[
https://issues.apache.org/jira/browse/MAPREDUCE-5446?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Jason Lowe updated MAPREDUCE-5446:
----------------------------------
Attachment: MAPREDUCE-5446.patch
Here's a patch that has MRApp.waitForState using waitForServiceToStop when
waiting for the STOPPED state so it properly waits until the service has fully
stopped.
My preference would be to rip out MRApp.waitForState(Service.STATE) completely
and replace it with something like waitForStopped since it only appears to be
used for that case. However I'm not sure if there are downstream projects that
are using MRApp, so this patch plays it safe by preserving the existing MRApp
interface.
> TestJobHistoryEvents and TestJobHistoryParsing have race conditions
> -------------------------------------------------------------------
>
> Key: MAPREDUCE-5446
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-5446
> Project: Hadoop Map/Reduce
> Issue Type: Bug
> Components: mrv2, test
> Affects Versions: 2.1.0-beta
> Reporter: Jason Lowe
> Assignee: Jason Lowe
> Attachments: MAPREDUCE-5446.patch
>
>
> TestJobHistoryEvents and TestJobHistoryParsing are not properly waiting for
> MRApp to finish. Currently they are polling the service state looking for
> Service.STATE.STOPPED, but the service can appear to be in that state
> *before* it is fully stopped. This causes tests to finish with MRApp threads
> still in-flight, and those threads can conflict with subsequent tests when
> they collide in the filesystem.
--
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
