Advanced Containers

In the first part, we pulled and ran existing container images from Docker Hub. In this section, we will build an image from scratch for running some of our own Python3 code. Then, we will push that image back up to Docker Hub so others may find and use it. After going through this module, students should be able to:

Install and test code in a container interactively
Write a Dockerfile from scratch
Build a Docker image from a Dockerfile
Push a Docker image to Docker Hub
Volume mount external data inside a Docker container
Design Principles: Containers contribute to the portability of software projects

Getting Set Up

Scenario: You are a developer who has written some code for reading and parsing meteorite landing data in JSON format. You now want to distribute that code for others to use in what you know to be a stable production environment (including OS and dependency versions). End users may want to use this application on their local workstations, in the cloud, or on an HPC cluster.

The first step in a typical container development workflow entails installing and testing an application interactively within a running Docker container.

Note

We recommend doing this on your VMs. But, one of the most important features of Docker is that it is platform agnostic. These steps could be done anywhere Docker is installed.

To begin, make a new folder for this work and prepare to gather some important files.

[coe332-vm]$ mkdir -p coe-332/docker-exercise/
[coe332-vm]$ cd coe-332/docker-exercise/
[coe332-vm]$ pwd
/home/ubuntu/coe-332/docker-exercise

Specifically, you need your ml_data_analysis.py script and the input data file called Meteorite_Landings.json. You can make copies of your own, our download sample copies from the links below. You also need a Dockerfile, and we can just make an empty one with no contents for now.

[coe332-vm]$ pwd
/home/ubuntu/coe-332/docker-exercise
[coe332-vm]$ touch Dockerfile
[coe332-vm]$ wget https://raw.githubusercontent.com/tacc/coe-332-sp26/main/docs/unit05/scripts/Meteorite_Landings.json
[coe332-vm]$ wget https://raw.githubusercontent.com/tacc/coe-332-sp26/main/docs/unit05/scripts/ml_data_analysis.py
[coe332-vm]$ wget https://raw.githubusercontent.com/tacc/coe-332-sp26/main/docs/unit05/scripts/models.py
[coe332-vm]$ ls
Dockerfile  Meteorite_Landings.json  ml_data_analysis.py

Warning

It is important to carefully consider what files and folders are in the same PATH as a Dockerfile (known as the ‘build context’). The docker build process will index and send all files and folders in the same directory as the Dockerfile to the Docker daemon, so take care not to docker build at a root level.

Containerize Code Interactively

There are several questions you must ask yourself when preparing to containerize code for the first time:

What is an appropriate base image?
What dependencies are required for my program?
What is the installation process for my program?
What environment variables may be important?

We can work through these questions by performing an interactive installation of our Python script. Our development environment (e.g. the Jetstream VM) is a Linux server running Ubuntu 24.04 and Python 3.14. We know our code works there, so that is how we will containerize it. Use docker run to interactively attach to a fresh Python 3.14 container.

[coe332-vm]$ docker run --rm -it -v $PWD:/code python:3.14 /bin/bash
root@7ad568453e0b:/#

Here is an explanation of the options:

docker run       # run a container
--rm             # remove the container on exit
-it              # interactively attach terminal to inside of container
-v $PWD:/code    # mount the current directory to /code
ubuntu:20.04     # image and tag from Docker Hub
/bin/bash        # shell to start inside container

The command prompt will change, signaling you are now ‘inside’ the container. And, new to this example, we are using the -v flag which mounts the contents of our current directory ($PWD) inside the container in a folder in the root directory called (/code).

Check for and Install Necessary Dependencies

The first thing we will typically do is check if our dependenices are available in the container. If they are not available, we will use a package manager to install them. To run our Meterorite Landing code, we just need Python3 and a few simple Python libraries, which can be managed through uv.

root@edd3bf9e45ab:/# which python3
/usr/local/bin/python3
root@edd3bf9e45ab:/# python3 --version
Python 3.14

Python3 is available in this container (as expected). But what about uv? Run uv in your terminal and you will see output like the following:

root@91b9c9caa283:/# uv
bash: uv: command not found

We can see that uv is not installed in this (or any) python base imgae by default. Let’s install in our container just as we did at the beginning of the course

Run the following inside of your container to install uv

root@91b9c9caa283:/# curl -LsSf https://astral.sh/uv/install.sh | sh

and make sure that we add it to our $PATH

root@91b9c9caa283:/# source $HOME/.local/bin/env

Run uv help to ensure you have correctly installed it

root@91b9c9caa283:/# uv help
An extremely fast Python package manager.

Usage: uv [OPTIONS] <COMMAND>

Commands:
auth                       Manage authentication
run                        Run a command or script
init                       Create a new project
...

