Skip to content
stacks

Deploying a py-worker

Docker image

Section titled “Docker image”
FROM aurostack.dev/wesnetech/python:3.14
WORKDIR /app
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY . .
CMD ["python", "worker.py"]

.env is excluded from the image. Pass everything at runtime, including HOME_PATH=/app.

Logs go to stdout (INFO and above, so docker logs shows them) and to rotating files under $HOME_PATH/logs; mount a volume at /app/logs to keep the files.

CI

Section titled “CI”

.forgejo/workflows/build-amd-push.yml builds and pushes the image on every push to main, tagged latest and <yy_mm_dd>_<random>, using the repository secrets REGISTRY_URL, REGISTRY, DK_USER and DK_ACCESS_TOKEN (see node-worker’s CI). It doesn’t check out submodules.

Production checklist

Section titled “Production checklist”
  • HOME_PATH=/app (or wherever the code lives), APP_ENV=production.
  • DB_* and REDIS_* identical to the API’s production values.
  • Models updated to match the API’s current schema.
  • Telemetry: OPENOBSERVE_ORG, OPENOBSERVE_TOKEN, a distinct OTEL_SERVICE_NAME.
  • Temporal: address, namespace, TEMPORAL_TLS_*, and the python task queue (or whichever queue the API starts Python workflows on).