Run Your Own Container

The custom-container-app notebook registers a container app in two dapi calls and runs a published container image on a Stampede3 compute node.

What the notebook does¶

1. Register. ds.apps.new("my-container", template="container") writes the app files and ds.apps.deploy() registers the app under your account. The image and command are job parameters, so this one app runs every container you build.

2. Run. A job names CONTAINER_IMAGE=docker://ghcr.io/kks32/python-container-s3:latest, an image GitHub Actions builds and publishes from its repository, and the compute node pulls and runs it with the input directory bind-mounted at /data.

3. Results. The containerized analysis computes a soil layer’s amplification spectrum, checks its resonant peak against f0 = Vs/4H, and the archived report comes back through get_output_content.

Swap in your own image by pushing a repository with a Dockerfile; Custom Containers covers building and publishing, and Apps documents the app files the template writes.