[
https://issues.apache.org/jira/browse/YARN-5090?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15287231#comment-15287231
]
Wangda Tan commented on YARN-5090:
----------------------------------
Thanks [~asuresh],
Patch generally looks good, only few suggestions:
1) Instead of putting e2e tests to yarn-client project, could we put the tests
to TestDistributedSchedulingService?
2) Most of the test failures look like not related, the only one may need to be
more careful is: TestMiniYarnClusterNodeUtilization. Sometimes failures on
Jenkins cannot be reproduced locally, my understanding is Jenkins VMs are
sometimes slower than local laptop.
> Add End-to-End test-cases for DistributedScheduling using MiniYarnCluster
> --------------------------------------------------------------------------
>
> Key: YARN-5090
> URL: https://issues.apache.org/jira/browse/YARN-5090
> Project: Hadoop YARN
> Issue Type: Sub-task
> Reporter: Arun Suresh
> Assignee: Arun Suresh
> Attachments: YARN-5090.001.patch, YARN-5090.002.patch,
> YARN-5090.002.patch, mvn_test.txt
>
>
> Validates End2End Distributed Scheduling flow which includes
> # the AM specifying *OPPORTUNISTIC* containers in its resource requests
> (YARN-2882, YARN-4335)
> # the Request intercepted by the {{AMRMProxyService}} running on the NM and
> handled by the {{LocalScheduler}} RequestInterceptor (YARN-2884, YARN-2885)
> # use of the {{DistributedSchedulingProtocol}} to talk to the
> {{DistributedSchedulingService}} running on the RM to obtain list of lightly
> loaded Nodes (YARN-2885, YARN-4412)
> # use of the {{OpportunisticContainerAllocator}} to create Container tokens
> which are returned to the AM (YARN-2885)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
