[
https://issues.apache.org/jira/browse/OOZIE-3087?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Peter Cseh updated OOZIE-3087:
------------------------------
Description:
§Right now rerun has two options:
- rerun failed
- rerun all but specify which nodes to skip
We should create an option that says "start from the last failed node and force
rerun everything after".
Imagine a scenario where there is a fork where every action in the fork goes
to the join node in both the <ok> and the <error> transition. This can happen
to avoid killing every action in the fork when one of them fails. Later they
can use a decision node to check if any of the actions failed in the fork.
If one of the actions in the fork fails and the WF is rerun with the "failed
only" option, the decision node won't get re-executed as it haven't failed
before. The new option would solve this issue
was:
Right now rerun has two options:
- rerun failed
- rerun all but specify which nodes to skip
We should create an option that says "start from the last failed node and
force rerun everything after".
Imagine a scenario where there is a fork where every action in the fork goes to
the join node in both the <ok> and the <error> transition. This can happen to
avoid killing every action in the fork when one of them fails. Later they can
use a decision node to check if any of the actions failed in the fork.
If one of the actions in the fork fails and the WF is rerun with the "failed
only" option, the decision node won't get re-executed as it haven't failed
before. The new option would solve this issue
> Add option to rerun "failed and subsequent nodes" to rerun
> ----------------------------------------------------------
>
> Key: OOZIE-3087
> URL: https://issues.apache.org/jira/browse/OOZIE-3087
> Project: Oozie
> Issue Type: New Feature
> Components: core
> Reporter: Peter Cseh
> Priority: Major
>
> §Right now rerun has two options:
> - rerun failed
> - rerun all but specify which nodes to skip
> We should create an option that says "start from the last failed node and
> force rerun everything after".
> Imagine a scenario where there is a fork where every action in the fork goes
> to the join node in both the <ok> and the <error> transition. This can happen
> to avoid killing every action in the fork when one of them fails. Later they
> can use a decision node to check if any of the actions failed in the fork.
> If one of the actions in the fork fails and the WF is rerun with the "failed
> only" option, the decision node won't get re-executed as it haven't failed
> before. The new option would solve this issue
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
