[
https://issues.apache.org/jira/browse/YARN-5176?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15319139#comment-15319139
]
Arun Suresh commented on YARN-5176:
-----------------------------------
Thanks for the patch [~kkaranasos].. and the thorough testing..
The original intent of the {{TestQueuingContainerManager}} being a subclass of
{{TestContainerManager}} was, I believe, to run all its testcases as well. But
I understand the rational for changing that hierarchy, given the different
configurations required for the specific testcases addresses in this patch.
I propose we have a parametrized testcase similar to what is done for the
Schedulers to test the common scenarios for both ContainerManagers. Can we have
a JIRA to track that ?
A minor nit otherwise is that the current test failure seems to not be a timing
issue. If that is a valid state, maybe you should assert that the final state
can be either *DONE* or *CONTAINER_CLEANEDUP_AFTER_KILL*
+1 pending the above..
> More test cases for queuing of containers at the NM
> ---------------------------------------------------
>
> Key: YARN-5176
> URL: https://issues.apache.org/jira/browse/YARN-5176
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Konstantinos Karanasos
> Assignee: Konstantinos Karanasos
> Attachments: YARN-5176.001.patch, YARN-5176.002.patch
>
>
> Extending {{TestQueuingContainerManagerImpl}} to include more test cases for
> the queuing of containers at the NM.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
