gin-metrics

gin-gonic/gin metrics exporter for Prometheus.

中文

• Introduction
• Grafana
• Installation
• Usage
• Custom Metric
• Metric with separate port
• Contributing

Introduction

gin-metrics defines some metrics for gin http-server. There have easy way to use it.

Below is the detailed description for every metric.

Metric	Type	Description
gin_request_total	Counter	all the server received request num.
gin_request_uv	Counter	all the server received ip num.
gin_uri_request_total	Counter	all the server received request num with every uri.
gin_request_body_total	Counter	the server received request body size, unit byte.
gin_response_body_total	Counter	the server send response body size, unit byte.
gin_request_duration	Histogram	the time server took to handle the request.
gin_slow_request_total	Counter	the server handled slow requests counter, t=%d.

Grafana

Set the grafana directory for details.

Installation

$ go get github.com/penglongli/gin-metrics

Usage

Your can see some metrics across http://localhost:8080/metrics

package main

import (
	"github.com/gin-gonic/gin"

	"github.com/penglongli/gin-metrics/ginmetrics"
)

func main() {
	r := gin.Default()

	// get global Monitor object
	m := ginmetrics.GetMonitor()

	// +optional set metric path, default /debug/metrics
	m.SetMetricPath("/metrics")
	// +optional set slow time, default 5s
	m.SetSlowTime(10)
	// +optional set request duration, default {0.1, 0.3, 1.2, 5, 10}
	// used to p95, p99
	m.SetDuration([]float64{0.1, 0.3, 1.2, 5, 10})

	// set middleware for gin
	m.Use(r)

	r.GET("/product/:id", func(ctx *gin.Context) {
		ctx.JSON(200, map[string]string{
			"productId": ctx.Param("id"),
		})
	})

	_ = r.Run()
}

Custom Metric

gin-metric provides ways to custom your own metric.

Gauge

With Gauge type metric, you can use three functions to change it's value.

And you should define a Gauge Metric first,

gaugeMetric := &ginmetrics.Metric{
    Type:        ginmetrics.Gauge,
    Name:        "example_gauge_metric",
    Description: "an example of gauge type metric",
    Labels:      []string{"label1"},
}

// Add metric to global monitor object
_ = ginmetrics.GetMonitor().AddMetric(gaugeMetric)

SetGaugeValue

SetGaugeValue will setting metric value directly。

_ = ginmetrics.GetMonitor().GetMetric("example_gauge_metric").SetGaugeValue([]string{"label_value1"}, 0.1)

Inc

Inc will increase 1 to metric value

_ = ginmetrics.GetMonitor().GetMetric("example_gauge_metric").Inc([]string{"label_value1"})

Add

Add will add float64 num to metric value

_ = ginmetrics.GetMonitor().GetMetric("example_gauge_metric").Add([]string{"label_value1"}, 0.2)

Counter

With Counter type metric, you can use Inc and Add function, don't use SetGaugeValue.

Histogram and Summary

For Histogram and Summary type metric, should use Observe function.

Metric with separate port

For some users, they don't want to merge the port of the metric with the port of the application.

So we provide a way to separate the metric port. Here is the example.

func main() {
	appRouter := gin.Default()
	metricRouter := gin.Default()

	m := ginmetrics.GetMonitor()
	// use metric middleware without expose metric path
	m.UseWithoutExposingEndpoint(appRouter)
	// set metric path expose to metric router
	m.Expose(metricRouter)

	appRouter.GET("/product/:id", func(ctx *gin.Context) {
		ctx.JSON(200, map[string]string{
			"productId": ctx.Param("id"),
		})
	})
	go func() {
		_ = metricRouter.Run(":8081")
	}()
	_ = appRouter.Run(":8080")
}

Contributing

If someone has a problem or suggestions, you can submit new issues or new pull requests.