SLC Python Meetup (July 2024): How to Containerize a Python Application

July 10, 2024

Presenter: David Gardner

Overview

This talk will explain what Docker is, why developers would want to containerize their Python application, and then demonstrate how to dockerize two simple Python applications:

• a terminal-based random movie picker
• a FastAPI server that returns json

Talk Slides

https://docs.google.com/presentation/d/e/2PACX-1vT2ceINVuEUgAbog6BCAE1kdi9cnpsLTd2xmNHIVYQaG8J5L4iTVFc9jSRatMZggfbbAO-xsYlRrmUB/pub?start=false&loop=true&delayms=5000

Demo

Pre-requisites

1. Install Docker Desktop (if installed)
2. Install Python 3 (if not installed)
3. clone this repo to your machine

Demo1: Containerize a random movie picker python app

1. cd into example1 directory
2. set up a Python env running Python 3.9: python3.9 -m venv .venv (if you don't have python 3.9 already installed on your machine, you'll need to install it using your preferred package manager. I use homebrew, so I ran: brew install [email protected])
3. activate the new virtual env: . .venv/bin/activate
4. confirm the Python version is 3.9: python --version
5. install the required packages into your virtual env: pip install requests bs4
6. play around with the "movie picker app" to get familiar with how it works: python main.py
7. containerize this movie picker app using: docker build -t random-movie-picker . (-t parameter gives the container image a tag of "random-movie-picker", making it easier to identify)
8. run the newly containerized app: docker run -i -t random-movie-picker ( -i parameter ensures the running container interactive while the -t parameter creates a pseudo-terminal connection to the Docker container)
9. Done! The movie picker app is now running inside a Docker container.
10. clean up by hitting command-C to stop the container and typing deactivate to end the virtual env python session

Demo2: Containerize a FastAPI python server

1. cd into the example2 directory
2. set up a Python env running Python 3.12: python3.12 -m venv .venv (if you don't have python 3.9 already installed on your machine, you'll need to install it using your preferred package manager. I use homebrew, so I ran: brew install [email protected])
3. activate the new virtual env: . .venv/bin/activate
4. confirm the Python version is 3.12: python --version
5. install the fastAPI and uvicorn packages into your virtual env: pip install fastapi uvicorn
6. in the root directory, start the uvicorn server to understand how it works: uvicorn app.main:app. Open the link where Uvicorn is running: 127.0.0.1:8000 and check the HTTP responses from /, /docs, and /items/foobar
7. run pip freeze --local > requirements.txt to generate the required packages file
8. containerize the api service: docker build -t fastapi-demo .
9. run the containerized api service: docker run -p 8000:8000 fastapi-demo (-p 8000:8000 tells docker to expose the local port 8000 to the container's port 8000)
10. open 0.0.0.0:8000/ on your browser (localhost:8000/ should work too) and interact with the service. Sending the localhost:8000/items/foobar should return {"item": "foobar"}
11. Done! The api service is running inside a Docker container. You can view the running containers on the "Docker Desktop > Containers" page
12. clean up by hitting command-C to stop the container and typing deactivate to end the virtual env python session

Notes

I didn't cover the .dockerignore file, but basically it works just like a .gitignore file, except it tells Docker what to ignore when copying files from the local "source" directory (.) into the docker image "working" directory.