# 性能分析

当我们把代码写完，单元测试，性能测试都做了，程序也能够运行起来，一切看起来都很美好。然而，光有这些是不足以保证服务能够长久稳定运行的，一些问题只有在长时间运行之后才会发现，比方说一些网络连接没有释放或者是协程陷入了阻塞或死循环，这些问题会在运行的过程中慢慢累积，到达某一瓶颈时爆发，影响到服务器系统的稳定性。

在golang中提供了对于应用程序运行阶段进行性能分析的工具pprof，根据场景的不同分为了两个库：`runtime/pprof`和`net/http/pprof`。

`runtime/pprof`用于为非守护形式运行的程序做性能分析

`net/http/pprof`用于为守护态的http服务程序做性能分析

接下来我们以一个http服务的例子来看如何做性能分析

例子：http服务性能分析

```go
package main

import (
	"fmt"
	"net/http"
	_ "net/http/pprof"

	"time"
)

func slow(w http.ResponseWriter, req *http.Request) {
	time.Sleep(time.Second * 5)
	w.Write([]byte("a slow return"))
}

func fast(w http.ResponseWriter, req *http.Request) {
	w.Write([]byte("a fast return"))
}

func endless(w http.ResponseWriter, req *http.Request) {
	for {
		time.Sleep(time.Second)
		if false {
			break
		}
	}
	w.Write([]byte("you will never see this message"))
}

func main() {
	http.HandleFunc("/slow", slow)

	http.HandleFunc("/fast", fast)

	http.HandleFunc("/endless", endless)

	if err := http.ListenAndServe("localhost:8080", nil); err != nil {
		fmt.Println(err)
	}
}

```

执行go run main.go启动服务

开启浏览器访问<http://localhost:8080/debug/pprof/可以看到如下内容>

![](/files/-LiGeN491TF0lkJYzOyP)

可以看到各项指标的当前运行统计数据，点击各项指标的链接可以查看到指标的当前详情

接下来使用go tool pprof交互式工具统计服务运行60秒间的性能情况

再开启一个新的控制台，执行命令

```
go tool pprof http://localhost:8080/debug/pprof/profile?seconds=60
```

输出如下

```
Fetching profile over HTTP from http://localhost:8080/debug/pprof/profile?seconds=60
```

这时候可以分别多次访问以下测试链接

```
http://localhost:8080/fast
http://localhost:8080/slow
http://localhost:8080/endless
```

60秒后交互式工具控制台会输出如下内容

```
Saved profile in C:\Users\wj361\pprof\pprof.samples.cpu.001.pb.gz
Type: cpu
Time: Jun 26, 2019 at 10:46am (CST)
Duration: 60s, Total samples = 30ms ( 0.05%)
Entering interactive mode (type "help" for commands, "o" for options)
(pprof)
```

说明分析报告已经生成完毕，可以输入help+Enter键查看支持的命令

比分说查看cpu开销前10的方法

```
(pprof) top10
Showing nodes accounting for 30ms, 100% of 30ms total
Showing top 10 nodes out of 15
      flat  flat%   sum%        cum   cum%
      10ms 33.33% 33.33%       10ms 33.33%  runtime.memmove
      10ms 33.33% 66.67%       10ms 33.33%  runtime.stdcall1
      10ms 33.33%   100%       10ms 33.33%  runtime.timerproc
         0     0%   100%       10ms 33.33%  net/http.(*conn).serve
         0     0%   100%       10ms 33.33%  net/http.(*connReader).startBackgroundRead
         0     0%   100%       10ms 33.33%  runtime.entersyscallblock_handoff
         0     0%   100%       10ms 33.33%  runtime.handoffp
         0     0%   100%       10ms 33.33%  runtime.mstart
         0     0%   100%       10ms 33.33%  runtime.newproc
         0     0%   100%       10ms 33.33%  runtime.newproc.func1
```

如果要查看其他的指标，也是以同样的方式运行交互式工具

```
go tool pprof http://localhost:8080/debug/pprof/{指标名称}?seconds=60
```

除了通过命令行交换工具查看分析报告外，还可以通过图形界面查看，这需要使用到graphviz工具

{% embed url="<https://www.graphviz.org/download/>" %}
graphviz下载地址
{% endembed %}

{% hint style="info" %}
需要将graphviz的二进制目录添加到环境变量$PATH中
{% endhint %}

开启cpu性能分析并启动web视图报告

```
go tool pprof -http=:8081 http://localhost:8080/debug/pprof/profile?seconds=60
```

分别多次访问以下测试链接

```
http://localhost:8080/fast
http://localhost:8080/slow
http://localhost:8080/endless
```

60秒后打开浏览器访问<http://localhost:8081/ui/，可以看到方法的调用链，其中框越大线越粗表示耗得时间更长>

![](/files/-LiGsB_QFEEJ_PonS9VY)

通过下拉菜单VIEW可以切换各种视图


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://golang-2.gitbook.io/handbook/ji-chu/ce-shi/xing-neng-fen-xi.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.