You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@falcon.apache.org by "Balu Vellanki (JIRA)" <ji...@apache.org> on 2015/08/04 03:07:04 UTC

[jira] [Commented] (FALCON-1366) Process vertex in graphDb is committed before process submission is complete

    [ https://issues.apache.org/jira/browse/FALCON-1366?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14652900#comment-14652900 ] 

Balu Vellanki commented on FALCON-1366:
---------------------------------------

The issue could not be reproduced with above steps. 

Upon further debugging with [~yzheng-hortonworks], we found that the real issue was with an inconsistent state in titan-db. For unknown reason, the DB failed to release lock on an existing process-entity vertex and hence did not allow adding new entity. 

Closing this as invalid.  

> Process vertex in graphDb is committed before process submission is complete
> ----------------------------------------------------------------------------
>
>                 Key: FALCON-1366
>                 URL: https://issues.apache.org/jira/browse/FALCON-1366
>             Project: Falcon
>          Issue Type: Bug
>    Affects Versions: 0.6.1
>            Reporter: Balu Vellanki
>            Assignee: Balu Vellanki
>            Priority: Critical
>             Fix For: 0.7
>
>
> To reproduce the problem, submit a feed "TestFeed" with validity start as "2015-07-31 08:56" and End as "2015-07-31 10:00". Now using the UI, submit a process with validity start "2015-07-30 08:56" that will write output to "TestFeed", this will fail with error 
> {code}
> Instance now(0,0) of feed TestFeed is before the start of feed Fri Jul 31 08:56:00 UTC 2015 for cluster primaryCluster
> {code}
> Now fix the validity time start and end of process, make it same as validity start and end of feed. Resubmit the process. It fails with Titan graphDB error. 
> {code}
> Could not commit transaction due to exception during persistence
> {code}
> This is happening because the process name is committed to graphDB even before the process submission is successful in the config store. This needs to be fixed. 



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