Now that uv is install, let’s initialize a uv environment in our /code directory and install the libraries our ml_data_analysis.py script depends on.

root@91b9c9caa283:/# uv init ./code

cd into the ./code directory.

root@91b9c9caa283:/# cd ./code

Add the packages

root@91b9c9caa283:/code# uv add pydantic pytest

Now that we have set up the container with the same packages as our VM, Let’s run our ml_data_analysis script and see if it works:

root@91b9c9caa283:/code# uv run python ml_data_analysis.py
83857.3
Northern & Eastern
Northern & Eastern
Northern & Western
Northern & Western
...

It works!

Warning

An important question to ask is: Does the versions of Python and other dependencies match the versions you are developing with in your local environment? If not, make sure to install the correct version of Python.

Install and Test Your Code

At this time, we should make a small edit to the code that will make it a little more flexible and more amenable to running in a container. Instead of hard coding the filename Meteorite_Landings.json in the script, let’s make a slight modification so we can pass the filename on the command line. In the script, add this line near the top of ml_data_analysis.py:

import sys

And change the with open... statements to these, as appropriate:

with open(sys.argv[1], 'r') as f:
    ml_data = json.load(f)

Note

You may need to install your favorite text editor with something like:

root@edd3bf9e45ab:/# apt-get update
root@edd3bf9e45ab:/# apt-get install vim

We now have functional versions of our script ‘installed’ in this container. Now would be a good time to execute the history command to see a record of the build process. When you are ready, type exit to exit the container and we can start writing these build steps into a Dockerfile.

Assemble a Dockerfile

After going through the build process interactively, we can translate our build steps into a Dockerfile using the directives described below. Open up your copy of Dockerfile with a text editor and enter the following:

The FROM Instruction

We can use the FROM instruction to start our new image from a known base image. This should be the first line of our Dockerfile. In our scenario, we want to match our development environment with Python3.14. We know our code works in that environment, so that is how we will containerize it for others to use:

FROM python:3.14

Base images typically take the form os:version. Avoid using the ‘latest’ version; it is hard to track where it came from and the identity of ‘latest’ can change.

Tip

Browse Docker Hub to discover other potentially useful base images. Keep an eye out for the ‘Official Image’ badge.

The RUN Instruction

We can install updates, install new software, or download code to our image by running commands with the RUN instruction. In our case, our dependencies were Python3 (built into our base image), uv, pydantic, and pytest (not using this yet). So, we will use a RUN instruction to install anything that is missing. The containerized version of uv is a little different that our local installation. Document can be found here

# Download the latest installer
ADD https://astral.sh/uv/install.sh /uv-installer.sh

# Run the installer then remove it
RUN sh /uv-installer.sh && rm /uv-installer.sh

# Ensure the installed binary is on the `PATH`
ENV PATH="/root/.local/bin/:$PATH"

# Initialize a uv project
RUN uv init /code

# NOTE: The WORKDIR instruction sets the working directory for any RUN, CMD, ENTRYPOINT, COPY and ADD instructions that follow it in the Dockerfile
WORKDIR /code

RUN uv add pydantic pytest

Each RUN instruction creates an intermediate image (called a ‘layer’). Too many layers makes the Docker image less performant, and makes building less efficient. We can minimize the number of layers by combining RUN instructions. Dependencies that are more likely to change over time (e.g. Python3 libraries) still might be better off in in their own RUN instruction in order to save time building later on.

The COPY Instruction

There are a couple different ways to get your source code inside the image. One way is to use a RUN instruction with wget to pull your code from the web. When you are developing, however, it is usually more practical to copy code in from the Docker build context using the COPY instruction. For example, we can copy our script to the root-level /code directory with the following instructions:

COPY ml_data_analysis.py /code/ml_data_analysis.py

COPY models.py /code/models.py

The ENV Instruction

Another useful instruction is the ENV instruction. This allows the image developer to set environment variables inside the container runtime. In our interactive build, we added the /code folder to the PATH. We can do this with ENV instructions as follows:

ENV PATH="/code:$PATH"

Putting It All Together

The contents of the final Dockerfile should look like:

FROM python:3.14

# Download the latest installer
ADD https://astral.sh/uv/install.sh /uv-installer.sh

# Run the installer then remove it
RUN sh /uv-installer.sh && rm /uv-installer.sh

# Ensure the installed binary is on the `PATH`
ENV PATH="/root/.local/bin/:$PATH"

# Initialize a uv project
RUN uv init /code

# NOTE: The WORKDIR instruction sets the working directory for any RUN, CMD, ENTRYPOINT, COPY and ADD instructions that follow it in the Dockerfile
WORKDIR /code

RUN uv add pydantic pytest

COPY ml_data_analysis.py /code/ml_data_analysis.py

COPY models.py /code/models.py

Build the Image

Once the Dockerfile is written and we are satisfied that we have minimized the number of layers, the next step is to build an image. Building a Docker image generally takes the form:

[coe332-vm]$ docker build -t <dockerhubusername>/<code>:<version> ./

The -t flag is used to name or ‘tag’ the image with a descriptive name and version. Optionally, you can preface the tag with your Docker Hub username. Adding that namespace allows you to push your image to a public registry and share it with others. The trailing dot ‘.’ in the line above simply indicates the location of the Dockerfile (a single ‘.’ means ‘the current directory’).

