轻量级压测工具hey介绍与实践-轻识

走过路过不要错过

点击蓝字关注我们

二进制安装

Linux 64-bit: https://storage.googleapis.com/hey-release/hey_linux_amd64
Mac 64-bit: https://storage.googleapis.com/hey-release/hey_darwin_amd64
Windows 64-bit: https://storage.googleapis.com/hey-release/hey_windows_amd64

源码安装

需要提交安装GO语言环境https://www.runoob.com/go/go-environment.html

git clone https://github.com/rakyll/hey.git
go get -u github.com/rakyll/hey
go install github.com/rakyll/hey
go build

hey压测简明示例

指定时长的get请求：客户端(-c)并发为2，持续发送请求2s (-c)
hey -z 5s -c 2 https://www.baidu.com/
指定请求总数的get请求：运行2000次(-n)，客户端并发为50(-c)
hey -n 2000 -c 50 https://www.baidu.com/
指定host的get请求：使用的cpu核数为2 (-cpus), 压测时长为5s(-z), 并发数为2
hey -z 5s -c 2 -cpus 2 -host "baidu.com" https://220.181.38.148
请求带header的get接口：压测时长为5s (-z), 客户端发送请求的速度为128QPS, 请求头用-H添加
hey -z 5s -q 128 -H "client-ip:0.0.0.0" -H "X-Up-Calling-Line-Id:X.L.Xia" https://www.baidu.com/
请求post请求
hey -z 5s -c 50 -m POST -H "info:firstname=xiuli; familyname=xia" -d "year=2020&month=1&day=21" https://www.baidu.com/
代理模式，需额外配置proxy：因部分ip频繁发请求有风险，故可用-x设置白名单代理向服务器发请求
hey -z 5s -c 10 -x "http://127.0.0.1:8001" http://baidu.com/
shell for循环实现压测
for i in `seq 10`; do curl -v http://baidu.com; done

使用方法

Usage: hey [options...]
Options:
// 指定运行的总请求数。默认值为200。
-n Number of requests to run. Default is 200.
// 客户端并发执行的请求数，默认为50。总请求数不能小于并发数。
-c Number of workers to run concurrently. Total number of requests cannot
      be smaller than the concurrency level. Default is 50.
// 客户端发送请求的速度限制，以每秒响应数QPS为单位，默认没有限制。
-q Rate limit, in queries per second (QPS) per worker. Default is no rate limit.
// 发送请求的持续时长，超时后程序停止并退出。若指定了持续时间，则忽略总请求数(-n)，例如-z 10s，-z 3m
-z Duration of application to send requests. When duration is reached,
      application stops and exits. If duration is specified, n is ignored.
      Examples: -z 10s -z 3m.
// 输出类型。若没有提供，则打印摘要。CSV是唯一支持的格式，结果以逗号分隔各个指标项。
-o Output type. If none provided, a summary is printed.
      "csv" is the only supported alternative. Dumps the response
      metrics in comma-separated values format.
// HTTP方法，例如GET，POST，PUT，DELETE，HEAD，OPTIONS方法
-m HTTP method, one of GET, POST, PUT, DELETE, HEAD, OPTIONS.
// HTTP请求头，可以用-H连续添加多个请求头。
-H Custom HTTP header. You can specify as many as needed by repeating the flag.
      For example, -H "Accept: text/html" -H "Content-Type: application/xml" .
// 每个请求的超时时间（以秒为单位）。默认值为20s，数值0代表永不超时。
-t Timeout for each request in seconds. Default is 20, use 0 for infinite.
// HTTP响应头
-A HTTP Accept header.
// HTTP请求正文
-d HTTP request body.
// HTTP请求正文文件
-D HTTP request body from file. For example, /home/user/file.txt or ./file.txt.
// 内容类型，默认为“ text / html”。
-T Content-type, defaults to "text/html".
// 基本身份验证，用户名：密码。
-a Basic authentication, username:password.
// HTTP代理地址作, 使用host:port格式。
-x HTTP Proxy address as host:port.
// 启用HTTP / 2
-h2 Enable HTTP/2.
// HTTP主机头
-host HTTP Host header.
// 禁用压缩。
-disable-compression Disable compression.
// 禁用保持活动状态，防止重新使用不同的HTTP请求之间的TCP连接。
-disable-keepalive    Disable keep-alive, prevents re-use of TCP
                        connections between different HTTP requests.
// 禁用HTTP重定向
-disable-redirects    Disable following of HTTP redirects
// 使用的cpu内核数。当前计算机的默认值为8核。
-cpus                 Number of used cpu cores.
                        (default for current machine is 8 cores)

