monitoring

Prometheus monitoring for Hertz

Usage Example

Server

import (
   "context"
	"time"

	"github.com/cloudwego/monitor-prometheus"
	"github.com/cloudwego/hertz/pkg/app"
	"github.com/cloudwego/hertz/pkg/app/server"
)

func main() {
...
	h := server.Default(server.WithHostPorts("127.0.0.1:8080"), server.WithTracer(prometheus.NewServerTracer(":9091", "/hertz")))

	h.GET("/metricGet", func(c context.Context, ctx *app.RequestContext) {
		ctx.String(200, "hello get")
	})

	h.POST("/metricPost", func(c context.Context, ctx *app.RequestContext) {
		time.Sleep(100 * time.Millisecond)
		ctx.String(200, "hello post")
	})

	h.Spin()

HOW-TO-RUN

1. install docker and start docker

2. change $inetIP to local ip in line 30 of prometheus.yml

3. run Prometheus and Grafana
   docker-compose up

4. run Hertz server
   go run main.go

5. run Hertz client
   go run client/main.go

6. visit http://localhost:3000, the account password is admin by default

7. configure Prometheus data sources

   1. Configuration
   2. Data Source
   3. Add data source
   4. Select Prometheus and fill the URL with http://prometheus:9090
   5. click Save & Test after configuration to test if it works

8. add dashboard Create -> dashboard, add monitoring metrics such as throughput and pct99 according to your needs, for example:

   • server throughput of succeed requests

   sum(rate(hertz_server_throughput{statusCode="200"}[1m])) by (method)

   • server latency pct99 of succeed requests

   histogram_quantile(0.9,sum(rate(hertz_server_latency_us_bucket{statusCode="200"}[1m]))by(le))

For more information about hertz monitoring, please click monitoring