抽象监控上报，增加推送到Prometheus的模块

[konglz]

Describe what this PR does / why we need it

• 抽象监控上报

• 增加推送到Prometheus的模块

Does this pull request fix one issue?

Describe how you did it

• 抽象监控上报，修改了sentinel-core

• 增加推送到Prometheus的模块：sentinel-transport-prometheus-push

Describe how to verify it

• 配置好Prometheus Server和Prometheus PushGateway

• 增加一对JVM参数，例如：-Dsentinel.prometheus.gateway=127.0.0.1:9091 -Dsentinel.web.port=8080

  • sentinel.prometheus.gateway是Prometheus PushGateway的ip和端口号，不传默认取127.0.0.1:9091

  • sentinel.web.port是应用的web端口号，如无web端口，随便传1个即可，不传默认取0

• 在Prometheus Server上收到5个标签，分别是：

  • SENTINEL_BLOCKED_QPS

  • SENTINEL_EXCEPTION_COUNT

  • SENTINEL_PASSED_QPS

  • SENTINEL_RT

  • SENTINEL_SUCCESS_QPS

• 在Prometheus Server上看到，每个标签包含如下label：

  • language=java

  • appName

  • host

  • port

  • instance=ip:port

  • resource

Special notes for reviews

[CLAassistant]

All committers have signed the CLA.

[codecov-io]

Codecov Report

❗ No coverage uploaded for pull request base (master@257dde1). Click here to learn what that means.
The diff coverage is 36.39%.

@@            Coverage Diff            @@
##             master     #167   +/-   ##
=========================================
  Coverage          ?   50.76%           
  Complexity        ?      822           
=========================================
  Files             ?      140           
  Lines             ?     4759           
  Branches          ?      674           
=========================================
  Hits              ?     2416           
  Misses            ?     2059           
  Partials          ?      284

Impacted Files	Coverage Δ	Complexity Δ
...entinel/node/metric/MetricSearcherDefaultImpl.java	10.46% <10.46%> (ø)	3 <3> (?)
.../csp/sentinel/node/metric/MetricTimerListener.java	92% <100%> (ø)	8 <1> (?)
.../sentinel/node/metric/MetricWriterDefaultImpl.java	48.34% <48.34%> (ø)	19 <19> (?)
...libaba/csp/sentinel/node/metric/MetricManager.java	52.17% <52.17%> (ø)	2 <2> (?)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 257dde1...29f2150. Read the comment docs.

[sczyh30]

Hi, thanks for your contribution. As this PR can bring breaking changes, you'll need to open an issue related to this feature, then illustrate the thoughts and design. It's not recommended to submit a large PR directly without discussion. A new feature must be well designed and discussed with project members in relevant issue. Regarding this feature we have to consider with standardization of metrics and monitoring described in our roadmap.

By the way, you can add your own fork to Awesome Sentinel first :-)

[CarpenterLee]

Thanks for your contribution. It's a good idea to store metrics to some existing monitoring systems, but before that, we should first to standardize our monitoring metrics. Now the metric's name is ambiguous and the monitor indexes are not rich enough, for example the MetricNode.exception is in fact exceptionQps other than total exception count. A standardization of metrics and monitoring will be designed in the latter version, welcome to join us for further discussion.

[sczyh30]

Hi, discussions for metrics and monitoring is welcomed in #211.

[sentinel-bot]

Ping @konglz . Conflict happens after merging a previous commit. Please rebase the branch against master and push it back again. Thanks a lot.

[pouchrobot]

ping @konglz
Conflict happens after merging a previous commit.
Please rebase the branch against master and push it back again. Thanks a lot.