[ 
https://issues.apache.org/jira/browse/HBASE-14082?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14628884#comment-14628884
 ] 

Elliott Clark commented on HBASE-14082:
---------------------------------------

bq.Is there any explicit stability promised for the JMX paths too to understand 
what options are open for addition of this data?
We've treated it as a breaking change before.  Transition from metrics1 to 
metrics2

bq.Elliott Clark the intention is to provide a way one can get region metrics 
rolled up by region server, region, replica, etc. putting it in another bean 
would not provide the desired region metrics differentiation by 
primary/secondary without having to replicate the whole bean. 
Sure there is. Have a bean that specifies which replica id this regionserver is 
hosting ( a region server can't host more than one replica of a region). Then 
when you poll jmx just join the metrics parsed from per region metrics with the 
replica id.

bq.Also, can you elaborate on how the addition of the replica ID would break 
your monitoring?
Internally we have a parser. I know there's also a tsdb tcollector that has the 
parsing. Pretty sure that other vendors also have the same.

> Add replica id to JMX metrics names
> -----------------------------------
>
>                 Key: HBASE-14082
>                 URL: https://issues.apache.org/jira/browse/HBASE-14082
>             Project: HBase
>          Issue Type: Improvement
>          Components: metrics
>            Reporter: Lei Chen
>            Assignee: Lei Chen
>         Attachments: HBASE-14082-v1.patch
>
>
> Today, via JMX, one cannot distinguish a primary region from a replica. A 
> possible solution is to add replica id to JMX metrics names. The benefits may 
> include, for example:
> # Knowing the latency of a read request on a replica region means the first 
> attempt to the primary region has timeout.
> # Write requests on replicas are due to the replication process, while the 
> ones on primary are from clients.
> # In case of looking for hot spots of read operations, replicas should be 
> excluded since TIMELINE reads are sent to all replicas.
> To implement, we can change the format of metrics names found at 
> {code}Hadoop->HBase->RegionServer->Regions->Attributes{code}
> from 
> {code}namespace_<namespace>_table_<tablename>_region_<regionname>_metric_<metricname>{code}
> to
> {code}namespace_<namespace>_table_<tablename>_region_<regionname>_replicaid_<replicaid>_metric_<metricname>{code}



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Reply via email to