[
https://issues.apache.org/jira/browse/OOZIE-2429?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15085752#comment-15085752
]
Ferenc Denes commented on OOZIE-2429:
-------------------------------------
Looks like that not only the the test is wrong (where the EventHandlerService
manipulates the event queue in the background), but there are really duplicate
COORDINATOR_ACTION - FAILURE messages. We have to modify the event generation
by checking if the coordinator action's state has really changed at the update
stage.
The test passes in some cases, where the background processes are slow to add
the last event. W have to make sure that all the related actions are complete
by the time we check the results.
Also the testForNoDuplicates checks 3 things. These should be separated out to
3 tests (interestingly one of the tests inside waits for an other one to
happen. This shows one of the problems with this practice).
> TestEventGeneration test is flakey
> ----------------------------------
>
> Key: OOZIE-2429
> URL: https://issues.apache.org/jira/browse/OOZIE-2429
> Project: Oozie
> Issue Type: Test
> Components: action, tests
> Affects Versions: trunk
> Reporter: Ferenc Denes
> Assignee: Ferenc Denes
> Priority: Minor
> Fix For: trunk
>
>
> TestEventGeneration's testForNoDuplicates fails tome to time depending on the
> circumstances of the test.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