To build the image, use:

[coe332-vm]$ docker build -t username/ml_data_analysis:1.0 ./

Note

Don’t forget to replace ‘username’ with your Docker Hub username.

Use docker images to ensure you see a copy of your image has been built. You can also use docker inspect to find out more information about the image.

[coe332-vm]$ docker images
REPOSITORY                 TAG        IMAGE ID       CREATED              SIZE
username/ml_data_analysis  1.0        2883079fad18   About a minute ago   446MB
...

[coe332-vm]$ docker inspect username/ml_data_analysis:1.0

If you need to rename your image, you can either re-tag it with docker tag, or you can remove it with docker rmi and build it again. Issue each of the commands on an empty command line to find out usage information.

Test the Image

We can test a newly-built image two ways: interactively and non-interactively. In interactive testing, we will use docker run to start a shell inside the image, just like we did when we were building it interactively. The difference this time is that we are NOT mounting the code inside with the -v flag, because the code is already in the container:

[coe332-vm]$ docker run --rm -it username/ml_data_analysis:1.0 /bin/bash

root@c5cf05edddcd:/code# uv run ml_data_analysis.py Meteorite_Landings.json
Traceback (most recent call last):
  File "/code/ml_data_analysis.py", line 96, in <module>
    main()
  File "/code/ml_data_analysis.py", line 82, in main
    with open(sys.argv[1], 'r') as f:
FileNotFoundError: [Errno 2] No such file or directory: 'Meteorite_Landings.json'

Here is an explanation of the options:

docker run      # run a container
--rm            # remove the container when we exit
-it             # interactively attach terminal to inside of container
username/...    # image and tag on local machine
/bin/bash       # shell to start inside container

Uh oh! We forgot about Meteorite_Landings.json! We get a FileNotFoundError in Python3. This is because we did not (1) copy the JSON file into the container at build time, nor did we (2) copy the JSON file into the container at run time.

We should pause at this moment to think about how we want to distribute this application. Should the data be encapsulated within? Or should we expect potential users to be bring their own data for analysis?

Let’s try again, but this time mount the data inside the container so we can access it. If we mount the current folder as, e.g., /data, then everything in the current folder will be available. In addition, if we write any new files inside the container to /data, those will be preserved and persist outside the container once it stops.

[coe332-vm]$ docker run --rm -it -v $PWD/Meteorite_Landings.json:/data/Meteorite_Landings.json username/ml_data_analysis:1.0 /bin/bash
...
### Same command as above, but easier to read:
[coe332-vm]$ docker run --rm \
                      -it \
                      -v $PWD/Meteorite_Landings.json:/data/Meteorite_Landings.json \
                      username/ml_data_analysis:1.0 \
                      /bin/bash

root@dc0d6bf1875c:/# pwd
/
root@dc0d6bf1875c:/# ls /data
Meteorite_Landings.json
root@dc0d6bf1875c:/# ls /code
ml_data_analysis.py
root@dc0d6bf1875c:/# ml_data_analysis.py /data/Meteorite_Landings.json
83857.3
Northern & Eastern
... etc

Everything looks like it works now! Next, exit the container and test the code non-interactively. Notice we are calling the container again with docker run, but instead of specifying an interactive (-it) run, we just issue the command as we want to call it on the command line. Also, notice the return of the -v flag, because we need to create a volume mount so that our data (Meteorite_Landings.json) is available inside the container.

[coe332-vm]$ docker run --rm \
                      -v $PWD/Meteorite_Landings.json:/data/Meteorite_Landings.json \
                      username/ml_data_analysis:1.0 \
                      ml_data_analysis.py /data/Meteorite_Landings.json
Northern & Eastern
... etc

Much simpler and cleaner! Our only local dependencies are the Docker runtime and some input data that we provide. Then we pull and run the image, mounting our data inside the container and executing the embedded Python3 script. Anyone with their own data could follow our same steps to replicate our work in their own environments.

Advanced Containers

Getting Set Up

Containerize Code Interactively

Check for and Install Necessary Dependencies

Install and Test Your Code

Assemble a Dockerfile

The FROM Instruction

The RUN Instruction

The COPY Instruction

The ENV Instruction

Putting It All Together

Build the Image

Test the Image

Share Your Docker Image

Additional Resources