Setting up Nginx

If you have previously had Nginx installed, and used the same location as before, then there’s a chance you do not need to mess with init.d. However if you need start/stop capability as well as allowing the server to auto start upon a reboot then perform the following steps. (If you installed nginx to a location other than /opt/nginx you may need to edit the file for the correct paths)

cd /etc/init.d
wget http://kbeezie.com/examples/nginx
chmod +x /etc/init.d/nginx
/usr/sbin/update-rc.d -f nginx defaults
/etc/init.d/nginx start

Now you can start, stop or restart nginx easily.

Configuring Nginx

By default the Nginx configuration file (/opt/nginx/conf/nginx.conf) looks something like this (commented lines were removed). Remember, Passenger already added the needed lines to the http { } block in your actual nginx.conf file.

worker_processes  1;
 
events {
    worker_connections  1024;
}
 
http {
    include       mime.types;
    default_type  application/octet-stream;
 
    sendfile        on;
    keepalive_timeout  65;
	gzip  on;
 
    server {
        listen       80;
        server_name  localhost;
 
        location / {
            root   html;
            index  index.html index.htm;
        }
 
        error_page   500 502 503 504  /50x.html;
        location = /50x.html {
            root   html;
        }
    }
}

Which would show you something like this when you visit the browser:

So we know it works… but its pretty much just pure static content. Let us change the default localhost into a Python application using Passenger. For this we’re going to use the default web folder provided by Nginx, the only difference is, we’ll need to create a Public folder under it.

cd /opt/nginx/html
mkdir public
vi passenger_wsgi.py

You don’t have to use vi to create your test Python script, you might even be an emacs guy cursing me for suggesting such monstrosity ^_^. The Python app below will simply output environment variables to the browser.

import sys
import os
 
def application(environ, start_response):
	start_response("200 OK", [])
	ret = ["%s: %s\n" % (key, value)
		for key, value in environ.iteritems()]
	return ret

Do note however, that the file is outside of the /public location, and that it is named passenger_wsgi.py. Placing the file elsewhere, or giving it a different filename will cause it not to be loaded by Passenger.

Now we need to tweak our Nginx configuration to load the Python WSGI application.

worker_processes  1;
 
events {
    worker_connections  1024;
}
 
http {
    include       mime.types;
    default_type  application/octet-stream;
 
    sendfile        on;
    keepalive_timeout  65;
	gzip  on;
 
	passenger_root /usr/local/lib/ruby/gems/1.9.1/gems/passenger-2.2.5;
	passenger_ruby /usr/local/bin/ruby;
 
    server {
        listen       80;
        server_name  localhost;
		root /opt/nginx/html/public/; #the folder name public must always be used
		passenger_enabled on; #this actually turns passenger on at this location
    }
}

A couple things you’ll notice, the passenger configuration parameters were already added by Passenger in the http { } block. In the above you’ll notice we don’t even use a location block in the server { } block. At a bare minimum we need the server_name, root path, and passenger_enabled line. (you don’t even need the listen 80; since 80 is the default port unless otherwise specified in the http {} block)

We’ll need to restart nginx for the changes to take effect.

# /etc/init.d/nginx restart
restarting nginx: nginx.

Load up the server in the browser, and you should be presented with enviroment outputs served up by your Python WSGI application:

You are now setup to easily deploy Python WSGI applications from Nginx. If you make a change to the script, you may need to restart nginx, as the application is stored in memory.

Using DJango and Web.py with Passenger

You’re probably wondering how would you deploy a framework such as Django or Web.py, with Passenger, especially since the main file must always be called passenger_wsgi.py. Instead of the test script above, you would simply use the code below and place your Django application in the same folder (above public). For example in the /opt/nginx/html folder you may see the following folders : django, public, and a folder with your django application name with the file passenger_wsgi.py along side them.

import os, sys
nginx_configuration= os.path.dirname(__file__)
project = os.path.dirname(nginx_configuration)
workspace = os.path.dirname(project)
sys.path.append(workspace) 
 
os.environ['DJANGO_SETTINGS_MODULE'] = 'testapp.settings'
import django.core.handlers.wsgi
application = django.core.handlers.wsgi.WSGIHandler()

Detailed Guide : Click for a more detailed guide to setting Django up with Passenger and Nginx

With Web.Py the documentation often calls for web.application which is self-hosted.

app = web.application(urls, globals())
if __name__ == "__main__": app.run()

Below is an example Web.py app as stored in a passenger_wsgi.py file. You’ll note that instead of the two lines above, it is now a single line with the modified function assigned to application. This is covered in Web.py’s documentation regarding Apache and mod_wsgi, so this modification should work with most WSGI servers.

import web
 
urls = (
  '/', 'index'
)
 
application = web.application(urls, globals()).wsgifunc()
 
class index:
    def GET(self):
        return "Hello, world!"

Further Information

For further information regarding Nginx itself, I strongly recommend taking a look at the Wiki

For more detailed instructions on using Passenger with Nginx, refer to the User Manual. The user manual will be useful if you wish to take it a step further and venture a look at building Ruby applications, of which you already have capabilities installed.

• Previous Entry: Scraping Google Front Page Results
• Next Entry: Nginx and Django