Container Tools

Keith VanderLinden
Calvin University

Demos

• Virtualization
  • Dev Containers
  • Docker
• GitHub Codespaces

• Goal: Demo the transition from script/local (back out) to script/cloud:
  • Referring to the infrastructure deployment diagram as appropriate.
  • Re-using the important terms throughout.
  • Configuring the dev-container appropriately.
  • Using the SLO project as an example.
  • We can run the script in VSCode or the WSL2 CLI.

1. Show hello.py running locally in CLI/Venv.
   • Show where we left off on the diagram.
   • Walk through the:
     • Docker container specified .devcontainer/Dockerfile.
       • There’s no Python Venv (recall this discussion earlier).
       • I use pip here; I couldn’t get Conda-based pre-builds to work for Tensorflow.
     • Additional VSCode configuration in .devcontainer/devcontainer.json.
       • This has the Dockerfile and some VSCode extensions3
2. Show VSCode running the local dev-container.
   • Show the transition from WSL2 to Docker on the diagram.
   • Show the Docker Desktop with the data562-slo container (and others), all exited.
   • Connect in VSCode and show the container running in:
     • VSCode.
     • Docker Desktop.
   • Run hello.py.
   • Walk through the new container environment.
     • Show the Linux file structure.
     • Show the container user.
     • Show Python, running without a Venv (because containers are isolated).
   • Rebuild the dev-container (between videos).
3. Show the dev-container running in GitHub Codespaces.
   • We can run it from:
     • GitHub Codespaces using a web-based VSCode.
     • local VSCode.
   • Show the limits of the free tier for personal codespaces.

• We’ve come full circle: cloud-local-cloud, running hello-world in countless ways - show diagram - but now:
  • We’re in control of what we’re deploying on the cloud (IaaS & PaaS).
  • We’ve got a reproducible, isolated development environment.
  • We’ll ultimately build our own cloud application (SaaS).

Snapshot

See the live demo.

Link a snapshot of the final state of the demo here, with the GitHub-cloud-based VSCode editor running hello.py from the GitHub repo.