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 "ruiliang (Jira)" <ji...@apache.org> on 2022/07/20 06:26:00 UTC
[jira] [Commented] (YARN-11209) yarn CapacityScheduler Unknown bug: Resource scheduling fails after RM runs for 3 days and resources cannot be fully used.
[ https://issues.apache.org/jira/browse/YARN-11209?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17568859#comment-17568859 ]
ruiliang commented on YARN-11209:
---------------------------------
In our scenario, the CodeCache problem causes the overall YARN scheduling efficiency to decrease. Do you want to increase the default yarn configuration to avoid the bug that causes CodeCache to trigger JVM?
The following figure shows the comparison of operating efficiency after adjusting parameters
-XX:CICompilerCount=8 -XX:ReservedCodeCacheSize=2G -XX:InitialCodeCacheSize=512M -XX:-UseCodeCacheFlushing
!QQ截图20220720142031.png!
> yarn CapacityScheduler Unknown bug: Resource scheduling fails after RM runs for 3 days and resources cannot be fully used.
> ---------------------------------------------------------------------------------------------------------------------------
>
> Key: YARN-11209
> URL: https://issues.apache.org/jira/browse/YARN-11209
> Project: Hadoop YARN
> Issue Type: Bug
> Components: capacity scheduler
> Affects Versions: 3.1.1
> Reporter: ruiliang
> Priority: Major
> Labels: capacity-scheduler, yarn
> Attachments: QQ截图20220720142031.png, image-2022-07-12-14-51-15-129.png, rm_20220711_1646_have_bug.jstack, rm_20220712_1424_run well.jstac
>
>
> * !image-2022-07-12-14-51-15-129.png|width=1667,height=670!
> * Here are the jStacks where resource scheduling does not reach 100%
> * rm_20220711_1646_have_bug.jstack
>
> restart rm After 3 days of good operation
> rm_20220712_1424_run well.jstack
> At present, I do not know what the problem is. Could you please tell me what configuration needs to be adjusted?There is no configuration change before and after the restart, but this problem is really prominent, request guidance
> CPU before and after comparison is obvious, other hard indicators are not clear display abnormal
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: yarn-issues-unsubscribe@hadoop.apache.org
For additional commands, e-mail: yarn-issues-help@hadoop.apache.org