Django on the cheap

Story on how to setup Django as cheaply as possible.
NB: Still work in progress.

This story will show how to create a Django application and deploy it to a hosting provider as cheaply as I could possible do it.

Prerequisites

I will not be free, but my aim is to have a working solution that costs less than 10 USD pr. month.

This will not be a complete from zero to hero tutorial, rather I'll assume that you know enough basic Django to be able to apply the shown yourself.

Overview

Local setup

  • Create Django
  • Dockerize Django
  • Postgres
  • NGINX and gunicorn

Github and github actions

  • Github repo
  • Github actions to test

Hosting provider

  • Dokku
  • Push to deploy
  • Local and dev/prod specific files

Database

  • ElephantHQ
  • Update settings

Workes (rabbitmq)

  • RabbitMQ instance locally
  • Worker locally
  • RabbitMQ hosted
  • Workers hosted

Local Setup

Create local Django application

First step is to setup a local working Django application. We will setup a virtual environment to hold our python and Django dependencies.

$ virtualenv venv -p python3
created virtual environment CPython3.10.4.final.0-64 in 257ms
  creator CPython3Posix(dest=/django_on_the_cheap/venv, clear=False, no_vcs_ignore=False, global=False)
  seeder FromAppData(download=False, pip=bundle, setuptools=bundle, wheel=bundle, via=copy, app_data_dir=/home/username/.local/share/virtualenv)
    added seed packages: pip==22.0.4, setuptools==62.1.0, wheel==0.37.1
  activators BashActivator,CShellActivator,FishActivator,NushellActivator,PowerShellActivator,PythonActivator
$ source venv/bin/activate
(venv) /django_on_the_cheap$ python --version
Python 3.10.4
(venv) /django_on_the_cheap$

In this case we are on Python 3.10 as I'm using an Ubuntu 22.04. Let's also install Django while we are at it:

$ pip install django
Collecting django
  Using cached Django-4.0.5-py3-none-any.whl (8.0 MB)
Collecting sqlparse>=0.2.2
  Using cached sqlparse-0.4.2-py3-none-any.whl (42 kB)
Collecting asgiref<4,>=3.4.1
  Using cached asgiref-3.5.2-py3-none-any.whl (22 kB)
Installing collected packages: sqlparse, asgiref, django
Successfully installed asgiref-3.5.2 django-4.0.5 sqlparse-0.4.2
WARNING: You are using pip version 22.0.4; however, version 22.1.2 is available.
You should consider upgrading via the '/home/username/django_on_the_cheap/venv/bin/python -m pip install --upgrade pip' command.

We will ignore the pip upgrade for now as it is not super important.

We initialize a Django project:

$ django-admin startproject conf .

I call the project "conf" for the reason that django-admin will create a folder with this name holding files like the settings.py file. Thus, I find that calling the project "conf" or "config" makes sense and avoids some nameing confusion down the line. The final "." just means that we will not create an extra layer of folders that django does by default. I find that for smaller projects this makes it a bit easier to manage.

Let's migrate the database and spin up the development server:

$ python manage.py migrate
Operations to perform:
  Apply all migrations: admin, auth, contenttypes, sessions
Running migrations:
  Applying contenttypes.0001_initial... OK
  Applying auth.0001_initial... OK
...
...
...
$ $ python manage.py runserver
Watching for file changes with StatReloader
Performing system checks...

System check identified no issues (0 silenced).
June 05, 2022 - 11:20:55
Django version 4.0.5, using settings 'conf.settings'
Starting development server at http://127.0.0.1:8000/
Quit the server with CONTROL-C.

And if you go to the localhost url, you should see the django index page.

We will finally do a pip freeze in order to generate a requirements file with all our dependencies:

$ pip freeze > requirements.txt

We also need to prepare for using static files. Add this to the settings.py file in the config folder:

STATICFILES_DIRS = [
    BASE_DIR / "static",
]
STATIC_ROOT = BASE_DIR / "staticfiles"

Dockerize the application

Now we want to run our Django app in a Docker container, and, we also want to use docker-compose to manage this, as we will add more containers in the future, like Nginx and a database. Docker-compose just makes this easier to manage.

