[
https://issues.apache.org/jira/browse/FALCON-1433?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14946467#comment-14946467
]
Peeyush Bishnoi commented on FALCON-1433:
-----------------------------------------
On further debugging this issue, I have found that if any action node in
sub-workflow fails then entire sub-workflow is getting executed on 1st retry.
This is similar to Pig/Hive query, where if any job fails in query, then entire
Pig/Hive query execute.
So, as such I have not seen any impact of property "oozie.wf.rerun.failnodes"
to "true" for ensuring that only failed action nodes execute from sub-workflow.
With this behavior, we should think to set "oozie.wf.rerun.failnodes" to
"false" and ensure that all instances must run till the defined number of
attempts in case of failure.
Thoughts please.
> Falcon retry instance do not proceed on failure
> -----------------------------------------------
>
> Key: FALCON-1433
> URL: https://issues.apache.org/jira/browse/FALCON-1433
> Project: Falcon
> Issue Type: Bug
> Components: rerun
> Reporter: Peeyush Bishnoi
> Assignee: Peeyush Bishnoi
> Priority: Critical
> Fix For: trunk
>
>
> When retry is specified in Falcon entity with number of attempts more than 1
> and if workflow failed, on retrying Falcon instance workflow (if action fail
> again) seems like stuck and further attempts also does not get retry.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
