prefect.yaml

# DataPusher+ Prefect deployment manifest.
#
# Operators register this deployment once during setup with
#   ckan datapusher_plus prefect-deploy
# or directly via
#   prefect deploy
# from the project root.
#
# The default targets a local ``process`` work pool — the analogue of the
# legacy single CKAN RQ worker — and uses the source code in the
# ckan-extension install. For Kubernetes / ECS / Docker push pools,
# replace the ``work_pool`` and add a ``build``/``push`` section.

name: datapusher-plus
prefect-version: ">=3.0,<4"

# Where Prefect should look for the flow entrypoint when a worker
# executes a run. ``set_working_directory`` is enough for the process
# worker; Docker/K8s deployments override this in their own pull steps.
pull:
- prefect.deployments.steps.set_working_directory:
    directory: .

deployments:
- name: datapusher-plus
  entrypoint: ckanext/datapusher_plus/jobs/prefect_flow.py:datapusher_plus_flow
  work_pool:
    name: datapusher-plus
    job_variables:
      env:
        # The CKAN environment file should be available to the worker
        # process. Operators override this in their work-pool config.
        CKAN_INI: "{{ $CKAN_INI }}"
  tags:
    - datapusher-plus
    - production
  description: |
    DataPusher+ ingestion flow. Downloads a CKAN resource, validates
    and analyses it with qsv, loads it into the Postgres datastore,
    and runs Jinja2 formulas. See the DataPusher+ README for the full
    stage descriptions and the resilience contract.