Make sure you have docker and docker-compose installed:

$ docker --version
Docker version 20.10.16, build aa7e414

$ docker-compose --version
docker-compose version 1.29.2, build 5becea4c

Create a docker-compose.yml file in the root of your project and add this:

version: '3.7'

services:
  django:
    build: 
      dockerfile: ./docker/local/Dockerfile
      context: .
    environment:
      - DEBUG=1
      - PYTHONDONTWRITEBYTECODE=1
      - PYTHONUNBUFFERED=1
      - PORT=8000
    ports:
      - 8000:8000
    volumes:
      - .:/app
    command: python manage.py runserver 0.0.0.0:8000

and create a folder called docker and within that folder, another folder called local. In there (in docker/folder) create a file called Dockerfile and put this in there:

FROM python:3.10

EXPOSE 8000

# Keeps Python from generating .pyc files in the container
ENV PYTHONDONTWRITEBYTECODE=1
# Turns off buffering for easier container logging
ENV PYTHONUNBUFFERED=1
ENV DEBUG 0

RUN apt-get update && apt-get upgrade -y && apt-get install -y \
  && apt-get clean

RUN pip install -U pip

# Install pip requirements
COPY . /app/
RUN python -m pip install -r /app/requirements.txt

# collect static files
WORKDIR /app
RUN python manage.py collectstatic --noinput

CMD python manage.py runserver

run

$ docker-compose up

and your application should be available on localhost:8000

Postgress database

We don't want to keep using the development sqlite database but rather use a Postgres database. This will also make it easier once we deploy to a production environment in the cloud. We do this in the docker-compose yml file.

Update the docker-compose.yml file to this:

version: '3.7'

services:
  db:
    image: postgres:12
    ports:
     - "5432:5432"
    environment:
     - POSTGRES_USER=postgres
     - POSTGRES_PASSWORD=12345
     - POSTGRES_DB=local_db
    volumes:
      - postgres_data:/var/lib/postgresql/data/

  django:
    build: 
      dockerfile: ./docker/local/Dockerfile
      context: .
    environment:
      - DEBUG=1
      - PYTHONDONTWRITEBYTECODE=1
      - PYTHONUNBUFFERED=1
      - PORT=8000
    ports:
      - 8000:8000
    volumes:
      - .:/app

volumes:
  postgres_data:

We are adding a postgres service and a volume to persist data. Do a new:

$ docker-compose up --build

We now have a running postgres database instance we can make Django connect to. Update the Database key in the settings.py file to:

DATABASES = {
    'default': {
        'ENGINE': 'django.db.backends.postgresql',
        'NAME': 'local_db', # set in docker-compose.yml as POSTGRES_DB
        'USER': 'postgres', # set in docker-compose.yml as POSTGRES_USER
        'PASSWORD': "12345", # set in docker-compose.yml as POSTGRES_PASSWORD
        'HOST': 'db', # set in docker-compose.yml as the service name
        'PORT': 5432 # default postgres port as the port number
    }
}

We also need to add postgres to our dependencies. Do this by adding psycopg2-binary to the requirements.txt file and run a docker-compose up --build again.

Now, however, we are getting this output:

django_1  | You have 18 unapplied migration(s). Your project may not work properly until you apply the migrations for app(s): admin, auth, contenttypes, sessions.
django_1  | Run 'python manage.py migrate' to apply them.

as we are using a new database it hasn't been migrated yet. For now, we do this manually by logging in to the django docker container and manually run migrate.

$ docker-compose run django bash
Creating django_on_the_cheap_django_run ... done
root@49efabfd6d54:/app# python manage.py migrate
Operations to perform:
  Apply all migrations: admin, auth, contenttypes, sessions
Running migrations:
  Applying contenttypes.0001_initial... OK
  Applying auth.0001_initial... OK
...
...
...

