[
https://issues.apache.org/jira/browse/YARN-6024?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15771071#comment-15771071
]
Wangda Tan commented on YARN-6024:
----------------------------------
The issue exists in branch-2.7, and not existed in branch-2.8/branch-2/trunk.
The root cause is, in AbstractCSQueue#canAssignToThisQueue:
{code}
if (Resources.lessThan(resourceCalculator, clusterResource,
newTotalWithoutReservedResource, currentLimitResource)) {
if (LOG.isDebugEnabled()) {
LOG.debug("try to use reserved: " + getQueueName()
+ " usedResources: " + queueUsage.getUsed()
+ ", clusterResources: " + clusterResource
+ ", reservedResources: " + resourceCouldBeUnreserved
+ ", capacity-without-reserved: "
+ newTotalWithoutReservedResource + ", maxLimitCapacity: "
+ currentLimitResource);
}
currentResourceLimits.setAmountNeededUnreserve(Resources.subtract(newTotalResource,
currentLimitResource));
return true;
}
{code}
It should be {{lessThanOrEqual}} instead of {{lessThan}}.
But we need to backport test cases to branch-2.8/branch-2/trunk to prevent
future changes.
> Capacity Scheduler continuous reservation looking doesn't work when queue's
> used+reserved = max
> -----------------------------------------------------------------------------------------------
>
> Key: YARN-6024
> URL: https://issues.apache.org/jira/browse/YARN-6024
> Project: Hadoop YARN
> Issue Type: Bug
> Reporter: Wangda Tan
> Assignee: Wangda Tan
>
> Found one corner case when continuous reservation looking doesn't work:
> When queue's used=max, the queue's capacity check fails.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
