You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@falcon.apache.org by "Venkatesh Seetharam (JIRA)" <ji...@apache.org> on 2013/07/03 19:54:21 UTC
[jira] [Commented] (FALCON-38) Falcon's parent workflow actions
(pre-processing & post-processing) should have mulitple retries
[ https://issues.apache.org/jira/browse/FALCON-38?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13699252#comment-13699252 ]
Venkatesh Seetharam commented on FALCON-38:
-------------------------------------------
Good catch. Post processing in itself an action could fail. AFAIK, oozie lets you configure reties based on the error codes which requires change to oozie-site conf. Does it let you add reties in the workflow action definition?
Also, does it make sense to listen to oozie workflow job end notifications as well which falcon can act upon failures and ignores successful ones.
> Falcon's parent workflow actions (pre-processing & post-processing) should have mulitple retries
> ------------------------------------------------------------------------------------------------
>
> Key: FALCON-38
> URL: https://issues.apache.org/jira/browse/FALCON-38
> Project: Falcon
> Issue Type: Bug
> Components: oozie, process
> Affects Versions: 0.3
> Environment: Oozie 3.2.X +, Falcon 0.1
> Reporter: Shaik Idris Ali
> Labels: failures, jms, oozie, retries,
> Original Estimate: 48h
> Remaining Estimate: 48h
>
> Oozie actions may fail for various transient and non-transient problems arising in Hadoop, Network etc.
> Falcon handles user-workflow action failures by sending a JMS notification to ActiveMQ (post-processing action in parent-wf), this message is used to provide automatic retries based on various retry policies defined in Process.xml. However, post-processing action of Falcon itself might fail and workflow may remain in killed/suspended state.
> Oozie provides OOTB retires that can be configured per action, we should leverage this to provide Falcon's parent workflow action reruns.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira