-
Notifications
You must be signed in to change notification settings - Fork 26
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
slow_cooker slow-down after about 5-10 minutes #59
Comments
Happy to dig into this: this is slow_cooker built from master? |
Unable to reproduce from master on a gcloud host to another gcloud host running nginx, lemme keep digging.
|
An overnight run on gcloud was ok, no slowdown noted. I'll try with Updated: That's a lot of sockets in TIME_WAIT, I wonder if we're eventually running into an issue opening new sockets. Let me figure out how to set some socket options in go's http client and we can tweak SO_LINGER Updated 2: No slowdown with |
Ok, I definitely see this on the mac with
|
It continues to drop in throughput, after 20 minutes it's down to 50%. After another 5 minutes, we're down to 25% throughput. this is troubling. |
Keeping total requests the same but reducing concurrency to 1 prevents the slow down from occurring.
|
Let me know if #60 helps with this. |
After about 5-10 minutes of running slow_cooker against a simple python http server, throughput drops by 25%.
:; slow_cooker -concurrency 4 -qps 30 http://localhost:8000
Restarting the python server without slow_cooker yields no improvement. Restarting slow_cooker, however, resumes at 120rps.
The same behavior is exhibited with and without -noreuse, presumably because the python server only speaks HTTP/1.0:
slow_cooker doesn't appear to be hitting any file descriptor limits:
lsof is pretty uninteresting
The text was updated successfully, but these errors were encountered: