[
https://issues.apache.org/jira/browse/HBASE-15163?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15116727#comment-15116727
]
Yu Li commented on HBASE-15163:
-------------------------------
bq. No they are continuous counting metrics. If the metrics snapshots once when
there's a spike and another time when there's not then the spike isn't lost,
just averaged out.
Let me confirm, from the codes in MetricsRegionServerWrapperImpl, the
read/writeRequestCount will be computed every
{{hbase.regionserver.metrics.period}}, 5s by default. This is what you mean
'continuous couting', right?
For the newly added metrics, they keep increasing and we'll get the value in
each snapshot, this is what you mean 'rate that's computed once per snapshot'?
But it's not a rate but a continuous count?:
{code}
@Override
public long getRpcGetRequestsCount() {
return regionServer.rpcServices.rpcGetRequestCount.get();
}
@Override
public GetResponse get(final RpcController controller,
final GetRequest request) throws ServiceException {
long before = EnvironmentEdgeManager.currentTime();
OperationQuota quota = null;
try {
checkOpen();
requestCount.increment();
rpcGetRequestCount.increment();
...
}
{code}
> Add sampling code and metrics for get/scan/multi/mutate count separately
> ------------------------------------------------------------------------
>
> Key: HBASE-15163
> URL: https://issues.apache.org/jira/browse/HBASE-15163
> Project: HBase
> Issue Type: Sub-task
> Reporter: Yu Li
> Assignee: Yu Li
> Attachments: DifferentRequestQPS.png, HBASE-15163.patch
>
>
> This way we could see QPS of different kinds of requests, to better analyze
> what's causing hot spot in system
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
