[
https://issues.apache.org/jira/browse/YARN-8774?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16624367#comment-16624367
]
Tao Yang edited comment on YARN-8774 at 9/22/18 12:17 AM:
----------------------------------------------------------
Thanks [~eepayne] for the review.
Attached v2 patch to improve the UT with adding check for killed state.
There are somewhere different in branch-2.8 (it also exist in branch-2.8) and
branch-2, especially in branch2.8.
Backported this patch to these two branches and attached them also for review.
was (Author: tao yang):
Thanks [~eepayne] for the review.
Attached v2 patch to improve the UT with adding check for killed state.
There are somewhat different in branch-2.8 (it also exist in branch-2.8) and
branch-2, especially in branch2.8.
Backported this patch to these two branches and attached them also for review.
> Memory leak when CapacityScheduler allocates from reserved container with
> non-default label
> -------------------------------------------------------------------------------------------
>
> Key: YARN-8774
> URL: https://issues.apache.org/jira/browse/YARN-8774
> Project: Hadoop YARN
> Issue Type: Bug
> Components: capacityscheduler
> Affects Versions: 3.2.0, 2.8.5
> Reporter: Tao Yang
> Assignee: Tao Yang
> Priority: Critical
> Attachments: YARN-8774.001.patch, YARN-8774.002.patch,
> YARN-8774.branch-2.001.patch, YARN-8774.branch-2.8.001.patch
>
>
> The cause is that the RMContainerImpl instance of reserved container lost its
> node label expression, when scheduler reserves containers for non-default
> node-label requests, it will be wrongly added into
> LeafQueue#ignorePartitionExclusivityRMContainers and never be removed.
> To reproduce this memory leak:
> (1) create reserved container
> RegularContainerAllocator#doAllocation: create RMContainerImpl instanceA
> (nodeLabelExpression="")
> LeafQueue#allocateResource: RMContainerImpl instanceA is put into
> LeafQueue#ignorePartitionExclusivityRMContainers
> (2) allocate from reserved container
> RegularContainerAllocator#doAllocation: create RMContainerImpl instanceB
> (nodeLabelExpression="test-label")
> (3) From now on, RMContainerImpl instanceA will be left in memory (be kept in
> LeafQueue#ignorePartitionExclusivityRMContainers) forever until RM restarted
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
