Render app does not respect gunicorn timeout

1

I have a Python (Django) app deployed using Render. I migrated over from Heroku yesterday.

This is my start command: gunicorn wsgi --worker-tmp-dir /dev/shm --workers=32 --threads=32 --log-level debug --timeout 10

As you can see, there is a 10 second timeout for requests. This used to be respected in Heroku. If a request took over 10 seconds, it would be terminated. However, I am not seeing this work in Render. Looking at my p99, there are requests that are taking more than 60 seconds.

I also have some application logs that inform me of request timings, and I am consistently seeing requests take over 10 seconds

2

the timeout attribute like that would absolutely be honoured, it’s at the python level so we’re simply running what you configure.

Per the gunicorn docs - is the worker silent? Settings — Gunicorn 23.0.0 documentation