Docker Tutorial

Files & notes for a series of docker tutorials for coworkers.

1. Getting Started with Docker

Install Docker Desktop
Open a terminal. Some options:

Windows: Lots of options
- Use the Docker Desktop terminal
- Or use the Windows Subsystem for Linux (WSL) with Ubuntu or another distro
- Or Windows Terminal (new default in Windows 11)
- Or VS Code terminal (probably running bash)
- Or Git Bash
- Or PowerShell
Mac: Use the Terminal app
Linux: I think you know already. By the way, is anyone using desktop Linux?

Run docker run hello-world to make sure Docker is installed and working.
- You may need to first add your career/user account to the group 'docker-users' in Local Users/Groups in Computer Management if you receive the error: 'docker client must be run with elevated privileges'
- After adding your account, you will need to click 'Apply' and log out to apply the changes

2. Running Python inside Docker

Run a Python container
```
$ docker run -it python:3.9
```
1. Try a python command such as print("Hello, world!")
  - What happens when you exit?
  - What does -it do? Could check with docker run --help or at https://docs.docker.com/reference/.
2. How about print(f"List: {",".join(["one", "two", "three"])}")?
  - What's an easy way to fix that, without changing the code?
3. Try a command that requires a library such as import pandas.
  - What happens? Any ideas how to fix it?

3. Theory

How are containers and virtual machines different? How are they similar?
What are images and containers?
What other container systems exist?

4. Dockerfiles

Dockerfiles let you customize images systematically, so they're convenient to reuse and share.

Create a file in your current directory named Dockerfile
```
FROM python:3.12
RUN pip install pandas
```
1. What does FROM do?
  - What is the grandparent image — in other words, what is python:3.12 based on?
2. What does RUN do?
Use it to build a custom Docker image and tag "python-pandas".
```
 $ docker build . -t python-pandas
```

Run a container from the new image.

$ docker run -it python-pandas

>>> import pandas as pd
>>> print(pd.__version__)

5. More Dockerfile details

Dockerfiles can do a lot for you. For example, grabbing a local file and then using it internally — here, requirements.txt. This is an example of infrastructure as code.

For more details, see the Dockerfile reference.

FROM python:3.12
WORKDIR /work
COPY requirements.txt .
RUN pip install -r requirements.txt --no-cache-dir

6. Mounting a volume

Docker containers don't have access to the host filesystem, and also they are ephemeral.

$ echo 'print("Hello, world!")' > hello.py
$ docker run -v $(pwd):/work -w /work python-pandas python hello.py

What does -v $(pwd):/work do?
What does -w /work do? What happens if you don't use it?
What does the python hello.py part do?

7. Docker images as software packaging

Open source software is often available as a Docker image, which makes it easy to install and run. For example:

Example: Jupyter Lab

Jupyter Lab (formerly "Jupyter Notebooks") is a popular tool for data science and machine learning. It's used by Purdue researchers to share code and reproduce results.

docker run -p 8888:8888 quay.io/jupyter/datascience-notebook:latest
Find the login URL in the output and open it in your browser. Should see the notebook interface.

Open a new notebook and draw a graph:

import matplotlib.pyplot as plt
plt.plot([0, 1, 2, 3, 5], [0, 1, 4, 9, 25]).show()

A cooler plot.

from matplotlib import pyplot as plt
from sklearn.datasets import make_checkerboard

n_clusters = (4, 3)
data, rows, columns = make_checkerboard(
    shape=(300, 300), n_clusters=n_clusters, noise=10, shuffle=False, random_state=42
)

plt.matshow(data, cmap=plt.cm.Blues)
plt.title("Original dataset")
_ = plt.show()

Juggling multiple containers with Docker Compose

See this repository for an example using message queues: mq-experiments.

wbbaker/docker-tutorial

About

Resources

Stars

Watchers

Forks

Releases

Contributors 2