If you try and go to localhost:8000 now, you will first get a warning relating to missing ssl (https), and second, if you bypass this warning, an error like:

Invalid HTTP_HOST header: '0.0.0.0:8000'. You may need to add '0.0.0.0' to ALLOWED_HOSTS.

So, we will update our ALLOWED_HOSTS in the settings.py file to include localhost.

ALLOWED_HOSTS = ["localhost", "0.0.0.0"]

Now you should be able to successfully load the django index page.

Gunicorn

For now we have used the development server that comes with Django. But we can't use this in production, so we wanna use something else, and in this case Gunicorn. So lets first add it to our requirements and rebuild our images.

requirements.txt

asgiref==3.5.2
Django==4.0.5
sqlparse==0.4.2
psycopg2-binary
gunicorn

and run docker-compose up --build

Update the Dockerfile to use the gunicorn instead of the development server:

WORKDIR /app
RUN python manage.py collectstatic --noinput

CMD gunicorn conf.wsgi --bind 0.0.0.0:8000

Nginx

Now, if you try and go to http://0.0.0.0:8000/admin you will notice that the static assets (css, js) are missing. We need to have nginx setup to serve these. Let's add nginx as a service to our docker-compose.yml file.

  nginx:
    build: 
      dockerfile: ./docker/local/nginx/Dockerfile
      context: .
    volumes:
      - static_volume:/app/staticfiles      
    ports:
      - 8000:80
    depends_on:
      - django
    networks:
      - nginx_network

We are going to make a few changes to both our Dockerfile and docker-compose.yml in order to setup nginx. The full docker-compose.yml should look like:

version: '3.7'

services:
  db:
    image: postgres:12
    ports:
     - "5432:5432"
    environment:
     - POSTGRES_USER=postgres
     - POSTGRES_PASSWORD=12345
     - POSTGRES_DB=local_db
    volumes:
      - postgres_data:/var/lib/postgresql/data/

  nginx:
    build: 
      dockerfile: ./docker/local/nginx/Dockerfile
      context: .
    volumes:
      - static_volume:/app/staticfiles      
    ports:
      - 8000:80
    depends_on:
      - django
    networks:
      - nginx_network

  django:
    build: 
      dockerfile: ./docker/local/Dockerfile
      context: .
    environment:
      - DEBUG=1
      - PYTHONDONTWRITEBYTECODE=1
      - PYTHONUNBUFFERED=1
      - PORT=8000
    expose:
      - 8000
    volumes:
      - .:/app
      - static_volume:/app/staticfiles
    depends_on:
      - db
    networks:
      - nginx_network

volumes:
  postgres_data:
  static_volume:

networks: 
  nginx_network:
    driver: bridge  

First: We are adding the nginx service.
Second: We are creating a static_volume volume to hold static assets that should be served by nginx.
Third: We are explicitly adding a network, the nginx_network (this is not strictly necessary).

We will also add a nginx config file which we copy into the nginx docker container. In the docker/local folder make a subfolder called "nginx", in here, we will have two files: Dockerfile and nginx.conf.
The Dockerfile should look like:

FROM nginx:1.21-alpine

RUN rm /etc/nginx/conf.d/default.conf
COPY ./docker/local/nginx/nginx.conf /etc/nginx/conf.d

and the nginx.conf:

upstream django_on_the_cheap {
    server django:8000;
}

server {

    listen 80 default_server;
    server_name _;

    location / {
        proxy_pass http://django_on_the_cheap;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header Host $host;
        proxy_redirect off;
    }

    location /static/ {
        alias /app/staticfiles/;
    }

}

We will make a minor update to the docker/local/Dockerfile as well, just to clean up a bit:

FROM python:3.10

ENV HOME=/
ENV APP_HOME=/app

# Keeps Python from generating .pyc files in the container
ENV PYTHONDONTWRITEBYTECODE=1
# Turns off buffering for easier container logging
ENV PYTHONUNBUFFERED=1
ENV DEBUG 0

RUN apt-get update && apt-get upgrade -y && apt-get install -y \
  && apt-get clean

RUN pip install -U pip

# Install pip requirements
COPY . $APP_HOME/
RUN python -m pip install -r $APP_HOME/requirements.txt

WORKDIR $APP_HOME/

CMD ["gunicorn", "--bind", "0.0.0.0:8000", "conf.wsgi:application"]

Now, do a docker-compose up --build and your website should show including the static assets.

