[
https://issues.apache.org/jira/browse/FALCON-584?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14092768#comment-14092768
]
Paul Isaychuk commented on FALCON-584:
--------------------------------------
Of course. So InstanceUtil.waitTillInstanceReachState(..) method, which is used
in this test, contains a validation ([these
lines|https://github.com/pisaychuk/incubator-falcon/blob/master/falcon-regression/merlin-core/src/main/java/org/apache/falcon/regression/core/util/InstanceUtil.java#L1084-L1086])
where job status is checked to avoid working with another statuses then
RUNNING, PREP, SUCCEEDED (mean that job is live). And then particular
instances(actions) statuses are checked . In case when we expect TIMEDOUT
status of that instances (actions) status of job would be DONEWITHERROR
(according to
[doc|https://oozie.apache.org/docs/3.1.3-incubating/BundleFunctionalSpec.html#a4.1._Bundle_Job_Status]
there can't be TIMEDOUT status of that job, so the only appropriate is
DONEWITHERROR status). So in our test when all instances reach TIMEDOUT status,
job reaches DONEWITHERROR and test fails on mentioned validation. As for me,
test passes in cases when job status hasn't been generated yet but instances
status are already TIMEDOUT.
Makes sense?
> Fix of ProcessInstanceRerunTest#testProcessInstanceRerunTimedOut
> ----------------------------------------------------------------
>
> Key: FALCON-584
> URL: https://issues.apache.org/jira/browse/FALCON-584
> Project: Falcon
> Issue Type: Bug
> Components: merlin
> Reporter: Paul Isaychuk
> Attachments: falcon-584.diff
>
>
> Test intermittently fails (expecting process instances to reach TIMEDOUT
> state) bundle job got into status DONEWITHERROR. But it should be so.
> TIMEDOUT status of instances can't generate another job status.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
