Spike in the memory usage - Heroku about daphne HOT 7 CLOSED

Memory usage of a worker is mostly dominated by Django and the app itself, though 500MB is quite a lot. Some questions:

• What version of Channels, Daphne, Django and Python are you using?
• Does the same thing happen with a normal WSGI process with, say, gunicorn with similar code?
• What sort of throughput are the workers getting?
• What channel layer are you using?

from daphne.

• What version of Channels, Daphne, Django and Python are you using?
  channels==0.14.0
  daphne==0.12.1
  Django==1.9.5
  Python : 2.7.1
• Does the same thing happen with a normal WSGI process with, say, gunicorn with similar code?
  Couldn't say for sure. I will try to reproduce the same load with gunicorn and will let you know
• What sort of throughput are the workers getting?
  throughput was around 200request per minute but I can assure you it was not that much.
• What channel layer are you using?
  Redis.

from daphne.

You're on old versions of channels and daphne; I'd at least try upgrading them first as some memory leaks were fixed a while ago.

from daphne.

from daphne.

I agree the footprint is too high, I want to make sure we can isolate it to the most recent release of channels and that it's not there when you run Django in normal WSGI mode, otherwise it's not a problem I can solve!

from daphne.

Closing due to lack of response.

from daphne.

Deploying Python Applications with Gunicorn fix this problem for me-

pip install gunicorn
pip freeze requirements.txt

Or, you can manually add it to your requirements.txt file.

Next, revise your application’s Procfile to use Gunicorn. Here’s an example Procfile

from daphne.