[
https://issues.apache.org/jira/browse/AMBARI-18694?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15616659#comment-15616659
]
Jaimin Jetly commented on AMBARI-18694:
---------------------------------------
+1 for the patch
> DataNode JVM heap settings should include CMSInitiatingOccupancy
> ----------------------------------------------------------------
>
> Key: AMBARI-18694
> URL: https://issues.apache.org/jira/browse/AMBARI-18694
> Project: Ambari
> Issue Type: Improvement
> Affects Versions: 2.2.2
> Reporter: Xiaobing Zhou
> Assignee: Xiaobing Zhou
> Fix For: 2.5.0
>
> 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 -XX:+UseConcMarkSweepGC
> {noformat}
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
