Add metrics naming convention for prometheus
Describe the proposal
Add prefixes (based on business scenarios)
For provider: dubbo.provider, such as dubbo.provider.rt.qps.seconds
For consumer: dubbo.consumer, such as dubbo. consumer.rt.qps.seconds
For registry: dubbo.registry, such as dubbo.registry.requests.total
Add suffix
Step 1: add base-units
- Time: suffix with
seconds, such asdubbo.rt.qps.seconds - Length: meters
- Bytes: bytes ...
Step 2: add suffix by metrics type
- Counter: suffix with
total, such asdubbo.rt.requests.total
Step 3: add other suffix
- The num of metrics:
count, such asdubbo.rt.seconds.count - Summary value:
sum, such asdubbo.rt.seconds.sum - Minimum value:
min, such asdubbo.rt.seconds.min - Maximum value:
max, such asdubbo.rt.seconds.max - Average value:
avg, such asdubbo.rt.seconds.avg - P99 value:
avg, such asdubbo.rt.seconds.p99 - P95 value:
avg, such asdubbo.rt.seconds.p95
It can be distinguished between the lower dubbo buried point specification and the Prometheus buried point specification, such as the second 可以区分下dubbo埋点规范和普罗米修斯埋点规范例如第二个
I change title "for prometheus".
我修改了标题加上了 prometheus
可以区分下场景,比如dubbo.registry. , dubbo.thread.
可以区分下场景,比如dubbo.registry. , dubbo.thread.
dubbo.thread 这个使用在哪个场景?我补充到文档上
可以区分下场景,比如dubbo.registry. , dubbo.thread.
I change title "for prometheus".
我修改了标题加上了 prometheus
可以区分下场景,比如dubbo.registry. , dubbo.thread.
dubbo.thread 这个使用在哪个场景?我补充到文档上
不同场景下的一种区分,比如 provider, dubbo.provider,类似