[
https://issues.apache.org/jira/browse/HDFS-11449?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
James Clampffer updated HDFS-11449:
-----------------------------------
Resolution: Fixed
Status: Resolved (was: Patch Available)
+1, committed to HDFS-8707. Should make it much easier to detect load skew on
clusters.
I think in the future it'd be best to also implement the functions used for
gathering IO statistics like hdfsFileGetReadStatistics. That could be extended
to report amount read per datanode as well as latency/bandwidth info if we want
to be clever about how we chose DNs for future reads.
> libhdfs++: Log Datanode read size when reading an HDFS block
> ------------------------------------------------------------
>
> Key: HDFS-11449
> URL: https://issues.apache.org/jira/browse/HDFS-11449
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Reporter: Xiaowei Zhu
> Assignee: Xiaowei Zhu
> Priority: Trivial
> Attachments: HDFS-11449.HDFS-8707.000.patch
>
>
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
