[
https://issues.apache.org/jira/browse/CASSANDRA-4009?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13267211#comment-13267211
]
Yuki Morishita commented on CASSANDRA-4009:
-------------------------------------------
Here is my current plan:
* Separate out metrics from current MBeans and move them to o.a.c.metrics
Currently most of MBeans have metrics, configurations and operations in the
same place. Our first step is to pull out metrics and reconstruct them into
o.a.c.metrics using Codahale Metrics. Current metrics interfaces will be marked
as deprecated but not removed. Configurations(properties like
minCompactionThreshold, which you can update through JMX) and operations stay
as they are.
* Migrate tools to use new o.a.c.metrics
And some versions later,
* Remove deprecated interface
o.a.c.metrics will have following classes.
* CacheMetrics
Cache metrics for each key/row cache. Metrics from CacheServiceMBean go here.
* ClientRequestMetrics (from 1.1.0)
Port metrics from StorageProxy to this metrics.
* ColumnFamilyMetrics
Metrics for each column family. Metrics from ColumnFamilyStoreMBean go here.
* CommitLogMetrics
Metrics from CommitLogMBean go here.
* CompactionMetrics
Metrics from CompactionManagerMBean go here.
* LatencyMetrics
Replacement for current LatencyTracker. Used by other metrics.
* MessagingMetrics
Metrics from MessagingServiceMBean go here.
* ThreadPoolMetrics
Metrics from JMXEnabledThreadPoolExecutorMBean go here.
* FailureDetectorMetrics, StreamingMetrics, DynamicEndpointSnitchMetrics
and so on...
Note that Codahale Metrics has own Histogram object but I'm thinking to stick
with current EstimatedHistogram since it's hard to implement recent histogram
(like recentSSTablesPerReadHistogram) which you have to reset histogram after
reading its values.
> Increase usage of Metrics and flesh out o.a.c.metrics
> -----------------------------------------------------
>
> Key: CASSANDRA-4009
> URL: https://issues.apache.org/jira/browse/CASSANDRA-4009
> Project: Cassandra
> Issue Type: Improvement
> Reporter: Brandon Williams
> Assignee: Yuki Morishita
> Fix For: 1.1.1
>
>
> With CASSANDRA-3671 we have begun using the Metrics packages to expose stats
> in a new JMX structure, intended to be more user-friendly (for example, you
> don't need to know what a StorageProxy is or does.) This ticket serves as a
> parent for subtasks to finish fleshing out the rest of the enhanced metrics.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
