[jira] [Commented] (FALCON-1720) Rerun API does not rerun succeeded instances

["pavan kumar kolamuri (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/FALCON-1720?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15082678#comment-15082678 ] 

pavan kumar kolamuri commented on FALCON-1720:
----------------------------------------------

I am explaining all cases here 
Normal Rerun -  Rerunning an instance without force, it will rerun only failed actions
Force Rerun - Rerunning an instance forcefully either by using force option or passing props , it will run all actions irrespective of status

Initially if you do force rerun or normal rerun then rerun will work fine without any issues, issue will come only when you do rerun twice, i will explain all those scenarios here.

1) Normal Rerun and Normal Rerun -  Both Current Falcon and Native Scheduler will work in this case.
2) Normal Rerun and Force Rerun - Current Falcon will work in this case and Native Scheduler won't work in this case
3) Force Rerun and Force Rerun -  Both Current Falcon and Native Scheduler will work in this case.
4) Force Rerun and Normal Rerun - Both Current Falcon and Native Scheduler won't work in this case, it will always do force rerun

Apart from these cases there is also properties option where user can pass workflow properties.
1) If Force flag and user properties are contradictory which should be prioritized ? In code it says force flag will be ignored in current but it is not happening with current falcon. I am thinking to make ignoring force option if user has set any of these props. 



> Rerun API does not rerun succeeded instances
> --------------------------------------------
>
>                 Key: FALCON-1720
>                 URL: https://issues.apache.org/jira/browse/FALCON-1720
>             Project: Falcon
>          Issue Type: Bug
>          Components: scheduler
>    Affects Versions: 0.9
>            Reporter: Pragya Mittal
>            Assignee: pavan kumar kolamuri
>
> When entity is scheduled via native scheduler, rerun api does not force rerun the succeeded instances. Although oozie run count gets incremented but state of instance does not change from 'SUCCEEDED' to 'RUNNING'



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)