[
https://issues.apache.org/jira/browse/HDFS-5180?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13763663#comment-13763663
]
Todd Lipcon commented on HDFS-5180:
-----------------------------------
I don't know that the audit log is the right place for this type of
information. I'd rather see a separate "slow RPC log" implemented -- HBase has
a function like this and it has been helpful for troubleshooting places where
high latency spikes happen. Putting it at the RPC layer also helps because you
can take into account queueTime separately than processingTime, and handle it
generically for all services rather than having to hook into each audit event.
> Add time taken to process the command to audit log
> --------------------------------------------------
>
> Key: HDFS-5180
> URL: https://issues.apache.org/jira/browse/HDFS-5180
> Project: Hadoop HDFS
> Issue Type: Improvement
> Components: namenode
> Affects Versions: 3.0.0
> Reporter: Shinichi Yamashita
>
> Command and ugi are output now by audit log of NameNode. But it is not output
> for the processing time of command to audit log.
> For example, we must check which command is a problem when a trouble such as
> the slow down occurred in NameNode.
> It should add the processing time to audit log to know the abnormal sign.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
