A simple configuration for using Flask and Docker Compose, which ticks the following boxes:
- Launch in production mode, with source code baked into the image.
- Launch in development mode with auto reloader enabled, and source code mounted via bind mount.
- Load (secret) environment variables from
.env-prodor.env-devfile respectively.
Usage
Pre-reqs
You'll need to create the .env-prod and .env-dev files manually:
# .env-prod
FROM_ENV_FILE='Hey this loaded from prod env!'
and
# .env-dev
FROM_ENV_FILE='Hey this loaded from dev env!'
The .gitignore file contains a line which prevents these from being committed to the repo.
Prod
docker-compose.yml contains the production config and can be run with:
docker-compose up
Dev
docker-compose-dev.yml contains the development config, in an overide file This can be run with:
docker-compose -f docker-compose.yml -f docker-compose-dev.yml up
On Linux you could make an alias with:
alias docker-compose-dev='docker-compose -f docker-compose.yml -f docker-compose-dev.yml'
Now launching simply becomes:
docker-compose-dev up
Background
Prod
Some significant parts of the docker-compose.yml (prod) file are:
environment:
- FLASK_ENV=production
env_file:
- .env-prod
command: gunicorn --bind '0.0.0.0:5000' app:app
In prod, this launches the app with gunicorn, There is no volume mount in this file, so the source code which is run is that which is baked into the image. Therefor to take account of the latest changes to your code, prior to launching in prod with docker-compose up you'll need to do docker-compose build
Dev
Some significant parts of the docker-compose-dev.yml (dev) file are:
environment:
- FLASK_ENV=development
env_file:
- .env-dev
command: flask run -h 0.0.0.0
volumes:
- ./code:/code
These values overide those specified in the prod compose file, so secrets are loaded from the .env-dev file instead, and the server runs in dev mode which enables the live reloader. Because of the volume mount, you can edit the code, and the server should reload when changes happen.