Toggle navigation

HTTP Proxying

Many people prefer using a standalone Python HTTP server and proxying that server via nginx, Apache etc.

A very stable Python server is CherryPy. This part of the documentation shows you how to combine your WSGI application with the CherryPy WSGI server and how to configure the webserver for proxying.

Creating a [UNKNOWN NODE title_reference] server

To run your application you need a [UNKNOWN NODE title_reference] file that starts up the WSGI Server.

It looks something along these lines:

from cherrypy import wsgiserver
from yourapplication import make_app
server = wsgiserver.CherryPyWSGIServer(('localhost', 8080), make_app())
try:
    server.start()
except KeyboardInterrupt:
    server.stop()

If you now start the file the server will listen on [UNKNOWN NODE title_reference]. Keep in mind that WSGI applications behave slightly different for proxied setups. If you have not developed your application for proxying in mind, you can apply the ProxyFix middleware.

Configuring nginx

As an example we show here how to configure nginx to proxy to the server.

The basic nginx configuration looks like this:

location / {
    proxy_set_header        Host $host;
    proxy_set_header        X-Forwarded-For $proxy_add_x_forwarded_for;
    proxy_pass              http://127.0.0.1:8080;
    proxy_redirect          default;
}

Since Nginx doesn’t start your server for you, you have to do it by yourself. You can either write an [UNKNOWN NODE title_reference] script for that or execute it inside a screen session:

$ screen
$ python start-server.py