动手写prometheus的exporter-04-Summary（摘要）

1. 不带标签的Summary

语法

• 实例化

func NewSummary(opts SummaryOpts) Summary
1

• SummaryOpts 结构体

type SummaryOpts struct {
    Namespace   string
    Subsystem   string
    Name        string
    Help        string
    ConstLabels Labels
    Objectives  map[float64]float64
    MaxAge      time.Duration
    AgeBuckets  uint32
    BufCap      uint32
}
1
2
3
4
5
6
7
8
9
10
11

• Observe

func (Histogram) Observe(float64)
1

完整示例

一组学生的成绩，分别为分到59分（含）以下、69分（含）以下……等各bucket中

• 代码

package main

import (
	"flag"
	"fmt"
	"github.com/prometheus/client_golang/prometheus"
	"github.com/prometheus/client_golang/prometheus/promhttp"
	"log"
	"net/http"
)
var addr = flag.String("listen-address", ":1840", "The address to listen on for HTTP requests")

var (
	numSummary = prometheus.NewSummary(prometheus.SummaryOpts{
		Name: "Summary_num",
		Help: "Calculate quantile",
		Objectives: map[float64]float64{0.4:0,0.5:0,0.6:0,0.7:0,0.8:0,0.88:0,0.9:0,0.95:0},
	})

)


func init() {
	prometheus.MustRegister(numSummary)
}

func main() {
	flag.Parse()

	var numbers = [10]float64{100,200,300,400,500,600,700,800,900,1000}
	for i:=0;i<len(numbers);i++{
		numSummary.Observe(numbers[i])
		fmt.Printf("Insert number: %f \n", numbers[i])
	}

	http.Handle("/metrics", promhttp.Handler())
	log.Fatal(http.ListenAndServe(*addr, nil))
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38

• 执行结果

# HELP Summary_num Calculate quantile
# TYPE Summary_num summary
Summary_num{quantile="0.4"} 400
Summary_num{quantile="0.5"} 500
Summary_num{quantile="0.6"} 600
Summary_num{quantile="0.7"} 700
Summary_num{quantile="0.8"} 800
Summary_num{quantile="0.88"} 900
Summary_num{quantile="0.9"} 900
Summary_num{quantile="0.95"} 1000
Summary_num_sum 5500
Summary_num_count 10
1
2
3
4
5
6
7
8
9
10
11
12

结果说明：

• Summary_num{quantile="0.4"}：该数组从小到大排序，前40%（含）的成员最后一个是400

如：本例从小到大排序，前40%是：100,200，300，400。最后一个是400。

• Summary_num{quantile="0.8"}：该数组从小到大排序，前80%（含）的成员最后一个是800
• Summary_num{quantile="0.88"}：该数组从小到大排序，前88%（含）的成员最后一个是900

从小到大排序后，前80%（含）的最后一位是800，但前88%（含）不包含800了，而最后一个取到了900

• summary_count_sum：所用统计的数字（10个）和为5500
• scores_count_count：当前一共统计了10个数的结果

2. 带标签的Summary

语法

• 实例化

func NewSummaryVec(opts SummaryOpts, labelNames []string) *SummaryVec
1

• HistogramOpts结构体

type SummaryOpts struct {
    Namespace   string
    Subsystem   string
    Name        string
    Help        string
    ConstLabels Labels
    Objectives  map[float64]float64
    MaxAge      time.Duration
    AgeBuckets  uint32
    BufCap      uint32
}

1
2
3
4
5
6
7
8
9
10
11
12

• 打标签

func (v *HistogramVec) WithLabelValues(lvs ...string) Observer
1

• Observe

func (Histogram) Observe(float64)
1

传入Observe(float64)的数据将统计到Bucket

完整示例

package main

import (
	"flag"
	"github.com/prometheus/client_golang/prometheus"
	"github.com/prometheus/client_golang/prometheus/promhttp"
	"log"
	"net/http"
)
var addr = flag.String("listen-address", ":1840", "The address to listen on for HTTP requests")

var (
	numSummary = prometheus.NewSummaryVec(prometheus.SummaryOpts{
		Name: "Summary_num",
		Help: "Calculate quantile",
		Objectives: map[float64]float64{0.4:0,0.5:0,0.6:0,0.7:0,0.8:0,0.88:0,0.9:0,0.95:0},
	},[]string{"group"})

)


func init() {
	prometheus.MustRegister(numSummary)
}

func main() {
	flag.Parse()

	var numbers01 = [10]float64{100,200,300,400,500,600,700,800,900,1000}
	var numbers02 = [10]float64{1100,1200,1300,1400,1500,1600,1700,1800,1900,2000}
	for i:=0;i<10;i++{
		numSummary.WithLabelValues("group-01").Observe(numbers01[i])
		numSummary.WithLabelValues("group-02").Observe(numbers02[i])
	}

	http.Handle("/metrics", promhttp.Handler())
	log.Fatal(http.ListenAndServe(*addr, nil))
}
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38

• 输出

# HELP Summary_num Calculate quantile
# TYPE Summary_num summary
Summary_num{group="group-01",quantile="0.4"} 400
Summary_num{group="group-01",quantile="0.5"} 500
Summary_num{group="group-01",quantile="0.6"} 600
Summary_num{group="group-01",quantile="0.7"} 700
Summary_num{group="group-01",quantile="0.8"} 800
Summary_num{group="group-01",quantile="0.88"} 900
Summary_num{group="group-01",quantile="0.9"} 900
Summary_num{group="group-01",quantile="0.95"} 1000
Summary_num_sum{group="group-01"} 5500
Summary_num_count{group="group-01"} 10
Summary_num{group="group-02",quantile="0.4"} 1400
Summary_num{group="group-02",quantile="0.5"} 1500
Summary_num{group="group-02",quantile="0.6"} 1600
Summary_num{group="group-02",quantile="0.7"} 1700
Summary_num{group="group-02",quantile="0.8"} 1800
Summary_num{group="group-02",quantile="0.88"} 1900
Summary_num{group="group-02",quantile="0.9"} 1900
Summary_num{group="group-02",quantile="0.95"} 2000
Summary_num_sum{group="group-02"} 15500
Summary_num_count{group="group-02"} 10
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

说明：见上例

---