Linode Library Home
Linode Library RSS Feed
Home :: Web Servers :: Nginx :: Python uWSGI
Print View View Source

WSGI using uWSGI and nginx on Ubuntu 9.10 (Karmic)

Published: by

Warning

This document is an older, unmaintained guide. There may be a new guide available for this software.

The uWSGI server provides a non-FastCGI method for deploying Python applications with the nginx web server. In coordination with nginx, uWSGI offers great stability, flexibility, and performance. However, to deploy applications with uWSGI and nginx, you must compile nginx manually with the included uwsgi module.

Contents

Install uWSGI

Begin by issuing the following command to update your system and install dependencies for uWSGI:

apt-get update
apt-get upgrade
apt-get install build-essential psmisc python-dev libxml2 libxml2-dev python-setuptools

Issue the following sequence of commands to download and compile uWSGI:

cd /opt/
wget http://projects.unbit.it/downloads/uwsgi-0.9.6.5.tar.gz
tar -zxvf uwsgi-0.9.6.5.tar.gz
mv uwsgi-0.9.6.5/ uwsgi/
cd uwsgi/
python setup.py
make

Issue the following command to create a dedicated system user to run the uwsgi processes:

adduser --system --no-create-home --disabled-login --disabled-password --group uwsgi

Send the following sequence of commands to set the required file permissions:

chown -R uwsgi:uwsgi /opt/uwsgi
touch /var/log/uwsgi.log
chown uwsgi /var/log/uwsgi.log

Compile nginx with uWSGI Support

Issue the following commands to download and compile nginx with support for the uwsgi protocol. If you previously installed nginx from Ubuntu packages, remove them at this juncture. The following command sequence mirrors the procedure defined in the installation guide for nginx for compiling nginx from source:

apt-get install libpcre3-dev build-essential libssl-dev
cd /opt/
wget http://nginx.org/download/nginx-1.0.0.tar.gz
tar -zxvf nginx-1.0.0.tar.gz
cd /opt/nginx-1.0.0/
./configure --prefix=/opt/nginx --user=nginx --group=nginx --with-http_ssl_module
make
make install
adduser --system --no-create-home --disabled-login --disabled-password --group nginx
cp /opt/uwsgi/nginx/uwsgi_params /opt/nginx/conf/uwsgi_params
wget -O init-deb.sh http://library.linode.com/assets/652-init-deb.sh
mv init-deb.sh /etc/init.d/nginx
chmod +x /etc/init.d/nginx
/usr/sbin/update-rc.d -f nginx defaults
/etc/init.d/nginx start

Configure uWSGI

Issue the following command to download an init script to manage the uWSGI process, located at /etc/init.d/uwsgi:

cd /opt/
wget -O init-deb.sh http://library.linode.com/assets/651-uwsgi-init-deb.sh
mv /opt/init-deb.sh /etc/init.d/uwsgi
chmod +x /etc/init.d/uwsgi

Create an /etc/default/uwsgi file to specify specific settings for your Python application. The MODULE specifies the name of the Python module that contains your wsgi specification. Consider the following example:

File excerpt:/etc/default/uwsgi

PYTHONPATH=/srv/www/ducklington.org/application
MODULE=wsgi_configuration_module

If you want to deploy a "Hello World" application, insert the following code into the /srv/www/ducklington.org/application/wsgi_configuration_module.py file:

File:/srv/www/ducklington.org/application/wsgi_configuration_module.py

import os
import sys

sys.path.append('/srv/www/ducklington.org/application')

os.environ['PYTHON_EGG_CACHE'] = '/srv/www/ducklington.org/.python-egg'

def application(environ, start_response):
    status = '200 OK'
    output = 'Hello World!'

    response_headers = [('Content-type', 'text/plain'),
                        ('Content-Length', str(len(output)))]
    start_response(status, response_headers)

    return [output]

Issue the following commands to make this init script executable, ensure that uWSGI is restarted following the next reboot sequence, and start the service:

/usr/sbin/update-rc.d -f uwsgi defaults
/etc/init.d/uwsgi start

Configure nginx Server

Create an nginx server configuration that resembles the following for the site where the uWSGI app will be accessible:

File excerpt:nginx virtual host configuration

server {
    listen   80;
    server_name www.ducklington.org ducklington.org;
    access_log /srv/www/ducklington.org/logs/access.log;
    error_log /srv/www/ducklington.org/logs/error.log;

    location / {
        include        uwsgi_params;
        uwsgi_pass     127.0.0.1:9001;
    }

    location /static {
        root   /srv/www/ducklington.org/public_html/static/;
        index  index.html index.htm;
    }
}

All requests to URLs ending in /static will be served directly from the /srv/www/ducklington.org/public_html/static directory. Restart the web server by issuing the following command:

/etc/init.d/nginx restart

Additional Application Servers

If the Python application you've deployed requires more application resources than a single Linode instance can provide, all of the methods for deploying a uWSGI application server are easily scaled to rely on multiple uSWGI instances that run on additional Linodes with the request load balanced using nginx's upstream capability. See our documentation of proxy and software load balancing with nginx for more information. For a basic example configuration, see the following example:

File excerpt:nginx configuration

upstream uwsgicluster {
     server 127.0.0.1:9001;
     server 192.168.100.101:9001;
     server 192.168.100.102:9001;
     server 192.168.100.103:9001;
     server 192.168.100.104:9001;
}

server {
    listen   80;
    server_name www.ducklington.org ducklington.org;
    access_log /srv/www/ducklington.org/logs/access.log;
    error_log /srv/www/ducklington.org/logs/error.log;

    location / {
        include        uwsgi_params;
        uwsgi_pass     uwsgicluster;
    }

    location /static {
        root   /srv/www/ducklington.org/public_html/static/;
        index  index.html index.htm;
    }
}

In this example, we create the uwsgicluster upstream, which has five components. One runs on the local interface, and four run on the local network interface of distinct Linodes (the 192.168. addresses or the private "back end" network). The application servers that run on those dedicated application servers are identical to the application servers described above. However, the application server process must be configured to bind to the appropriate network interface to be capable of responding to requests.

More Information

You may wish to consult the following resources for additional information on this topic. While these are provided in the hope that they will be useful, please note that we cannot vouch for the accuracy or timeliness of externally hosted materials.

Creative Commons License

This guide is licensed under a Creative Commons Attribution-NoDerivs 3.0 United States License.

Last edited by System on Friday, April 29th, 2011 (r1460).