[
https://issues.apache.org/jira/browse/OOZIE-2129?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14342919#comment-14342919
]
Jaydeep Vishwakarma commented on OOZIE-2129:
--------------------------------------------
1. Added back the launch time in filter.
3. Yes this patch will fix the issue. I meant without this patch it will not
work.
For 2 and 5 : "they have to implement some code" propagation.xml is the way
for the launcher to share details in case they need to take advantage of it,
since there are no other simple ways or established contract to share info in a
backward compatible way.
6. Yes, when any coord_action/workflow/launcher are running, oozie will not
allow to rerun. The issues comes when the job launched by launcher is in
running state and coord_action/workflow/launcher is in killed state, In this
case, before rerunning the coord_action/workflow/launcher none of the previous
launched job is running on cluster. Check my second comment in the same jira
that explain in what cases this condition can occur.
> Duplicate child jobs per instance
> ---------------------------------
>
> Key: OOZIE-2129
> URL: https://issues.apache.org/jira/browse/OOZIE-2129
> Project: Oozie
> Issue Type: Bug
> Reporter: Shwetha G S
> Assignee: Jaydeep Vishwakarma
> Attachments: OOZIE-2129-v0.patch, OOZIE-2129-v1.patch,
> OOZIE-2129-v2.patch
>
>
> OOZIE-1722 ensures that child job is killed at launcher restart. But this
> doesn't work for java actions as the tag is not passed to the child job.
> In case of coord action rerun, new workflow is created and hence new tag. So,
> it doesn't ensure old child job is killed at launcher start
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
