Maxime Petazzoni created OOZIE-873:
--------------------------------------
Summary: Change kill node to fail node yielding the FAILED state
instead of the KILLED state
Key: OOZIE-873
URL: https://issues.apache.org/jira/browse/OOZIE-873
Project: Oozie
Issue Type: Bug
Components: workflow
Affects Versions: trunk
Reporter: Maxime Petazzoni
Fix For: trunk
With the current kill node, ending a workflow in the KILLED state, it is
impossible from a monitoring standpoint to differentiate between a workflow
that was killed by the command-line from a workflow that died in error because
one of its node jumped to the kill node.
I think it would make more sense to use the FAILED state better and change the
kill node to a fail node, yielding the FAILED state for the workflow when the
fail node is reached.
The same discussion actually arises for the use of the FAILED state. I'm not
sure what the FAILED state is used for right now though.
Changing the name of the kill node to "fail" also means it's not a backward
compatible change (we can make the kill node through a FAILED state during a
given deprecation period though).
Discussion is open; I can be on the hook for the actual implementation once we
have an agreement on the path to follow on this one.
/Maxime
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
