Prometheus之Exporter开发

提示：扫一扫查出行
【扫一扫了解最新限行尾号】
复制提示

Prometheus开发Exporter简介

Exporter 本身是一个http 服务，其指标结果只要符合 Prometheus 规范就可以被 Prometheus 使用。

Prometheus中metric的类型

Prometheus的Client Library提供度量的四种基本类型包括

// Counter 计数器
// Gauge 仪表盘
// Histogram 直方图
// Summary 概要// Prometheus中metric的格式
// 格式：<metric name>{<label name>=<label value>, ...}
// 例如：api_http_requests_total{method="POST", handler="/messages"}// metric name: 唯一标识,命名遵循[a-zA-Z_:][a-zA-Z0-9_:]*.


Counter

Counter类型好比计数器,用于统计类似于: cpu时间,api访问总次数,异常发生次数等等场景,这些指标的特点就是增加不减少.

因此当我们需要统计cpu的使用率时,我们需要使用rate{}函数计算该counter在过去一段时间每个时间序列上的每秒的平均增长率.

一个Counter标识一个累计度量,只增不减,重启后恢复为0,适用于访问次数统计,异常次数统计等场景.

Gauge

Gauge类型,英文直译的话叫"计量器",但是和Counter的翻译太类似了,因此我个人更喜欢使用"仪表盘"这个称呼,仪表盘的特点就是可以增加或者减少的,因此Gauge适合用于如: 当前内存使用率,当前cpu使用率,当前温度,当前速度等等一系列的监控指标

Gauge表示可变化的度量值,适用于CPU,内存使用率等

Histogram

Histogram柱状图这个比较直接,更多的是用于统计一些数据分布的情况,用于计算在一定范围的分布情况,同时还提供了度量指标值的总和.

Histogram对指标的范围性(区间)统计,比如内存在0%-30%, 30%-70%之间的采样次数

Histogram包含三个指标

<basename>: 度量值名称
<basename>_count: 样本反正总次数
<basename>_sum: 样本发生次数中值的总和
<basename>_bucket{le="+Inf"}: 每个区间的样本数


Summary

Summary摘要和Histogram柱状图比较类似,主要用于计算在一定时间窗口范围内度量指标对象的总数以及所有对量指标的总和.

和histogram类似,提供次数和总和,同时提供每个滑动窗口中的分位数.

Histogram和Summary的对比

Prometheus中的Jobs和INSTANCES

Instances

// 仅采集的API endpoint


Jobs

// 相同目的的Instances
// 例如: 四个节点上的api-server
job: api-server
instance 1: 1.2.3.4:5670
instance 2: 1.2.3.4:5671
instance 3: 5.6.7.8:5670
instance 4: 5.6.7.8:5671


Prometheus拉去目标数据时, 会自动给目标的时序上增加一些标签,用于唯一标识,如果时序中本省已经包含,那么取决于honor_labels

// Job: 会增加Job名称
// instance: 增加host: port


开发一个简单Exporter

Prometheus为开发提供了客户端工具,用于为自己的中间件开发Exporter,对接Prometheus, 目前支持go,java,python,ruby

监听HTTP请求返回一行字符串

lexporter_request_count{user="admin"} 1000

package mainimport (
"fmt"
"net/http"
)func HelloHandler(w http.ResponseWriter, r *http.Request) {
fmt.Fprintf(w, "lexporter_request_count{user=\"admin\"} 1000" )
}func main () {
http.HandleFunc("/metrics", HelloHandler)
http.ListenAndServe(":8000", nil)
}


配置Prometheus,将exporter到Prometheus中

计数器

Counter 类型代表一种样本数据单调递增的指标，即只增不减，除非监控系统发生了重置。例如，你可以使用 counter 类型的指标来表示服务的请求数、已完成的任务数、错误发生的次数等。

我们对程序进行改造，累计计算count的数量

example1

package mainimport (
"fmt"
"net/http"
)type Counter struct {
count int64
}func (c *Counter) Add(count int64) int64 {
c.count += count
return c.count
}var counter = new(Counter)func HelloHandler(w http.ResponseWriter, r *http.Request) {
fmt.Fprintf(w, "lexporter_request_count{user=\"admin\"} %d",counter.Add(10) )
}func main () {
http.HandleFunc("/metrics", HelloHandler)
http.ListenAndServe(":8000", nil)
}


example2

package mainimport (
"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
"net/http"
)func main() {// counter
requestTotal := prometheus.NewCounter(prometheus.CounterOpts{
Name: "request_total",
Help: "request total",
})codeStatus := prometheus.NewCounterVec(prometheus.CounterOpts{
Name: "status_code_total",
Help: "status_code total",
},[]string{"code"})requestTotal.Add(10)
codeStatus.WithLabelValues("200").Add(10)
codeStatus.WithLabelValues("500").Add(20)
codeStatus.WithLabelValues("404").Add(30)prometheus.MustRegister(requestTotal)
prometheus.MustRegister(codeStatus)// 暴露
http.Handle("/metrics", promhttp.Handler())
http.ListenAndServe(":8888", nil)
}

