wrk VS wrk2

> My initial requirement was to send requests with unique parameters. To the best of my knowledge, no tool could do this.

wrk does this with lua. https://github.com/wg/wrk/blob/master/src/wrk.lua

Also even things like the venerable jmeter supported pulling parameters from a csv file.

We used wrk2, a fantastic tool to generate stable load on the server. Wrk2 improves on wrk and adds the ability to generate sustained load at a specific rate (-R parameter).

I am using a tool called wrk to make a bunch of requests to my website, and changing the concurrency of the requests drastically changes the result, with concurrency set to 1, I get 894 requests made in 10 seconds, and when I set concurrency to 10, I get 8549 requests. This linear scaling stops when increasing the concurrency more drastically (18805 requests for 100 and 19814 for 500 concurrency).

Im creating multithreaded async http server for learning purposes and i'm facing problem when benchmarking with wrk. I get socket read errors on every connection:

We use locust at work but I HIGHLY recommend wrk for a very robust yet simple load testing tool.

https://github.com/wg/wrk

And of course, this talk by Gil Tene is fantastic if you're interested in load testing stats https://www.youtube.com/watch?v=lJ8ydIuPFeU

Good morning what tool do you use to test your infra in terms of http load ? A tool that works, I tested : - https://github.com/tsenart/vegeta but it returns 0 errors or a http_net error from Golang - LOIC (Low Orbit Ion Canon) https://github.com/NewEraCracker/LOIC but the requests do not appear in my nginx logs and I feel no slowdown - Apache Jmeter https://jmeter.apache.org/ but I can't drop my infra and I have Java socket closed errors - K6 https://k6.io/ but I can't bring down my infra with - wrk https://github.com/wg/wrk no matter what parameter I put it doesn't make enough requests per second, I put the same parameters as on a tutorial and I don't get the same result...

You could try to just blast it with wrk or bombardier. Can easily get around 50k requests on consumer machine.

Sure, first I did the load testing on the same machine. The same machine ran web servers, php-fpm and ab / wrk programs.

There are http benchmarking tools like wrk [0]. You don't need a ddos service for that.

[0] https://github.com/wg/wrk

> This runs a benchmark for 30 seconds, using 2 threads, keeping 100 HTTP connections open, and a constant throughput of 2000 requests per second (total, across all connections combined).

Some distros include `ab`[2] which is also good, but wrk2 improves on it (and on wrk version 1) in multiple ways, so that's what I use myself.

[1] https://github.com/giltene/wrk2

[2] https://httpd.apache.org/docs/2.2/programs/ab.html

I've had my eyes on wrk2 [1]

1. https://github.com/giltene/wrk2

But I am curious, what does HN use? Any tips?

We used wrk2, a fantastic tool to generate stable load on the server. Wrk2 improves on wrk and adds the ability to generate sustained load at a specific rate (-R parameter).

I use tools like vegeta and wrk2 to answer those questions.

Needs (2015).

I loved the talks from Gil Tene.

I always reach for his fork of wrk whenever I need to test throughput:

https://github.com/giltene/wrk2

That's why a lot of people just use something like wrk or wrk2 (highly recommended to run it on a separate machine) and benchmark the ability to serve actual requests.

To get the most out of preload, I preloaded all files that the experimental endpoint needs to include. As a benchmarking tool, I use wrk2 — a more advanced Apache Benchmark analog — to keep it simple and provide more flexibility to generate loads similar to a real-life one.

i use https://github.com/giltene/wrk2 pretty regularly.

it has decent lua hooks to customize behavior but i use it in the dumbest way possible to hammer a server at a fixed rate with the same payload over and over.

i run it by hand after a big change to the server to make sure nothing obviously regressed. i used to run it nightly in a jenkins job but 99% of the time no one looked at results. it was nice to see if assumptions on load a single node could handle didn't hold anymore.

I use wrk2 because it overcomes coordinated omission.