一文带你玩转Golang Prometheus Eexporter开发

目录

• 1. Gauge指标类型
• 1.1 不带label的基本例子
• 1.2 带有固定label指标的例子
• 1.3 带有非固定label指标的例子
• 2. Counter指标类型
• 2.1 不带label的基本例子
• 2.2 带有固定label指标的例子
• 2.3 带有非固定label指标的例子
• 3. Histogram指标类型
• 3.1 不带label的基本例子
• 3.2 带固定label的例子
• 3.3 带有非固定label的例子
• 4. Summary指标类型
• 4.1 不带label的例子
• 4.2 带有固定label的例子
• 4.3 带有非固定label的例子
• 5. 值的修改
• 5.1 基于事件的触发来修改值，比如每访问1次/api就增1
• 5.2 基于时间周期的触发来修改值，比如下面的示例中，是每间隔1秒获取cpu负载指标
• 5.3 基于每访问一次获取指标的URI才修改值，比如每次访问/metrics才去修改某些指标的值

本篇内容有点长，代码有点多。有兴趣的可以坚持看下去，并动手实践，没兴趣的可以划走。本文分两大块，一是搞清楚prometheus四种类型的指标Counter，Gauge，Histogram，Summary用golang语言如何构造这4种类型对应的指标，二是搞清楚修改指标值的场景和方式。

指标类型	类别	描述	可应用场景
Counter	计数类	使用在累计指标单调递增或递减情况下，只能在目标重启后自动归零	服务请求处理数量、已完成任务数量、错误数量
Guage	测量类	使用可增可减的的数据情况下	当前内存/CPU使用情况、并发请求数量
Histogram	直方图类	使用统计指标信息在不同区间内的统计数量	延迟时间、响应大小。例如：0-1秒内的延迟时间、、0-5秒内的延迟时间、例如0-1kb之内的响应大小、0-5kb之内的响应大小
Summary	摘要类	类似于直方图，在客户端对百分位进行统计	延迟时间、响应大小。例如：超过百分之多少的人要满足需求的话，需要多长时间完成。

1. Gauge指标类型

1.1 不带label的基本例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
cpuUsage:=prometheus.NewGauge(prometheus.GaugeOpts{
Name:"cpu_usage",//指标名称
Help:"thisistestmetricscpuusage",//帮助信息
})
//给指标设置值
cpuUsage.Set(89.56)
//注册指标
prometheus.MustRegister(cpuUsage)
//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

1javascript.2 带有固定label指标的例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
cpuUsage:=prometheus.NewGauge(prometheus.GaugeOpts{
Name:"cpu_usage",//指标名称
Help:"thisistestmetricscpuusage",//帮助信息
ConstLabels:prometheus.Labels{"MAChineType":"host"},//label
})
//给指标设置值
cpuUsage.Set(89.56)
//注册指标
prometheus.MustRegister(cpuUsage)
//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

1.3 带有非固定label指标的例子

packagemain

import(
"net/http"

"github.com/prometheu开发者_Go培训s/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义带有不固定label的指标
mtu:=prometheus.NewGaugeVec(prometheus.GaugeOpts{
Name:"interface_mtu",
Help:"网卡接口MTU",
},[]string{"interface","Machinetype"})

//给指标设置值
mtu.WithLabelValues("lo","host").Set(1500)
mtu.WithLabelValues("ens32","host").Set(1500)
mtu.WithLabelValues("eth0","host").Set(1500)

//注册指标
prometheus.MustRegister(mtu)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

2. Counter指标类型

2.1 不带label的基本例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
reqTotal:=prometheus.NewCounter(prometheus.CounterOpts{
Name:"current_request_total",
Help:"当前请求总数",
})
//注册指标
prometheus.MustRegister(reqTotal)

//设置值
reqTotal.Add(10)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

2.2 带有固定label指标的例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/编程客栈prometheus/promhttp"
)

funcmain(){
//定义指标
suceReqTotal:=prometheus.NewCounter(prometheus.CounterOpts{
Name:"sucess_request_total",
Help:"请求成功的总数",
ConstLabels:prometheus.Labels{"StatusCode":"200"},
})
//注册指标
prometheus.MustRegister(suceReqTotal)

//设置值
suceReqTotal.Add(5675)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

2.3 带有非固定label指标的例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
pathReqTotal:=prometheus.NewCounterVec(prometheus.CounterOpts{
Name:"path_request_total",
Help:"path请求总数",
},[]string{"path"})
//注册指标
prometheus.MustRegister(pathReqTotal)

//设置值
pathReqTotal.WithLabelValues("/token").Add(37)
pathReqTotal.WithLabelValues("/auth").Add(23)
pathReqTotal.WithLabelValues("/user").Add(90)
pathReqTotal.WithLabelValues("/api").Add(67)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

3. Histogram指标类型

3.1 不带label的基本例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
reqDelay:=prometheus.NewHistogram(prometheus.HistogramOpts{
Name:"request_delay",
Help:"请求延迟，单位秒",
Buckets:prometheus.LinearBuckets(0,3,6),//调用LinearBuckets生成区间，从0开始，宽度3，共6个Bucket
})

//注册指标
prometheus.MustRegister(reqDelay)

//设置值
reqDelay.Observe(6)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

3.2 带固定label的例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
reqDelay:=prometheus.NewHistogram(prometheus.HistogramOpts{
Name:"request_delay",
Help:"请求延迟，单位秒",
Buckets:prometheus.LinearBuckets(0,3,6),//调用LinearBuckets生成区间，从0开始，宽度3，共6个Bucket
ConstLabels:prometheus.Labels{"path":"/api"},
})

