Development Containers in Visual Studio Code

PSTAT 234 (Fall 2025)

Sang-Yun Oh

University of California, Santa Barbara

Development Containers Basics

Jetstream2 virtual machine (JS2VM) instance performs all your work.
JS2VM runs Docker engine on Ubuntu Linux OS
Docker builds/runs containers.
Container has all software and libraries you need.
Development container configuration is in .devcontainer/devcontainer.json
Container definition is in .devcontainer/Containerfile (or Dockerfile)

Devcontainer on Remote Host (Jetstream2 VM)

Can you explain the diagram above?

Filesystems and Files for Development Containers

Jetstream2 VM Filesystem

/home/exouser/my-project/
├── .devcontainer         
│   ├── devcontainer.json 
│   └── Containerfile
├── .gitignore            
├── README.md             
└── ...

Container Filesystem

/home/jovyan/work/
├── .devcontainer           
│   ├── devcontainer.json   # devcontainer config
│   └── Containerfile       # container definition
├── .gitignore              # for git
├── README.md               
└── ...                     # other files

Target directory in container is set in .devcontainer/devcontainer.json.

"workspaceMount": "source=${localWorkspaceFolder},target=/home/jovyan/work,type=bind", 
"workspaceFolder": "/home/jovyan/work",

Working with Files

Assume project files are in /home/exouser/my-project/ on JS2VM and mounted to /home/jovyan/work/ in the container,

Changes to files in /home/jovyan/work/ in the container are reflected in /home/exouser/my-project/ on JS2VM, and vice versa.
File /home/exouser/my-project/README.md on JS2VM and /home/jovyan/work/README.md in the container are the same file.
You can edit files in either JS2VM or the container.
Other than this mounted directory, container filesystem is ephemeral.
JS2VM filesystem is persistent unless VM is deleted.

Container Filesystem is Ephemeral

In general, container filesystem is ephemeral. Other than files mounted from a persistent storage, changes to other files in the container are lost when the container is rebuilt or deleted.

Modifying Container Image

Dockerfile or Containerfile defines the container image.
Editing .devcontainer/Dockerfile changes container image definition.
Rebuilding the container image applies the changes to your container.
Rebuilding the container does not affect files in persistent storage.
Modify system software, python/R packages, OS configuration, etc.

References

Dockerfile reference on Turing Way (Community 2022)
Ten Simple Rules for Writing Dockerfiles for Reproducible Data Science (Nüst et al. 2020)
PSTAT documentation

Modifying Devcontainer Configuration

.devcontainer/devcontainer.json defines the container configuration.
Editing .devcontainer/devcontainer.json changes container configuration.
Rebuilding the container applies the changes to your container.
Modify how container is built, run, and configured.
VS Code extensions, settings, and other configurations.

References

Development Container Specification
Ten Simple Rules for Writing Dockerfiles for Reproducible Data Science (Nüst et al. 2020)

Container Image Examples

Container URL	`REGISTRY`	`NAMESPACE`	`IMAGE_NAME`	`TAG`
`docker.io/ubuntu:22.04`	`docker.io`	`library`	`ubuntu`	`22.04`
`docker.io/intel/oneapi:latest`	`docker.io`	`intel`	`oneapi`	`latest`
`quay.io/jupyter/r-notebook:r-4.4.2`	`quay.io`	`jupyter`	`r-notebook`	`r-4.4.2`
`nvcr.io/nvidia/pytorch:24.12-py3`	`nvcr.io`	`nvidia`	`pytorch`	`24.12-py3`

Tutorials on Development Containers Usage

Exercises

Install tmux by modifying Dockerfile and rebuild the container.
Change the FROM image to another Jupyter Docker Image

GitHub Copilot

GitHub Copilot can assist you in writing and modifying your development container configuration files. You can use it to:

Generate boilerplate code for your Containerfile or devcontainer.json.
Get suggestions for VS Code extensions to include in your development container.
Find and fix issues in your configuration files.

References

Community, The Turing Way. 2022. “The Turing Way: A Handbook for Reproducible, Ethical and Collaborative Research.” Zenodo. https://doi.org/10.5281/ZENODO.7625728.

Nüst, Daniel, Vanessa Sochat, Ben Marwick, Stephen J. Eglen, Tim Head, Tony Hirst, and Benjamin D. Evans. 2020. “Ten Simple Rules for Writing Dockerfiles for Reproducible Data Science.” Edited by Scott Markel. PLOS Computational Biology 16 (11): e1008316. https://doi.org/10.1371/journal.pcbi.1008316.