[
https://issues.apache.org/jira/browse/HDFS-11461?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15886585#comment-15886585
]
Hanisha Koneru commented on HDFS-11461:
---------------------------------------
The intention here is to catch slow disks, as opposed to overloaded disks. So
it would not matter if the disk traffic is actually from HDFS or not. We want
to find out the overall latency of disk over a long time interval to avoid
false positives due to transient traffic.
This information will only be exposed via jmx so that admins can get a
reference on potentially slow disks. They can then run diagnostics or take
whatever action as deemed fit.
> DataNode Disk Outlier Detection
> -------------------------------
>
> Key: HDFS-11461
> URL: https://issues.apache.org/jira/browse/HDFS-11461
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: hdfs
> Reporter: Hanisha Koneru
> Assignee: Hanisha Koneru
> Attachments: HDFS-11461.000.patch
>
>
> Similar to how DataNodes collect peer performance statistics, we can collect
> disk performance statistics per datanode and detect outliers among them, if
> any.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
