linkxzhou/http_bench

golang HTTP stress testing tool, support single and distributed, http/1, http/2 and http/3.

Hi 👋, 我是linkxzhou

🙋 自我介绍

✍️  大家好，我是linkxzhou! 目前是某大厂是高级程序员，方向是做后台和云原生方向，正在转向AI工程领域，个人微信公众号是《周末程序猿》（微信，知乎，掘金同号），里面积累了一些技术文章，如果您感兴趣可以关注!

Github Stats

http_bench's Issues

关于-c参数执行报错的问题

以下是必现的问题
./http_bench -c 1000 -d 20s -url https://www.test.com
-c参数好像超过500就一定会报错,500就正常，我的机器CPU是8核
[ERROR] err: Get "https://www.test.com/kkk": dial tcp 192.168.1.3:443: i/o timeout (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https:/www.test.com/kkk": dial tcp 192.168.1.3:443: i/o timeout (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": dial tcp 192.168.1.3:443: i/o timeout (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": dial tcp 192.168.1.3:443: i/o timeout (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": context deadline exceeded (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": dial tcp 192.168.1.3:443: i/o timeout (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": dial tcp 192.168.1.3:443: i/o timeout (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": context deadline exceeded (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": context deadline exceeded (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": context deadline exceeded (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": context deadline exceeded (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": dial tcp 192.168.1.3:443: i/o timeout (Client.Timeout exceeded while awaiting headers)
[ERROR] err: Get "https://www.test.com/kkk": context deadline exceeded (Client.Timeout exceeded while awaiting headers)

每秒控制速率不够精准？

测试了一下，感觉每次实际打出来的rps 和设置的 总是有误差，有时候差距还比较大，控制不够精准

working with URLs from file uses only the first line (so it seems not random)

I think reading URLs from a file doesn't work as expected.
(I expected all the lines to be used and even at random.)

http_bench -n 1000 -c 100 -m GET -url-file stress.txt -verbose 1
My stress.txt

http://localhost
http://localhost/todo
http://localhost/todo/1
http://localhost/country
http://localhost/country/1
http://localhost/language
http://localhost/language/1
http://localhost/user
http://localhost/user/1

I can see that multiple lines are recognised:

Request params: {
        "sequence_id": 1670661308,
        "cmd": 0,
        "request_method": "GET",
        "request_body": "",
        "request_script_body": "",
        "request_httptype": "http1",
        "n": 1000,
        "c": 100,
        "duration": 10,
        "timeout": 3000,
        "qps": 0,
        "disable_compression": false,
        "disable_keepalives": false,
        "auth_username": "",
        "auth_password": "",
        "headers": null,
        "urls": [
                "http://localhost/todo",
                "http://localhost/todo/1",
                "http://localhost/country",
                "http://localhost/country/1",
                "http://localhost/language",
                "http://localhost/language/1",
                "http://localhost/user",
                "http://localhost/user/1",
                "http://localhost"
        ],
        "output": ""

but when I look at my webserver logging, I can see that only the first URL gets all the hits.
When I shuffle my stress.txt URLs, another first line gets all the hist

change usage of github.com/lucas-clemente/quic-go

change

github.com/lucas-clemente/quic-go

into

github.com/quic-go/quic-go

(and perhaps to the latest v0.32.0)

support http/3?

support gRPC?