//注册指标
prometheus.MustRegister(reqDelay)

//设置值
reqDelay.Observe(6)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

3.3 带有非固定label的例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
reqDelay:=prometheus.NewHistogramVec(prometheus.HistogramOpts{
Name:"request_delay",
Help:"请求延迟，单位秒",
Buckets:prometheus.LinearBuckets(0,3,6),//调用LinearBuckets生成区间，从0开始，宽度3，共6个Bucket
},[]string{"path"})

//注册指标
prometheus.MustRegister(reqDelay)

//设置值
reqDelay.WithLabelValues("/api").Observe(6)
reqDelay.WithLabelValues("/user").Observe(3)
reqDelay.WithLabelValues("/delete").Observe(2)
reqDelay.WithLabelValues("/get_token").Observe(13)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

4. Summary指标类型

4.1 不带label的例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
reqDelay:=prometheus.NewSummary(prometheus.SummaryOpts{
Name:"request_delay",
Help:"请求延迟",
Objectives:map[float64]float64{0.5:0.05,0.90:0.01,0.99:0.001},//百分比:精度
})

//注册指标
prometheus.MustRegister(reqDelay)

//设置值
reqDelay.Observe(4)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

4.2 带有固定label的例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometandroidheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
reqDelay:=prometheus.NewSummary(prometheus.SummaryOpts{
Name:"request_delay",
Help:"请求延迟编程",
Objectives:map[float64]float64{0.5:0.05,0.90:0.01,0.99:0.001},//百分比:精度
ConstLabels:prometheus.Labels{"path":"/api"},
})

//注册指标
prometheus.MustRegister(reqDelay)

//设置值
reqDelay.Observe(4)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

4.3 带有非固定label的例子

packagemain

import(
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
//定义指标
reqDelay:=prometheus.NewSummaryVec(prometheus.SummaryOpts{
Name:"request_delay",
Help:"请求延迟",
Objectives:map[float64]float64{0.5:0.05,0.90:0.01,0.99:0.001},//百分比:精度
},[]string{"path"})

//注册指标
prometheus.MustRegister(reqDelay)

//设置值
reqDelay.WithLabelValues("/api").Observe(4)
reqDelay.WithLabelValues("/token").Observe(2)
reqDelay.WithLabelValues("/user编程客栈").Observe(3)

//暴露指标
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

5. 值的修改

5.1 基于事件的触发来修改值，比如每访问1次/api就增1

基于事件的触发对指标的值进行修改，通常大多数是来自业务方面的指标需求，如自研的应用需要暴露相关指标给promethus进行监控、展示，那么指标采集的代码（指标定义、设置值）就要嵌入到自研应用代码里了。

packagemain

import(
"fmt"
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
)

funcmain(){
urlRequestTotal:=prometheus.NewCounterVec(prometheus.CounterOpts{
Name:"urlRequestTotal",
Help:"PATH请求累计单位次",
},[]string{"path"})

http.HandleFunc("/api",func(whttp.ResponseWriter,r*http.Request){
urlRequestTotal.WithLabelValues(r.URL.Path).Inc()//使用Inc函数进行增1
fmt.Fprintf(w,"Welcometotheapi")
})

prometheus.MustRegister(urlRequestTotal)
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

5.2 基于时间周期的触发来修改值，比如下面的示例中，是每间隔1秒获取cpu负载指标

基于时间周期的触发，可以是多少秒、分、时、日、月、周。

packagemain

import(
"net/http"
"time"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
"github.com/shirou/gopsutil/load"
)

funcmain(){

cpuUsage:=prometheus.NewGaugeVec(prometheus.GaugeOpts{
Name:"CpuLoad",
Help:"CPU负载",
},[]string{"time"})

//开启一个子协程执行该匿名函数内的逻辑来给指标设置值，且每秒获取一次
gofunc(){
forrangetime.Tick(time.Second){
info,_:=load.Avg()
cpuUsage.WithLabelValues("Load1").Set(float64(info.Load1))
cpuUsage.WithLabelValues("Load5").Set(float64(info.Load5))
cpuUsage.WithLabelValues("Load15").Set(float64(info.Load15))
}
}()

prometheus.MustRegister(cpuUsage)

http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

5.3 基于每访问一次获取指标的URI才修改值，比如每次访问/metrics才去修改某些指标的值

下面的这个示例是，每访问一次/metrics就获取一次内存总容量

packagemain

import(
"fmt"
"net/http"

"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
"github.com/shirou/gopsutil/mem"
)

funcmain(){
MemTotal:=prometheus.NewGaugeFunc(prometheus.GaugeOpts{
Name:"MemTotal",
Help:"内存总容量单位GB",
},func()float64{
fmt.Println("callMemTotal...")
info,_:=mem.VirtualMemory()
returnfloat64(info.Total/1024/1024/1024)
})
prometheus.MustRegister(MemTotal)
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":9900",nil)
}

目前只有Gauge和Counter的指标类型有对应的函数，分别是NewGaugeFunc和NewCounterFunc，而且是固定的label。

到此这篇关于一文带你玩转Golang Prometheus Eexporter开发的文章就介绍到这了,更多相关Golang Prometheus Eexporter开发内容请搜索我们以前的文章或继续浏览下面的相关文章希望大家以后多多支持我们！