It’s pretty common that when I want to deploy a Django project that I’ll want to use the Nginx + uWSGI combo. For reference I’m going to include a few rules for setting this up on an Ubuntu system.
For this to work you should know two things. The first is that I expect your django project to reside in /var/www/project and that your project’s virtual environment is at /var/www/project/env. I also expect that you’ve installed python and you can run your django project with the standard python manage.py runserver 0.0.0.0:8000 command. This tutorial will replace that command.
To get started you want to install Nginx, uWSGI, and the uWSGI python plugin.
sudo apt-get update
sudo apt-get install -y nginx uwsgi uwsgi-plugin-python
After installing you’ll want to write an Nginx configuration file for your django project. In this case the upstream should point to the unix socket that you want to use to serve your django project.
# upstream django server
upstream django {
server unix:///run/uwsgi/app/uwsgi/socket;
}
# configuration of the server
server {
# the domain name it will serve for
server_name localhost;
listen 8000;
charset utf-8;
# max upload size
client_max_body_size 15M;
# Django static
location /static {
alias /www/var/project/static;
expires off;
# fix issue with Vagrant and sendfile
sendfile off;
}
# Django app
location / {
uwsgi_pass django;
include /etc/nginx/uwsgi_params;
}
}
Link this file and start Nginx:
ln -s /www/var/project/deploy/nginx.conf /etc/nginx/sites-enabled/
sudo /etc/init.d/nginx start
Next you need to make a python uWSGI file. Place this file at /var/www/project/wsgi.py. It should be at the same level as your settings.py file. In Django 1.5 and later this file is generated for you when you start a new django project.
import os
os.environ.setdefault("DJANGO_SETTINGS_MODULE", "settings")
from django.core.wsgi import get_wsgi_application
application = get_wsgi_application()
Then you want to write yourself a uWSGI ini file. The unix socket is determined by default by the uwsgi process and resides at /run/uwsgi/app/uwsgi/socket.
[uwsgi]
##
# Django-related settings
##
# the base directory (full path)
chdir = /var/www/project
# Django's wsgi file
module = wsgi
# the virtualenv (full path)
home = /var/www/project/env
##
# process-related settings
##
# master
master = true
# maximum number of worker processes
processes = 10
# clear environment on exit
vacuum = true
# automatically kill workers on master's death
no-orphans = true
# set mode of created UNIX socket
chmod-socket = 660
# place timestamps into log
log-date = true
# user identifier of uWSGI processes
uid = www-data
# group identifier of uWSGI processes
gid = www-data
# daemonize and log output
daemonize = /var/log/uwsgi/project.log
Then link the uWSGI file and start uWSGI.
ln -s /var/www/deploy/uwsgi.ini /etc/uwsgi/apps-enabled/
sudo /etc/init.d/uwsgi start
Now you should be able to browse to http://0.0.0.0:8000 and see your django project! If you make changes and want to reload you can simply run sudo /etc/init.d/uwsgi reload.
One thing that I noticed while deploying was that my static files did not serve completely, meaning that my javascript files seemed to be cut off. After doing a little bit of searching it appears that a bug with Vagrant, nginx, and shared folders means you need to add sendfile off; to the static file section of your nginx.conf. I initially thought this was a problem with grunt, which I use for generating static files, but after checking there was simply no problem with the generated files.