-
Notifications
You must be signed in to change notification settings - Fork 3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Locust never gets past 10 req/s, despite the server being much quicker than that #223
Comments
You need to adjust the default wait times. |
Ah, excellent, that fixes it, I do now get 600 or so. Sorry, I clearly didn't understand the docs there, I'd skimmed past those assuming they were for managing timeouts. Seems a bit strange that 0 and 0 aren't the default there to me, but fair enough, I guess that's a tradeoff for realistic users vs pure benchmarking. Might be nice to include the waiting time in the results explicitly though, so it's clearer for other foolish people like me. Thanks! |
No worries, it can be made more clear. |
For future reference, you can use the option This will allow you to properly simulate a large number of users without having to change their timeout. |
@pimterry can u explain what does this min_wait and max_wait do ? do we need to decrease this times or increase to achieve high rps ? |
@shaikshakeel did you check the documentation? |
@shaikshakeel |
Hey there, I also have this problem. I deployed locust on my local kubernetes cluster. The host is also inside the cluster. Cpu usage is 25% and ram usage is 400 mb when I start locuat with 30.000 ccu and 1000 hatch rate i get 400 req/s. There are jsut two get calls inside the test. Min wait time 100 and max is 150. Are there any troubleshooting steps i can go through to fond the problem ? |
please ask general questions in the Slack channel... this issue is closed |
Why do we need to specify wait time when there is only one task in my Taskset. |
The wait time I believe just means how long to wait between task execution, even if the same task is repeated over and over, and doesn't have to do with different tasks. |
I've got a very simple locustfile, running against a very simple local server, and locust tells me it can do 10req/s, while ab tells me it can do 600+.
My locustfile:
My server is a very simple flask app, running locally under CherryPy, and returning a fixed string value.
If I run this with
locust -c 10 -r 10 -n 1000 --no-web -f locustfile.py
(1000 requests with 10 users, all hatched immediately) I end up with:It hovers around 10, but doesn't pass 10 requests per second at any point. Each request only takes 6ms avg, suggesting each user could do at least 100/s, so 10 users -> at least 1,000 max from locust itself. That should definitely make my server the bottleneck.
Unfortunately though, my server can do way more than 10 requests/s. If I use
ab
instead, running:(1000 requests, with 10 threads)
I get :
About 600 requests per second, much more in line with what I'd expect.
Why does Locust say I can only do 10 requests per second, when I can safely do 600? Am I missing something obvious in my configuration, or is ab doing something enormously different from locust that causes this effect? Seems like it should be easy to reproduce the same basic result in this case with both tools.
The text was updated successfully, but these errors were encountered: