Workflow execution backend for pypsa-app, but can be used independently. Dispatches Snakemake workflows to compute targets via a FastAPI microservice.
Clones workflow repos onto compute targets, launches Snakemake inside pixi environments, monitors execution by polling, and streams logs back via SSE.
Supported compute targets:
- SLURM over SSH: connects to an HPC head node via ssh, runs jobs as SLURM submissions
- Local: runs workflows on the same machine
cp compose.example.yml compose.yml
# Adjust settings and remove unneeded compute targets
docker compose up -dThe container image is published at ghcr.io/pypsa/snakedispatch:latest.
All configuration lives in a single YAML file (default path: ./config.yaml).
The file must contain exactly one compute target key (slurm_ssh or local) with its settings. App-level settings can optionally be added at the top level. See config/ for example configurations.
With the dev server running, interactive API docs are available at /docs (Swagger UI) and /redoc.
