十一，微服務監控

監控 PROMETHEUS

Prometheus是由SoundCloud開發的開源監控報警系統和時序列資料庫(TSDB)，使用Go語言開發，是Google BorgMon監控系統的開源版本，prometheus存盤的是時序資料，即按相同時序(相同名稱和標簽)，以時間維度存盤連續的資料的集合，

專案中應用

完整代碼：

https://github.com/Justin02180218/micro-kit

組態檔

prometheus:
  namespace: "library"
  subsystem: "user_service"

config.go

type PrometheusConfig struct {
    Namespace string `json:"namespace" yaml:"namespace"`
    Subsystem string `json:"subsystem" yaml:"subsystem"`
}

pkg/monitors

在 pkg 下新建目錄 monitors，創建 prometheus.go 檔案：

代碼如下：

type PrometheusParams struct {
    Counter   *kitprometheus.Counter
    Summary   *kitprometheus.Summary
    Gauge     *kitprometheus.Gauge
    Histogram *kitprometheus.Histogram
}
func MakePrometheusParams(conf *configs.PrometheusConfig) *PrometheusParams {
    fieldKeys := []string{"method"}
    requestCounter := kitprometheus.NewCounterFrom(stdprometheus.CounterOpts{
        Namespace: conf.Namespace,
        Subsystem: conf.Subsystem,
        Name:      "request_count",
        Help:      "Number of requests received.",
    }, fieldKeys)
    requestLatency := kitprometheus.NewSummaryFrom(stdprometheus.SummaryOpts{
        Namespace: conf.Namespace,
        Subsystem: conf.Subsystem,
        Name:      "request_latency_microseconds",
        Help:      "Total duration of requests in microseconds.",
    }, fieldKeys)
    return &PrometheusParams{requestCounter, requestLatency, nil, nil}
}

service層

在 service 層創建 user_service_metrics.go，以 library-user-service 為例，其他微服務相同，

func MetricsMiddleware(prometheusParams *monitors.PrometheusParams) ServiceMiddleware {
    return func(next UserService) UserService {
        return metricsMiddleware{next, prometheusParams.Counter, prometheusParams.Summary}
    }
}

type metricsMiddleware struct {
    UserService
    requestCount   metrics.Counter
    requestLatency metrics.Histogram
}

func (mw metricsMiddleware) Register(ctx context.Context, vo *dto.RegisterUser) (*dto.UserInfo, error) {
    defer func(begin time.Time) {
        lvs := []string{"method", "Register"}
        mw.requestCount.With(lvs...).Add(1)
        mw.requestLatency.With(lvs...).Observe(time.Since(begin).Seconds())
    }(time.Now())
    return mw.UserService.Register(ctx, vo)
}

func (mw metricsMiddleware) FindByID(ctx context.Context, id uint64) (*dto.UserInfo, error) {
    defer func(begin time.Time) {
        lvs := []string{"method", "FindByID"}
        mw.requestCount.With(lvs...).Add(1)
        mw.requestLatency.With(lvs...).Observe(time.Since(begin).Seconds())
    }(time.Now())
    return mw.UserService.FindByID(ctx, id)
}

func (mw metricsMiddleware) FindByEmail(ctx context.Context, email string) (*dto.UserInfo, error) {
    defer func(begin time.Time) {
        lvs := []string{"method", "FindByEmail"}
        mw.requestCount.With(lvs...).Add(1)
        mw.requestLatency.With(lvs...).Observe(time.Since(begin).Seconds())
    }(time.Now())
    return mw.UserService.FindByEmail(ctx, email)
}

func (mw metricsMiddleware) FindBooksByUserID(ctx context.Context, id uint64) (interface{}, error) {
    defer func(begin time.Time) {
        lvs := []string{"method", "FindBooksByUserID"}
        mw.requestCount.With(lvs...).Add(1)
        mw.requestLatency.With(lvs...).Observe(time.Since(begin).Seconds())
    }(time.Now())
    return mw.UserService.FindBooksByUserID(ctx, id)
}

func (mw metricsMiddleware) HealthCheck() bool {
    defer func(begin time.Time) {
        lvs := []string{"method", "HealthCheck"}
        mw.requestCount.With(lvs...).Add(1)
        mw.requestLatency.With(lvs...).Observe(time.Since(begin).Seconds())
    }(time.Now())
    return mw.UserService.HealthCheck()
}

transport層

修改 user_transport.go，以 library-user-service 為例，其他微服務相同，

r.GET("/metrics", func(c *gin.Context) {
    promhttp.Handler().ServeHTTP(c.Writer, c.Request)
})

修改 main.go

以 library-user-service 為例，其他微服務相同，

prometheusParams := monitors.MakePrometheusParams(configs.Conf.PrometheusConfig)

userService = service.MetricsMiddleware(prometheusParams)(userService)

啟動 Prometheus

在工程目錄下建立 deploy/prometheus 目錄，創建 prometheus.yaml 組態檔：

global:
  scrape_interval: 15s
  external_labels:
    monitor: 'library-monitor'
scrape_configs:
  - job_name: 'prometheus'
    scrape_interval: 5s
    static_configs:
     - targets: ['IP:9090']
       labels:
        group: 'local'
  
  - job_name: 'library'
    scrape_interval: 5s
    static_configs:
      # 配置微服務的IP和Port  
      - targets: ['IP:10086', 'IP:10087', 'IP:10089']
        labels:
          group: 'library'

創建 docker-compose 檔案 prometheus-grafana.yaml：

version: '3'
services:    
 prometheus:
  image: prom/prometheus
  ports:
    - 9090:9090
  volumes:
    - /micro-kit/deploy/prometheus/prometheus.yaml:/etc/prometheus/prometheus.yml
 grafana:
  image: grafana/grafana
  ports: 
    - 3000:3000
  environment:
    - GF_SECURITY_ADMIN_PASSWORD=123456
  volumes: 
    - /grafana_db:/var/lib/grafana grafana/grafana

執行命令：

docker-compose -f deploy/prometheus/prometheus-grafana.yaml up -d

依次啟動 consul、zipkin、三個微服務和網關，然后在地址欄輸入 http://localhost:9090/targets，可以看到 prometheus 成功收集到服務資訊：

下一篇文章，我們將三個微服務和一個網關構建成Docker鏡像，在容器中啟動整個應用，

完整代碼：

https://github.com/Justin02180218/micro-kit

更多【分布式專輯】【架構實戰專輯】系列文章，請關注公眾號