These are instructions for serving Django and Flask applications in the simplest form. These are based on an Ubuntu Server.
- Examples I'll Use
- Step 1: Setup Your Project
- Step 2: Setup Nginx
- Step 3: Setup Gunicorn
- Step 4: Reload Everything
- Optional: Dotfiles
For these examples I will use the following.
- This is based on an Ubuntu Box.
- ubuntu as my linux user
- www-data as the default server group
- foo as my Flask or Django project name
1: Your server needs a virtual environment. I will be using virtualenvwrapper.
sudo apt-get install virtualenvwrapper
My project path will be /home/ubuntu/projects/foo
2: So the first thing to do is make the virtualenvwrapper:
mkdir ~/projects && cd ~/projects
mkvirtualenv foo
(or, if using my dotfiles below):
mkproject foo
workon foo
3: Place your Django or Flask project in /home/ubuntu/projects/foo
1: First install nginx, and make a new file for our virtual host.
sudo apt-get install nginx
sudo touch /etc/nginx/sites-available/foo
sudo ln -s /etc/nginx/sites-available/foo/ etc/nginx/sites-enabled/foo
2a: For Flask, Add the following
server {
location / {
proxy_pass http://localhost:8000;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
}
location /static {
alias /home/ubuntu/projects/foo/static/;
}
}
2b: For Django Add the following (Whatever your Paths are):
server {
server_name www.foo.com;
return 301 $scheme://foo.com$request_uri;
}
server {
listen 80;
server_name foo.com;
charset utf-8;
location /media {
alias /home/ubuntu/projects/foo/foo/media/;
}
location /static {
alias /home/ubuntu/projects/foo/foo/static/;
}
location / {
proxy_pass http://127.0.0.1:8000;
proxy_set_header Host $host;
proxy_set_header X-Real-IP $remote_addr;
proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
}
}
Notice that this will redirect all traffic to www.foo.com to foo.com.
3: Restart Nginx
sudo service nginx restart
This is the Python WSGI HTTP server.
1: You should have a requirements.txt of some sort in your project that you made for PIP packages. Simply include gunicorn in your project's requirements.txt (or include a version like I do below):
...
gunicorn==19.4.5
2: Go to your project, and make sure you are in the virtual enviroment to install it.
workon foo
pip install -r requirements.txt
To manually test while still in your virtual environment do one of the following commands below. Then load your webserver's IP address or domain name if you pointed it. (Services like AWS need you to open port 80 in your security policy for HTTP)
3a: For Flask, if you created a default app module, simply run it as:
gunicorn app:app --bind localhost:8000
3b: For Django, go to the folder your wsgi.py file is and run:
gunicorn --bind localhost:8000
4: If all is working as it should you can CTRL+C to cancel it. If you have problems, see how you run it here: http://docs.gunicorn.org/en/stable/run.html
Supervisor is a process controller that makes it easy to manage running programs/tasks and keep them alive.
1: Install supervisor globally.
sudo apt-get install supervisor
2: Create a project file for supervisor
sudo vim /etc/supervisor/conf.d/foo.conf
3a: Enter the following for Flask:
[program:foo]
command = /home/ubuntu/.virtualenvs/foo/bin/gunicorn app:app --bind 0.0.0.0:8000
directory = /home/ubuntu/projects/foo
autorestart=true
3b: Enter the following for Django:
[program:foo]
command = /home/ubuntu/.virtualenvs/foo/bin/gunicorn --bind 0.0.0.0:8000
directory = /home/ubuntu/projects/foo
autorestart=true
1: Let's stop any gunicorn instances and stop supervisor to make sure that it loads the default configuration file and we start with a clean slate
sudo pkill gunicorn
sudo service supervisor stop
sudo unlink /var/run/supervisor.sock
2: Load the default configuration for supervisor and re-read the folders
sudo supervisord -c /etc/supervisor/supervisord.conf
3: Now update supervisor and start our application
sudo supervisorctl update
sudo supervisorctl start foo
4: You are done. To see many of the options provided in the CLI or configuration files visit: http://supervisord.org/running.html
Note there are two CLI commands supervisor and supervisorctl.
First, sudo supervisorctl reread - Only updates the changes. It does not restart any of the managed applications, even if their configuration has changed. New application configurations cannot be started with this command.
Second, sudo supervisorctl update - Restarts the applications whose configuration has changed. After the update command, new application configurations becomes available to start, but do not start automatically until the supervisor service restarts or system reboots (Even if autostart is on) you still need to run it.
Third, sudo supervisorctl start foo to run the program.
sudo supervisorctl stop foo to stop the program.
To debug a problem, either check the job, or check the logs:
supervisorctl tail foo
cd /var/log/supervisor/ # tail logs here.
1: (Skip this if you don't need it) If you'd like to see my dotfiles which make virtualenvwrapper easier to manage, visit my config-ubuntu repository.
2: On your server, you can do the following:
git clone git@github.com:JREAM/config-ubuntu.git
cd config-ubuntu
sudo ./install.sh
Type a Command: dot
3: Otherwise, You can simply copy the .virtualenvs folder and the .bashrc file where you see the block called PYTHON. You will need to change the paths in .bashrc and the .virtualenvs folder. This setup was introduced to me by Dan Sackett.