It is really exciting to build our own projects and see it working on production environment, right? Some developers might think deploying a Django app to a server is a nightmare. In this post I will show how to prepare your application to have different configurations and deploy it to Heroku.
Requirements: You must install Heroku CLI and GIT.
Heroku is the PaaS that will host our project. Go to the dashboard and create your new app. Go to the dashboard and create your new app.
Once you have created you app, it is necessart to use the terminal to set up your credentials. We need the Heroku CLI to access the app settings (check the docs to install it properly).
After installing you can login:
Later on we will use a PostgreSQL. Heroku helps us to provide a pg database by running the command. It is quite easy and you can learn about that in the docs.
Hint: The command
$ heroku apps is very handy and allows us to see the apps we have.
Preparing our application
When it comes to managing local and production environments I like to use python decouple because it allows me to use env vars and, thus, to have different configurations. Plus, I use dj_database_url to make it easy to use different databases with a single configuration string.
1. At first, create two files: .env and .env-sample. The first will be used in our local environment and must be ignored by GIT. The second works as a template:
2. Open your settings.py, import the modules we will use and use the configuration bellow:
The first two lines import the module while the other just load the values using python-decouple lib. The
cast parameter is mandatory because Python will treat the .env file content as string and it is important to cast the value to the proper data type.
Pay attention to the
DATABASES variables. The first uses an Sqlite in the local environment but in Heroku we will configure it to store the Postgres configuration url.
Deploying to Heroku
We have created our app in Heroku, attached a Postgres database and configured the settings to accept different configuration strings. Now it is time to deploy our application!
Make sure your requirements.txt file contains all dependencies
It must have, at least: dj-database-url, gunicorn, psycopg2, python-decouple and pytz. This file is extremely important because Heroku looks for the requirements file to install the dependencies.
Create a Procfile in the app root directory
A Procfile (yes, without any extension) tells Heroku how to run your application. Pay attention that **
Create runtime.txt in the app root directory
The file will tell to Heroku which version of python our project uses.
Configure your production env vars
Do you remember that our settings.py looks for an .env file? Well, python-decouple respect the environment variables precedence over config files. Thus, if there is an env var set in production the config directive will not look for an .env file.
For each key in .env file we will use the Heroku CLI to set the env var in production. We need to setup four variables: SECRET_KEY, DEBUG, ALLOWED_HOSTS and DATABASE_URL. Open the terminal and execute:
- SECRET_KEY is surrounded by single quotes because we have to explicitly define it as a string to heroku;
by your app name;
- the string .herokuapp.com must be in ALLOWED_HOSTS otherwise Django will not be executed in Heroku.
How to find the DATABASE_URL? Go to your app dashboard > Resources > Heroku Postgres Database. A new tab will open and allow you to check your DATABASE CREDENTIALS.
Deploy your app
In your Heroku app dashboard will contain instructions to deploy it. By using git you can deploy it easily:
If you face any problems just check the terminal because Heroku shows expressive messages that can help you. For example, during the deploy I got the message below:
By checking that I realized I forgot to set the STATIC_ROOT variable in settings.py and install a WSGI middleware to provide the static files.
Note: Django is not meant to provide static files. Use AWS S3 to do that for production-ready projects.
That’s all folks!