[
https://issues.apache.org/jira/browse/HADOOP-1562?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
David Bowen updated HADOOP-1562:
--------------------------------
Attachment: 1562_2.patch
Here is an updated patch with the following changes:
1 There is now a static method for initiating JVM metric reporting. There
is actually no need to keep a reference to the JvmMetrics object (since it will
be referenced by a Timer that it creates, and that timer runs for the lifetime
of the process). So I took out the fields that were keeping such references
(and which were causing warnings).
2 Where possible, I put the call to initiate JVM metrics inside the class
that is collecting other metrics, so as to minimize the intrusion of
metrics-related stuff in the main body of the code.
3 DFS metrics are now tagged with a session id if one is provided in the
hadoop-site.xml file. This will be the case when HOD provisions DFS.
4 Added JVM metric reporting to the task sub-processes (i.e.
TaskTracker.Child).
> Report Java VM metrics
> ----------------------
>
> Key: HADOOP-1562
> URL: https://issues.apache.org/jira/browse/HADOOP-1562
> Project: Hadoop
> Issue Type: New Feature
> Components: metrics
> Reporter: David Bowen
> Assignee: David Bowen
> Attachments: 1562.patch, 1562_1.patch, 1562_2.patch
>
>
> It would be useful to have each Java process in Hadoop (JobTracker,
> TaskTracker, NameNode and DataNode) report some Java VM metrics. E.g.
> heap/non-heap memory used/committed, number of garbage collections and
> percentage of time spent in GC, number of threads that are
> runnable/blocked/waiting/etc.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
