Deploy FastAPI

This guide deploys a FastAPI application with Uvicorn and connects it to a managed PostgreSQL database. For the shared deploy flow, see Deploy from GitHub.

Prepare Your App

Run with Uvicorn

Uvicorn defaults to 127.0.0.1, which is unreachable in a container, so bind 0.0.0.0 and the port you expose. Because the start command runs as a direct exec, wrap it in sh -c so $PORT expands:

sh -c "uvicorn main:app --host 0.0.0.0 --port $PORT"

Replace main:app with the path to your app object. Add uvicorn (or fastapi[standard]) to requirements.txt.

For more concurrency in production, run Uvicorn workers under Gunicorn:

sh -c "gunicorn main:app --workers 2 --worker-class uvicorn.workers.UvicornWorker --bind 0.0.0.0:$PORT"

Set the worker count to the CPU cores of your instance size.

Deploy

Create an application from your repository. Buildpacks runs pip install -r requirements.txt, or use a Dockerfile. Set the start command above, and expose the port.

Connect a Database

Create a managed PostgreSQL database and copy its connection string into a DATABASE_URL environment variable. Connections require SSL, so ensure the string ends with ?sslmode=require.

import os
from sqlalchemy import create_engine

engine = create_engine(os.environ["DATABASE_URL"])

Map environment variables to typed settings with pydantic-settings.

Run Migrations

Out Plane has no pre deploy hook. Run Alembic once from the browser terminal with alembic upgrade head, or fold it into the start command:

sh -c "alembic upgrade head && uvicorn main:app --host 0.0.0.0 --port $PORT"

If you run more than one instance, prefer running migrations once manually so they do not race.

Production Notes

• Bind 0.0.0.0 and use $PORT. Hardcoding 8000 means no traffic reaches the app.
• Avoid blocking, CPU bound work inside async routes with a single worker. It stalls the event loop. Add workers or offload the work.

Next Steps