[
https://issues.apache.org/jira/browse/YARN-3487?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14495137#comment-14495137
]
Wangda Tan commented on YARN-3487:
----------------------------------
[~jlowe], removing lock of checkAccess may cause consumer get partial queue's
ACL when reinitialize happens. See {{YarnAuthorizationProvider}},
{{CapacityScheduler#setQueueAcls}} will set ACL one by one, so checkAccess's
invoker can pull incomplete result.
To avoid this, one possible solute might be: add a reintialize_lock to CS, it
will be locked when doing reinitialize, and checkAccess will try to acquire it.
> CapacityScheduler scheduler lock obtained unnecessarily
> -------------------------------------------------------
>
> Key: YARN-3487
> URL: https://issues.apache.org/jira/browse/YARN-3487
> Project: Hadoop YARN
> Issue Type: Bug
> Components: capacityscheduler
> Affects Versions: 2.6.0
> Reporter: Jason Lowe
> Assignee: Jason Lowe
> Priority: Critical
> Attachments: YARN-3487.001.patch, YARN-3487.002.patch
>
>
> Recently saw a significant slowdown of applications on a large cluster, and
> we noticed there were a large number of blocked threads on the RM. Most of
> the blocked threads were waiting for the CapacityScheduler lock while calling
> getQueueInfo.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
