ambari-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Andrew Onischuk (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (AMBARI-18006) Tune HDFS opts parameters to trigger GC more predicitably
Date Wed, 03 Aug 2016 14:38:20 GMT

    [ https://issues.apache.org/jira/browse/AMBARI-18006?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15406003#comment-15406003
] 

Andrew Onischuk commented on AMBARI-18006:
------------------------------------------

The above failure is caused by jenkins bug. Ran the tests manually:
{noformat}
[INFO] Rat check: Summary of files. Unapproved: 0 unknown: 0 generated: 0 approved: 148 licence.
[INFO] ------------------------------------------------------------------------
[INFO] Reactor Summary:
[INFO] 
[INFO] Ambari Views ...................................... SUCCESS [2.150s]
[INFO] Ambari Metrics Common ............................. SUCCESS [1.209s]
[INFO] Ambari Server ..................................... SUCCESS [1:02.528s]
[INFO] Ambari Agent ...................................... SUCCESS [14.786s]
[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 1:21.439s
[INFO] Finished at: Wed Aug 03 17:36:37 EEST 2016
[INFO] Final Memory: 86M/1115M
[INFO] ------------------------------------------------------------------------
{noformat}

> Tune HDFS opts parameters to trigger GC more predicitably
> ---------------------------------------------------------
>
>                 Key: AMBARI-18006
>                 URL: https://issues.apache.org/jira/browse/AMBARI-18006
>             Project: Ambari
>          Issue Type: Bug
>            Reporter: Andrew Onischuk
>            Assignee: Andrew Onischuk
>             Fix For: 3.0.0
>
>         Attachments: AMBARI-18006.patch
>
>
> DN Heap usage alert is set to 80% warning and 90% critical. This alert is
> causing a lot of alerts to be fired off and needs to be tuned.
> One thing we need to make sure is that DN gc is happening. I am not sure at
> what % of heap usage will the gc be fired but we need to make sure our alerts
> are for a % higher than that as when i manually ran a GC the heap usage went
> below the threshold.
> Based on discussions with hdfs dev's it was determined we should be adding the
> following
>     
>     
>     
>     XX:+UseCMSInitiatingOccupancyOnly and -XX:CMSInitiatingOccupancyFraction=<percent>
>     
> to the namenode and datanode opts. Based on this setting hopefully we can also
> determine what % to set the alerts to.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message