Gauge(固定label和非固定label)

package mainimport (
"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
"net/http"
)func main() {
// counter
// guage
// historgram
// summary// metrics name(label=label_value) metrics_valu
// 有lable
// label/label_value 固定// 无label(固定lable)
// lable/label_value 变化的
cpu := prometheus.NewGauge(prometheus.GaugeOpts{
Name: "cpu",
Help: "cpu total",// 没有lable和固定lable一样的
ConstLabels: prometheus.Labels{"a":"xxx"},
})// 非固定label
disk := prometheus.NewGaugeVec(prometheus.GaugeOpts{
Name: "disk",
Help: "disk total",},[]string{"mount"})cpu.Set(2222)
disk.WithLabelValues("/mnt/sda1:").Set(100)
disk.WithLabelValues("/mnt/sda2:").Set(200)
disk.WithLabelValues("/mnt/sda3:").Set(200)
disk.WithLabelValues("/mnt/sda4:").Set(200)// 注册指标信息
prometheus.MustRegister(cpu)
prometheus.MustRegister(disk)// 暴露
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":8888",nil)}


historgram

example1

package mainimport (
"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
"net/http"
)func main() {// historgram
request_time := prometheus.NewHistogramVec(prometheus.HistogramOpts{
Name: "request_time",
Help: "request time",
},[]string{"url"})prometheus.MustRegister(request_time)
request_time.WithLabelValues("/aaa").Observe(6)// 暴露
http.Handle("/metrics", promhttp.Handler())
http.ListenAndServe(":8888", nil)
}

summary

package mainimport (
"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
"net/http"
)func main() {// summary
requestsummary := prometheus.NewSummaryVec(prometheus.SummaryOpts{
Name: "request_time_summary",
Help: "request time summary",
Objectives: map[float64]float64{0.5: 0.05,0.9:0.09},
},[]string{"url"})prometheus.MustRegister(requestsummary)requestsummary.WithLabelValues("/aaa").Observe(6)
requestsummary.WithLabelValues("/aaa").Observe(2)// 暴露
http.Handle("/metrics", promhttp.Handler())
http.ListenAndServe(":8888", nil)
}

值的修改(事件触发或者时间触发)

package mainimport (
"fmt"
"github.com/prometheus/client_golang/prometheus"
"github.com/prometheus/client_golang/prometheus/promhttp"
"math/rand"
"net/http"
"strconv"
"time"
)func main() {
// 无label(固定lable)
// lable/label_value 变化的
cpu := prometheus.NewGauge(prometheus.GaugeOpts{
Name: "cpu",
Help: "cpu total",// 没有lable和固定lable一样的
ConstLabels: prometheus.Labels{"a":"xxx"},
})// 非固定label
disk := prometheus.NewGaugeVec(prometheus.GaugeOpts{
Name: "disk",
Help: "disk total",},[]string{"mount"})cpu.Set(2222)
disk.WithLabelValues("/mnt/sda1:").Set(100)
disk.WithLabelValues("/mnt/sda2:").Set(200)
disk.WithLabelValues("/mnt/sda3:").Set(200)
disk.WithLabelValues("/mnt/sda4:").Set(200)codeStatus := prometheus.NewCounterVec(prometheus.CounterOpts{
Name: "status_code_total",
Help: "status_code total",
},[]string{"code"})codeStatus.WithLabelValues("200").Add(10)
codeStatus.WithLabelValues("500").Add(20)
codeStatus.WithLabelValues("404").Add(30)
// counter
requestTotal := prometheus.NewCounter(prometheus.CounterOpts{
Name: "request_total",
Help: "request total",
})requestTotal.Add(10)// 值的修改
// 修改的时间 => 触发
// 时间触发
// 磁盘使用, cpu使用,内存使用
go func() {
for range time.Tick(time.Second) {
disk.WithLabelValues("/mnt/sda1").Set(float64(rand.Int()))
}
}()// 事件触发,业务请求
http.HandleFunc("/",func(w http.ResponseWriter,r *http.Request){
requestTotal.Inc()
codeStatus.WithLabelValues(strconv.Itoa(rand.Intn(5) * 100)).Add(1)
fmt.Fprintf(w,"hi")
})// 注册指标信息
prometheus.MustRegister(cpu)
prometheus.MustRegister(disk)
prometheus.MustRegister(requestTotal)// 暴露
http.Handle("/metrics",promhttp.Handler())
http.ListenAndServe(":8888",nil)
}