[
https://issues.apache.org/jira/browse/YARN-11848?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
chenruotao updated YARN-11848:
------------------------------
Target Version/s: 3.4.1, 3.1.4 (was: 3.1.4, 3.4.1)
Description:
when nodemanager restart, there are too many logs like
"2025-03-23 01:53:05,883 WARN [RM Event dispatcher] rmnode.RMNodeImpl
(RMNodeImpl.java:handleRunningAppOnNode(791)) - Cannot get RMApp by
appId=application_1737511638592_8413719, just added it to finishedApplications
list for cleanup"
print in resourcemanager.
when too many nodemanagers are restarted at one time, resourcemanager will be
blocked.
I think nodemanager should be able to determine whether an app has fininshed on
local and reduce the number of apps reported to resourcemanager.
https://github.com/apache/hadoop/pull/7861
was:
when nodemanager restart, there are too many logs like
"2025-03-23 01:53:05,883 WARN [RM Event dispatcher] rmnode.RMNodeImpl
(RMNodeImpl.java:handleRunningAppOnNode(791)) - Cannot get RMApp by
appId=application_1737511638592_8413719, just added it to finishedApplications
list for cleanup"
print in resourcemanager.
when too many nodemanagers are restarted at one time, resourcemanager will be
blocked.
I think nodemanager should be able to determine whether an app has fininshed on
local and reduce the number of apps reported to resourcemanager.
> Reduce the number of apps reported when nodemanager restart
> -----------------------------------------------------------
>
> Key: YARN-11848
> URL: https://issues.apache.org/jira/browse/YARN-11848
> Project: Hadoop YARN
> Issue Type: Improvement
> Components: nodemanager
> Affects Versions: 3.1.4, 3.4.1
> Reporter: chenruotao
> Priority: Major
> Labels: pull-request-available
>
> when nodemanager restart, there are too many logs like
> "2025-03-23 01:53:05,883 WARN [RM Event dispatcher] rmnode.RMNodeImpl
> (RMNodeImpl.java:handleRunningAppOnNode(791)) - Cannot get RMApp by
> appId=application_1737511638592_8413719, just added it to
> finishedApplications list for cleanup"
> print in resourcemanager.
> when too many nodemanagers are restarted at one time, resourcemanager will be
> blocked.
> I think nodemanager should be able to determine whether an app has fininshed
> on local and reduce the number of apps reported to resourcemanager.
> https://github.com/apache/hadoop/pull/7861
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: yarn-issues-h...@hadoop.apache.org
