Deploy multiple Flask applications with Nginx reverse proxy and Gunicorn

May 23rd, 2016

I often use the Python framework Flask for my projects because it is simple and I can focus on my domain logic. I usually also use Gunicorn to serve my applications with Nginx as a reverse proxy. My Gunicorn tasks are executed by an upstart script:

description "mytask"

start on runlevel [2345]
stop on runlevel [!2345]

respawn
setuid user
setgid user

env PATH=/path/to/flask/venv/
chdir /path/to/my/app
exec gunicorn --workers 3 -b unix:mysock.sock wsgi:app

My app is started by wsgi.py in my app folder:

from app import app

if __name__ == "__main__":
    app.run()

The upstart tasks create mysock.sock UNIX domain socket files in my apps folders. They will be used by Nginx to serve my applications.

In /etc/nginx/sites-available, I create my config file:

server {
    listen 80 default_server;
    server_name example.com;

    location /my-first-app/ {
        prox_pass http://unix:/path/to/my/app1/mysock1.sock:/;
    }

    location /my-second-app/ {
        prox_pass http://unix:/path/to/my/app2/mysock2.sock:/;
    }

    #...
}

The ":/" part means that if you try to access http://example.com/my-first-app/my-route/, your app will receive the route "/my-route" and not "/my-first-app/my-route" that would not match the following route in your Flask app:

@app.route('/my-route', methods=['GET'])
def my_route():
    pass

I hope this basic tutorial helped. Feedbacks in comments are welcome!