You are viewing a plain text version of this content. The canonical link for it is here.
Posted to common-issues@hadoop.apache.org by "Dmytro Molkov (JIRA)" <ji...@apache.org> on 2009/12/24 02:24:29 UTC
[jira] Created: (HADOOP-6470) JMX Context for Metrics
JMX Context for Metrics
-----------------------
Key: HADOOP-6470
URL: https://issues.apache.org/jira/browse/HADOOP-6470
Project: Hadoop Common
Issue Type: New Feature
Reporter: Dmytro Molkov
The way metrics are currently exposed to the JMX in the NameNode is not helpful, since only the current counters in the record can be fetched and without any context those number mean little.
For example the number of files created equal to 150 only means that in the last period there were 150 files created but when the new period will end is unknown so fetching 150 again will either mean another 150 files or we are fetching the same time period.
One of the solutions for this problem will be to have a JMX context that will accumulate the data (being child class of AbstractMetricsContext) and expose different records to the JMX through custom MBeans. This way the information fetched from the JMX will represent the state of things in a more meaningful way.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Assigned: (HADOOP-6470) JMX Context for Metrics
Posted by "dhruba borthakur (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-6470?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
dhruba borthakur reassigned HADOOP-6470:
----------------------------------------
Assignee: Dmytro Molkov
> JMX Context for Metrics
> -----------------------
>
> Key: HADOOP-6470
> URL: https://issues.apache.org/jira/browse/HADOOP-6470
> Project: Hadoop Common
> Issue Type: New Feature
> Reporter: Dmytro Molkov
> Assignee: Dmytro Molkov
> Attachments: JMX.patch
>
>
> The way metrics are currently exposed to the JMX in the NameNode is not helpful, since only the current counters in the record can be fetched and without any context those number mean little.
> For example the number of files created equal to 150 only means that in the last period there were 150 files created but when the new period will end is unknown so fetching 150 again will either mean another 150 files or we are fetching the same time period.
> One of the solutions for this problem will be to have a JMX context that will accumulate the data (being child class of AbstractMetricsContext) and expose different records to the JMX through custom MBeans. This way the information fetched from the JMX will represent the state of things in a more meaningful way.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
[jira] Updated: (HADOOP-6470) JMX Context for Metrics
Posted by "Dmytro Molkov (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/HADOOP-6470?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Dmytro Molkov updated HADOOP-6470:
----------------------------------
Attachment: JMX.patch
I created the first iteration of the context for JMX, any comments are welcome.
Also it would be helpful if someone can come up with a way to test this context.
> JMX Context for Metrics
> -----------------------
>
> Key: HADOOP-6470
> URL: https://issues.apache.org/jira/browse/HADOOP-6470
> Project: Hadoop Common
> Issue Type: New Feature
> Reporter: Dmytro Molkov
> Attachments: JMX.patch
>
>
> The way metrics are currently exposed to the JMX in the NameNode is not helpful, since only the current counters in the record can be fetched and without any context those number mean little.
> For example the number of files created equal to 150 only means that in the last period there were 150 files created but when the new period will end is unknown so fetching 150 again will either mean another 150 files or we are fetching the same time period.
> One of the solutions for this problem will be to have a JMX context that will accumulate the data (being child class of AbstractMetricsContext) and expose different records to the JMX through custom MBeans. This way the information fetched from the JMX will represent the state of things in a more meaningful way.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.