[ 
https://issues.apache.org/jira/browse/HBASE-5186?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13195690#comment-13195690
 ] 

Phabricator commented on HBASE-5186:
------------------------------------

tedyu has commented on the revision "HBASE-5186 [jira] Add metrics to 
ThriftServer".

INLINE COMMENTS
  src/main/java/org/apache/hadoop/hbase/thrift/ThriftMetrics.java:2 Year is not 
needed.
  src/main/java/org/apache/hadoop/hbase/thrift/ThriftMetrics.java:44 Would 
ThriftMetrics be created for thrift2 ?
  Fine if you plan to do it in another JIRA.
  src/main/java/org/apache/hadoop/hbase/thrift/ThriftMetrics.java:53 As the 
code on line 102 shows, "hbase.thrift.slow.response.nano.second" may be a 
better name for this config.
  src/main/java/org/apache/hadoop/hbase/thrift/CallQueue.java:57 Would 
timeInQueue be a better method name ?
  src/main/java/org/apache/hadoop/hbase/thrift/CallQueue.java:29 Does the call 
queue length in metrics match 'total elements removed' in javadoc ?
  src/main/java/org/apache/hadoop/hbase/thrift/CallQueue.java:128 According to 
http://docs.oracle.com/javase/6/docs/api/java/util/concurrent/BlockingQueue.html#drainTo%28java.util.Collection%29:

  Attempts to drain a queue to itself result in IllegalArgumentException.

  We should maintain the above semantics.
  src/main/java/org/apache/hadoop/hbase/thrift/CallQueue.java:153 Can we add 
the new Call object to underlyingQueue ?
  This way we don't need to create toBeAdded

REVISION DETAIL
  https://reviews.facebook.net/D1461

                
> Add metrics to ThriftServer
> ---------------------------
>
>                 Key: HBASE-5186
>                 URL: https://issues.apache.org/jira/browse/HBASE-5186
>             Project: HBase
>          Issue Type: Improvement
>            Reporter: Scott Chen
>            Assignee: Scott Chen
>         Attachments: HBASE-5186.D1461.1.patch, HBASE-5186.D1461.2.patch, 
> HBASE-5186.D1461.3.patch, HBASE-5186.D1461.4.patch, HBASE-5186.D1461.5.patch
>
>
> It will be useful to have some metrics (queue length, waiting time, 
> processing time ...) similar to Hadoop RPC server. This allows us to monitor 
> system health also provide a tool to diagnose the problem where thrift calls 
> are slow.

--
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

        

Reply via email to