Skip to content

Docker Image

The lume-services repository is packaged with a Dockerfile for composing the image used to execute individual workflows. The file was modeled after that packaged with Prefect with the addition of conda for environment resolution and already-installed lume-services.

When a job is submitted to the scheduling service, the agent creates a container using this image and executes a workflow using a module path (all flows are required to use module storage).

Notes:

This project might be helpful for future development: https://github.com/conda-forge/conda-docker-feedstock