[
https://issues.apache.org/jira/browse/HDFS-14264?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16899452#comment-16899452
]
Amithsha commented on HDFS-14264:
---------------------------------
Yes this is due to no of disks and the usage, in our environment of 4.1Tb * 13
hard disks per node taking around 25min to report the FBR to NN. Since the yarn
also cohosted on the same node, at regular interval datanode utilising high
disk I/O and causing interupts to yarn applications.
> Datanode du -sk command is slow
> -------------------------------
>
> Key: HDFS-14264
> URL: https://issues.apache.org/jira/browse/HDFS-14264
> Project: Hadoop HDFS
> Issue Type: Improvement
> Affects Versions: 2.7.1
> Reporter: Amithsha
> Priority: Major
>
> Datanode consuming more time on du -sk command as well as creating heavy IO
> on disk. In our prod systems, each disk of dfs usage is 3Tb, to caculate it,
> the datanode will spend 10-20min Avg time. Also nodemanagers are running on
> the same box during this du -sk operation could see heavy IO on disk.
> Datanode should cache the usage and also not to be cleared by any other
> process.
--
This message was sent by Atlassian JIRA
(v7.6.14#76016)
---------------------------------------------------------------------
To unsubscribe, e-mail: hdfs-issues-unsubscr...@hadoop.apache.org
For additional commands, e-mail: hdfs-issues-h...@hadoop.apache.org
