What is the clear way to run flask application with gevent backend server and utilize all processor cores? I have idea to run multiple copies of flask application where gevent WSGIServer listen one port in diapason 5000..5003 (for 4 processes) and nginx as load balancer.
But I'm not sure that this way is the best and may be there are some other ways to do it. For example, master process listen one port and workers process incoming connections.
I'll take a shot!
location / {
include proxy_params;
proxy_pass http://127.0.0.1:5000;
}
This is a simple flask app that i will be using for this example.
myapp.py:
from flask import Flask
app = Flask(__name__)
@app.route("/")
def hello():
return "Hello World!"
if __name__ == "__main__":
app.run()
Okay so I know that you said that you wanted to use gevent, but if you are willing to compromise on that a little bit I think you would be very happy with this setup.
[uwsgi]
master = true
plugin = python
http-socket = 127.0.0.1:5000
workers = 4
wsgi-file = myapp.py
callable = app
If you must have gevent you might like this little setup
config.py:
import multiprocessing
workers = multiprocessing.cpu_count()
bind = "127.0.0.1:5000"
worker_class = 'gevent'
worker_connections = 30
Then you can run:
gunicorn -c config.py myapp:app
Thats right you have a worker for each cpu and 30 connections per worker.
See if that works for you.
If you are really sold on using nginx as a load balancer try something like this in your http section
upstream backend {
server 127.0.0.1:5000;
server 127.0.0.1:5002;
server 127.0.0.1:5003;
server 127.0.0.1:5004;
}
then one of these in the server section
location / {
include proxy_params;
proxy_pass http://backend;
}
Good Luck buddy!