Local setup conclusions

We now have a fully working Django setup running a production like infrastructure.

Github and github actions

Github repo

Lets make our folder into a git repository and make a commit.

First, we create a .gitignore file in our root folder to ensure we don't commit stuff we don't want to:

venv/
staticfiles/

*.pyc
__pycache__/
db.sqlite3

now do:

$ git init
$ git status
On branch master

No commits yet

Untracked files:
  (use "git add <file>..." to include in what will be committed)
    .gitignore
    conf/
    docker-compose.yml
    docker/
    manage.py
    requirements.txt

nothing added to commit but untracked files present (use "git add" to track)

All looks well, so lets add and commit

$ git add .
$ git commit -m "Initial commit with working local docker setup"
[master (root-commit) 0989c45] Initial commit with working local docker setup
 12 files changed, 308 insertions(+)
 create mode 100644 .gitignore
 create mode 100644 conf/__init__.py
 create mode 100644 conf/asgi.py
 create mode 100644 conf/settings.py
 create mode 100644 conf/urls.py
 create mode 100644 conf/wsgi.py
 create mode 100644 docker-compose.yml
 create mode 100644 docker/local/Dockerfile
 create mode 100644 docker/local/nginx/Dockerfile
 create mode 100644 docker/local/nginx/nginx.conf
 create mode 100755 manage.py
 create mode 100644 requirements.txt

Make sure you have created an empty Github repository, then add this repository and push:

$ git remote add origin git@github.com:perwagner/django_on_a_dime.git
$ git push -u origin master
Enumerating objects: 18, done.
Counting objects: 100% (18/18), done.
Delta compression using up to 8 threads
Compressing objects: 100% (16/16), done.
Writing objects: 100% (18/18), 4.24 KiB | 4.24 MiB/s, done.
Total 18 (delta 1), reused 0 (delta 0), pack-reused 0
remote: Resolving deltas: 100% (1/1), done.
To github.com:perwagner/django_on_a_dime.git
 * [new branch]      master -> master
Branch 'master' set up to track remote branch 'master' from 'origin'.

Github action and tests

Alt Text

Hosting Provider

Vultr

create new user on the vps:

$ sudo adduser username

$ sudo usermod -aG sudo username

on you local do this to setup ssh key:

ssh-copy-id username@123.321.123.32

If you get tired of remembering the IP address, you can add this to your host file (/etc/hosts):

$ sudo nano /etc/hosts

123.321.123.32     dokku

and now you can login to your server with:

$ ssh username@dokku

Dokku on VPS

Log in to the server and do a:

$ sudo apt-get update
$ sudo apt-get upgrade

$ wget https://raw.githubusercontent.com/dokku/dokku/v0.28.1/bootstrap.sh
$ sudo DOKKU_TAG=v0.30.1 bash bootstrap.sh

Next, we add our public key from the .ssh folder

$ echo "your-public-key-contents-here" | sudo dokku ssh-keys:add admin
$ dokku domains:set-global 123.321.123.32

Now, login to the dokku server and create an app:

$ dokku apps:create myapp

and lets also allow the firewall to let through http trafic:

$ sudo ufw allow http

In your local repository, you can now add your server as a remote:

$ git remote add production dokku@123.321.123.32:myapp
$ git push production

"production" is the name of the remote. It is important that the username is "dokku", otherwise it won't work.

Set the domain for your app so it will serve on port 80 by default:

$ dokku domains:set myapp 123.321.123.32

References

Github repo for django on a dime