[
https://issues.apache.org/jira/browse/MAPREDUCE-2558?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13042846#comment-13042846
]
Luke Lu commented on MAPREDUCE-2558:
------------------------------------
Comments on patch v2:
The queue level instrumentation class configuration is not a requirement. If
you don't want to test the mechanism I'd suggest that you remove the related
code (you can still keep the instrumentation methods, it's the abstract class
and class loading stuff that are unnecessary).
Looking up the queue from queue manager for metrics for every metrics update
seems unnecessary as well. You can add a final Queue field in JIP and
initialize it in the JIP ctor.
You need a unique name (which also maps to a MBean name) for every queue
metrics so you don't have naming conflicts when you have multiple queues. An
MBean/jconsole friendly QueueInstrumentation#create would look like this:
{code}
return ms.register("QueueMetrics,q="+ queueName, "Queue metrics",
new QueueMetricsSource(queueName, conf));
{code}
> Add queue-level metrics 0.20-security branch
> --------------------------------------------
>
> Key: MAPREDUCE-2558
> URL: https://issues.apache.org/jira/browse/MAPREDUCE-2558
> Project: Hadoop Map/Reduce
> Issue Type: New Feature
> Components: jobtracker
> Affects Versions: 0.20.204.0
> Reporter: Jeffrey Naisbitt
> Assignee: Jeffrey Naisbitt
> Fix For: 0.20.205.0
>
> Attachments: queue-metrics-v2.patch, queue-metrics.patch
>
>
> We would like to record and present the jobtracker metrics on a per-queue
> basis.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
