ambari-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Xiaobing Zhou (JIRA)" <>
Subject [jira] [Commented] (AMBARI-18694) DataNode JVM heap settings should include CMSInitiatingOccupancy
Date Fri, 28 Oct 2016 20:14:59 GMT


Xiaobing Zhou commented on AMBARI-18694:

I posted v001 patch by changing -XX:ConcGCThreads=8 to -XX:ConcGCThreads=4, which is more
optimal per offline discussion with [~jaimin], thanks.

> DataNode JVM heap settings should include CMSInitiatingOccupancy
> ----------------------------------------------------------------
>                 Key: AMBARI-18694
>                 URL:
>             Project: Ambari
>          Issue Type: Improvement
>            Reporter: Xiaobing Zhou
>            Assignee: Xiaobing Zhou
>         Attachments: AMBARI-18694.000.patch, AMBARI-18694.001.patch
> As HDFS-11047 reported, DirectoryScanner does scan by deep copying FinalizedReplica.
In a deployment with 500,000+ blocks, we've seen the DN heap usage being accumulated to high
peaks very quickly. Deep copies of FinalizedReplica will make DN heap usage even worse if
directory scans are scheduled more frequently. 
> Another factor is that huge number of ScanInfo instances corresponding to HDFS blocks
are lingering in garbage to eat many heap memories until a full GC takes place.
> This proposes adding JVM settings to force GC more frequently to release DataNode heap
consumed as a result of two aforementioned reasons, i.e. add the options to HADOOP_DATANODE_OPTS
> {noformat}
> -XX:CMSInitiatingOccupancyFraction=70 -XX:+UseCMSInitiatingOccupancyOnly -XX:ConcGCThreads=8
> {noformat}

This message was sent by Atlassian JIRA

View raw message