[
https://issues.apache.org/jira/browse/YARN-7562?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16268643#comment-16268643
]
chuanjie.duan commented on YARN-7562:
-------------------------------------
Sorry, I didn't paste full configuration, I have updated. What i expect is:
user cim submit job, match root.cim queue(leaf queue).
algo like a department, testa and testb are teams under department, each
department has only one hadoop user -- algo, so algo sometimes would "set
mapreduce.job.queuename=root.algo.testa " and submit their job, expect queue
is "root.algo.testa" not a parent queue "root.algo".
In my opinion, the rule can be more intelligent:
case user, primaryGroup, secondaryGroupExistingQueue, nestedUserQueue , it
should skip if queuename is a parent; (parent queue is meanless for user)
case specified, returning an "not a leaf queue" exception is ok.
I updated my patch
> queuePlacementPolicy should not match parent queue
> --------------------------------------------------
>
> Key: YARN-7562
> URL: https://issues.apache.org/jira/browse/YARN-7562
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: fairscheduler, resourcemanager
> Affects Versions: 2.7.1
> Reporter: chuanjie.duan
> Attachments: YARN-7562.002.patch, YARN-7562.patch
>
>
> User algo submit a mapreduce job, console log said "root.algo is not a leaf
> queue exception".
> root.algo is a parent queue, it's meanless for me. Not sure why parent queue
> added before
> <queue name="default">
> <minResources>3000 mb, 1 vcores</minResources>
> <maxResources>24000 mb, 8 vcores</maxResources>
> <maxRunningApps>4</maxRunningApps>
> <weight>1</weight>
> <schedulingMode>fifo</schedulingMode>
> </queue>
> <queue name="cim">
> <minResources>3000 mb, 1 vcores</minResources>
> <maxResources>24000 mb, 8 vcores</maxResources>
> <maxRunningApps>4</maxRunningApps>
> <weight>1</weight>
> <schedulingMode>fifo</schedulingMode>
> </queue>
> <queue name="algo">
> <minSharePreemptionTimeout>300</minSharePreemptionTimeout>
> <minResources>40000 mb, 10 vcores</minResources>
> <maxResources>200000 mb, 60 vcores</maxResources>
> <queue name="testb">
> <minSharePreemptionTimeout>300</minSharePreemptionTimeout>
> <minResources>40000 mb, 10 vcores</minResources>
> <maxResources>100000 mb, 30 vcores</maxResources>
> <maxRunningApps>20</maxRunningApps>
> <schedulingMode>fifo</schedulingMode>
> <weight>4</weight>
> </queue>
> <queue name="testa">
> <minSharePreemptionTimeout>300</minSharePreemptionTimeout>
> <minResources>40000 mb, 10 vcores</minResources>
> <maxResources>100000 mb, 30 vcores</maxResources>
> <maxRunningApps>20</maxRunningApps>
> <schedulingMode>fifo</schedulingMode>
> <weight>4</weight>
> </queue>
> </queue>
> <queuePlacementPolicy>
> <rule name="user" create="false" />
> <rule name="specified" />
> <rule name="default" queue="default"/>
> </queuePlacementPolicy>
--
This message was sent by Atlassian JIRA
(v6.4.14#64029)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
