[
https://issues.apache.org/jira/browse/YARN-4722?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15159194#comment-15159194
]
Sangjin Lee commented on YARN-4722:
-----------------------------------
We've been seeing this too, and it looks like a good improvement over the
current behavior.
A small question: do you think is volatile is needed for these variables? If
these variables are accessed strictly from a single thread, then volatile can
be removed? Probably it's not a big deal either way, but it might be good to
see whether it's needed or not.
> AsyncDispatcher logs redundant event queue sizes
> ------------------------------------------------
>
> Key: YARN-4722
> URL: https://issues.apache.org/jira/browse/YARN-4722
> Project: Hadoop YARN
> Issue Type: Bug
> Affects Versions: 2.6.0
> Reporter: Jason Lowe
> Assignee: Jason Lowe
> Attachments: YARN-4722.001.patch
>
>
> A fairly common occurrence in RM logs is a string of redundant event-queue
> logs like the following which does little except bloat the logs:
> {noformat}
> 2016-02-23 08:00:00,948 [IPC Server handler 36 on 8030] INFO
> event.AsyncDispatcher: Size of event-queue is 1000
> 2016-02-23 08:00:00,948 [IPC Server handler 36 on 8030] INFO
> event.AsyncDispatcher: Size of event-queue is 1000
> 2016-02-23 08:00:00,948 [IPC Server handler 36 on 8030] INFO
> event.AsyncDispatcher: Size of event-queue is 1000
> 2016-02-23 08:00:00,948 [IPC Server handler 36 on 8030] INFO
> event.AsyncDispatcher: Size of event-queue is 1000
> 2016-02-23 08:00:00,948 [IPC Server handler 36 on 8030] INFO
> event.AsyncDispatcher: Size of event-queue is 1000
> [...]
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
