Home

Description

The prefect-proxy is to be thought of as part of the DDP_backend application. The only reason it has been split off into a separate app in its own repo is because it uses the asynchronous Prefect API, and DDP_backend runs on Django which is still not fully async. Rather than mark a few Django endpoints as async and figuring out how sync_to_async() works, it was easier to move all that functionality into its own FastAPI app.

prefect-proxy does not use a database. It listens for HTTP requests from DDP_backend and forwards them to Prefect. It has no knowledge of the users, roles or organizations in DDP_backend.

The project structure is as follows:

• schemas.py describes the message payloads so that Pydantic can validate them
• helpers.py contains utility functions
• service.py is where all the work happens
• main.py is where the FastAPI routes are defined
• flows.py contains reusable Prefect flow definitions i.e. Python functions decorated with @flow

prefect-proxy requires certain environment variables to be set in a .env, the template for which is provided in .env.template. Any new environment variables which the application may require in future should be listed in .env.template along with default values which are not secret.