prometheus自定義metrics

項目地址 我的博客

平常項目開發過程當中爲了觀察項目的線上運行指標一般須要項目提供一系列指標信息，咱們目前用的prometheus，而項目固然要提供一系列prometheus metrics指標信息，可是默認官方golang的倉庫不是很好用。

• 全部指標所有保存在內存重啓後會所有清零
• 若是指標labels數量變動，沒法從新註冊，unregister貌似無效
• 沒法定時清除數據，好比我想觀察某些指標每一天的數據
• web服務不少通用的http接口指標須要重複定義並本身定義中間件收集

因而乎找到github上一個很好的項目，能夠知足上面全部需求能夠大多數通常項目開發過程當中的metrics需求，使用起來也比較簡單。下面來自項目READEME：

lib-metrics

GoDoc

Dependency

github.com/AlecAivazis/survey/v2 v2.0.1 // indirect
github.com/atotto/clipboard v0.1.2 // indirect
github.com/go-redis/redis v6.15.2+incompatible
github.com/golang/protobuf v1.3.1
github.com/onsi/ginkgo v1.8.0 // indirect
github.com/onsi/gomega v1.5.0 // indirect
github.com/prometheus/client_golang v1.0.0
github.com/prometheus/client_model v0.0.0-20190129233127-fd36f4220a90
github.com/prometheus/common v0.6.0
gopkg.in/robfig/cron.v2 v2.0.0-20150107220207-be2e0b0deed5
gopkg.in/yaml.v2 v2.2.2 // indirect
複製代碼

Initialization

init global collector gather

subsystem := "test"
metricsCollector = NewAutumnCollector(subsystem)
複製代碼

create local metric and registry

localMetrics = metricsCollector.NewAutumnLocalMetric("local_test", "local_type")
// Reset data at zero every day
localMetrics.ResetByCron("0 0 * * *")
metricsCollector.Register(localMetrics)
複製代碼

create metric base on redis and registry

redisMetric = metricsCollector.NewAutumnRedisMetric(client, "redis_test", "type")
// reset redis key
redisMetric.ResetMetricRedisKey("reset_redis_key_test")
// get now redis key
redisMetric.GetRedisKey()
// Reset data at zero every day
redisMetric.ResetByCron("0 0 * * *")
// registry metric
metricsCollector.Register(redisMetric)
複製代碼

Register HTTP handler middleware to customize metrics for collecting HTTP requests

// Collect interface responses
var standardRequestDuration = prometheus.NewHistogramVec(
	prometheus.HistogramOpts{
		Name:    "server_request_duration_seconds",
		Help:    "How much time costs(s)",
		Buckets: []float64{0.0005, 0.001, 0.002, 0.005, 0.010, 0.020, 0.050, 0.1, 0.5, 1, 5},
	}, []string{"code", "router"})

metricsCollector.RegisterPrometheus(standardRequestDuration)

// Pre-interceptor record start time
metricsCollector.UseBeforeHandlerHook(func(ctx *Context) {
	fmt.Println("this is before 1 start")
	start := time.Now()
	ctx.Set("start", start)
	fmt.Println("this is before 1 end")
})

// the post-interceptor calculates the response time and obtains information such as code, which is set by Observer.
metricsCollector.UseAfterHandlerHook(func(ctx *Context) {
	fmt.Println("this is after 1 start")
	start := ctx.GetTime("start")
	elapsed := time.Since(start).Seconds()
	standardRequestDuration.WithLabelValues(fmt.Sprint(ctx.ResponseWriter.Status()), ctx.Request.URL.Path).Observe(elapsed)
	fmt.Println("this is after 1 end")
})
複製代碼

Usage

// local metrics
localMetrics.Set(1,"1")
localMetrics.Incr("2")
localMetrics.Add(1,"3")

// redis metrics
redisMetric.Set(1,"1")
redisMetric.Incr("2")
redisMetric.Add(1,"3")

// Using native metric
standardRequestDuration.WithLabelValues(200, "/test").Observe(10)
複製代碼

Registry Handler

Register gin middleware and add metrics interface to collect HTTP metrics information

engine = gin.New()
// add gin middleware
engine.Use(func(ctx *gin.Context) {
	metricsCollector.Use(http.HandlerFunc(func(w http.ResponseWriter,r *http.Request) {
		ctx.Next()
	})).ServeHTTP(ctx.Writer,ctx.Request)
})
// Register gin interface to provide metrics interface
engine.Get(DefaultMetricPath, gin.WrapF(metricsCollector.Handler()))
複製代碼

Register native HTTP Middleware

engine = new(http.ServeMux)
h:=metricsCollector.Use(engine)
engine.Handle(DefaultMetricPath, metricsCollector.Handler())
http.ListenAndServe(":8080", h)
複製代碼

Built-in default HTTP service startup metrics interface

if err := metricsCollector.ListenAndServe(":8080",metricsCollector.Handler); err != nil && err != http.ErrServerClosed {
     panic(err)
}
複製代碼