse/HIVE-10959
> Project: Hive
> Issue Type: Bug
> Components: WebHCat
> Affects Versions: 0.15.0
> Reporter: Ivan Mitic
> Assignee: Ivan Mitic
> Attachments: HIVE-10959.patch
>
>
> Currently, Templeton launcher kills all child jobs (jobs tagged with the
> parent job's id) upon task retry.
> Upon templeton launcher task retry, templeton should reconnect to the running
> job and continue tracking its progress that way.
> This logic cannot be used for all job kinds (e.g. for jobs that are driven by
> the client side like regular hive). However, for MapReduceV2, and possibly
> Tez and HiveOnTez, this should be the default.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