Windows下命令行

hey_windows_amd64.exe -n 9000 -c 100 -m "POST" -T "application/json" -H "Authorization: Bearer eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJodHRwOi8vc2NoZW1hcy54bWxzb2FwLm9yZy93cy8yMDA1LzA1L2lkZW50aXR5L2NsYWltcy9uYW1laWRlbnRpZmllciI6IjEzNzI1MTUyNTI1Iiwic3ViIjoiMTM3MjUxNTI1MjUiLCJqdGkiOiIxZTU2MzlhNS02YjQzLTRmNTctYWFkYi0yN2UyNzcyOTliNTgiLCJpYXQiOjE1ODM3OTg0NzAsIm5iZiI6MTU4Mzc5ODQ3MCwiZXhwIjoxNTgzODg0ODcwLCJpc3MiOiJHREg1IiwiYXVkIjoiR0RINSJ9.n5sUWANx28LXmvipMMl0cAtorMCei5NxJXUDEnHcd7E"

http://localhost:8881/api/services/app/GZGoddessFestival/TurnOn

hey实战

对于多个实例的应用，如并发锁。实际业务场景中分布式锁。

nginx配置

upstream backend {
       server 127.0.0.1:5000;
        server 127.0.0.1:6001;
    }
    server {
        listen       8881;
        server_name localhost;
        #charset koi8-r;
        #access_log logs/host.access.log main;
        location / {
          proxy_pass http://backend;
        }
}

我们分别在本地部署两个相同版本的实例，分别端口是5000,6001，如上是我们反向代理的配置。对外监听的8881端口。

2个实例本地启动，dotnetcore应用默认是5000端口

dotnet Api.Web.Host.dll --urls=http://localhost:6001/
dotnet Api.Web.Host.dll

我们获取对应接口的URL，准备负载测试如下，9000请求，100并发， HTTP POST，自定义HTTP HEADER 用于认证：

clip_image002 clip_image002[4]

clip_image002[6]

以上是简单的测试摘要，HTTP STATUS 200 共计 7136个，HTTP 500 149个，最慢的请求19.8秒

Nginx的日志

6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.132 Safari/537.36"
127.0.0.1 - - [10/Mar/2020:07:56:45 +0800] "POST /api/services/app/Ability/TestRsa HTTP/1.1" 200 455 "http://localhost:8881/swagger/index.html" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.132 Safari/537.36"
127.0.0.1 - - [10/Mar/2020:07:56:48 +0800] "POST /api/services/app/Ability/TestRsa HTTP/1.1" 200 455 "http://localhost:8881/swagger/index.html" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.132 Safari/537.36"
127.0.0.1 - - [10/Mar/2020:07:58:09 +0800] "GET /api/services/app/GZGoddessFestival/TurnOn HTTP/1.1" 404 0 "-" "Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.132 Safari/537.36"
127.0.0.1 - - [10/Mar/2020:07:59:06 +0800] "POST /api/services/app/GZGoddessFestival/TurnOn HTTP/1.1" 401 206 "-" "Fiddler"
127.0.0.1 - - [10/Mar/2020:07:59:16 +0800] "POST /api/services/app/GZGoddessFestival/TurnOn HTTP/1.1" 401 206 "-" "hey/0.0.1"
127.0.0.1 - - [10/Mar/2020:07:59:16 +0800] "POST /api/services/app/GZGoddessFestival/TurnOn HTTP/1.1" 401 206 "-" "hey/0.0.1"
127.0.0.1 - - [10/Mar/2020:07:59:16 +0800] "POST /api/services/app/GZGoddessFestival/TurnOn HTTP/1.1" 401 206 "-" "hey/0.0.1"
127.0.0.1 - - [10/Mar/2020:07:59:16 +0800] "POST /api/services/app/GZGoddessFestival/TurnOn HTTP/1.1" 401 206 "-" "hey/0.0.1"

以上日志中我们也看到hey的请求，TurnOn这个接口是用于游戏抽奖，我们使用简单的工具实现负载测试，用于检测程序接口在并发的TPS，同时验证接口是否有奖品超发的情况。