You are viewing a plain text version of this content. The canonical link for it is here.
Posted to yarn-issues@hadoop.apache.org by "Xie YiFan (Jira)" <ji...@apache.org> on 2020/12/18 03:49:00 UTC
[jira] [Updated] (YARN-10537) Change type of LogAggregationService
threadPool
[ https://issues.apache.org/jira/browse/YARN-10537?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Xie YiFan updated YARN-10537:
-----------------------------
Priority: Minor (was: Major)
> Change type of LogAggregationService threadPool
> -----------------------------------------------
>
> Key: YARN-10537
> URL: https://issues.apache.org/jira/browse/YARN-10537
> Project: Hadoop YARN
> Issue Type: Improvement
> Reporter: Xie YiFan
> Priority: Minor
>
> Now, LogAggregationService threadPool is FixedThreadPool which of default threadPoolSize is 100. LogAggregationService will construct AppLogAggregator for new come application and submit to threadPool. AppLogAggregator do while loop unitl application finished. Some application may run very long time due to reason such as no enough resource or other. As result, it occupy one thread of threadPool. When this application size greater than threadPoolSize, the later short-live application can't upload logs until previous long-live application finished. So, i think we should replace FixedThreadPool to CachedThreadPool.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org