title	FastAPI
description	A FastAPI server
tags	fastapi uvicorn python postgresql

FastAPI Fullstack Template

A production-ready FastAPI template with PostgreSQL, fully async, and modern Python tooling.

CHANGELOG — See what's changed between versions.

Features

• FastAPI 0.139+ — JSON serialized natively by pydantic-core (Rust)
• Fully async — endpoints, CRUD, and database layer
• SQLAlchemy 2.0 with async engine and select() style queries
• UUIDv7 primary keys — time-ordered for B-tree locality (native in PostgreSQL 18)
• Alembic async migrations included
• Pydantic v2 for request/response validation and settings management
• Pure ASGI middleware for doc protection (no BaseHTTPMiddleware overhead)
• Typed lifespan state with proper startup/shutdown lifecycle
• uvloop + httptools for maximum ASGI performance
• GZIP compression for responses > 1KB
• Session-based auth protecting /docs and /redoc
• /health endpoint for platform healthchecks
• uv for fast, reproducible dependency management
• Locust load testing included
• Memory-tuned Docker image — multi-stage build, ~62MB idle (Railway bills by memory)

Project Structure

├── src/
│ ├── backend/
│ │ ├── main.py # FastAPI app entry point
│ │ ├── api/v1/endpoints/ # Route handlers
│ │ ├── core/ # Config, lifespan, middleware, routers
│ │ ├── crud/ # Service layer (async CRUD)
│ │ ├── dependencies/ # Database setup, DI
│ │ ├── models/ # SQLAlchemy ORM models
│ │ ├── schemas/ # Pydantic schemas
│ │ ├── security/ # Authentication
│ │ └── data/ # Seed data
│ └── frontend/
│ └── login/ # Login page templates & static files
├── migrations/ # Alembic async migrations
├── tests/ # Async test suite
├── alembic.ini # Alembic config (URL comes from app settings)
├── pyproject.toml # Dependencies & project config
├── Dockerfile # Multi-stage build with uv
├── locustfile.py # Load testing
└── .env.example # Environment variable template

Quick Start

Prerequisites

• Python 3.12+
• uv package manager

Setup

# Clone the repository
git clone https://github.com/yuting1214/Fullstack-FastAPI.git
cd Fullstack-FastAPI
# Install dependencies
uv sync
# Copy environment template
cp .env.example .env
# Edit .env with your values# Run in development mode (SQLite, auto-reload)
uv run python -m src.backend.main
# Run in production mode (PostgreSQL)
ENV_MODE=prod HOST=0.0.0.0 uv run python -m src.backend.main

Database Migrations

Tables are auto-created on startup, so the one-click deploy needs no manual step. For controlled schema evolution, use Alembic (the URL is resolved from the same ENV_MODE / DATABASE_URL settings the app uses):

uv run alembic upgrade head # apply migrations
uv run alembic revision --autogenerate -m "change"# generate a new migration

Running Tests

uv sync --extra dev
uv run pytest

Note: Tests use httpx.AsyncClient via ASGITransport, which does not trigger FastAPI's lifespan events. The shared tests/conftest.py handles DB table creation and teardown automatically. If you add new tests, the setup_db fixture runs via autouse — no manual setup needed.

Load Testing

uv run locust
# Open http://localhost:8089

Docker

docker build -t fullstack-fastapi .
docker run -p 5000:5000 fullstack-fastapi

API Endpoints

Method	Path	Description
GET	/	Onboarding message
GET	/health	Liveness probe (healthchecks)
GET	/login	Login page
POST	/login	Authenticate
GET	/logout	Clear session
POST	/api/v1/messages/	Create message
GET	/api/v1/messages/	List messages
GET	/api/v1/messages/{id}	Get message
PUT	/api/v1/messages/{id}	Update message
DELETE	/api/v1/messages/{id}	Delete message

Environment Variables

See .env.example for all available configuration options.

Runtime behavior is controlled by env vars (no CLI flags), so the app also runs under plain uvicorn src.backend.main:app:

Variable	Default	Purpose
ENV_MODE	dev	dev = SQLite + auto-reload, prod = PostgreSQL
HOST	127.0.0.1	Bind address (0.0.0.0 in the Docker image)
PORT	5000	Listen port (set automatically by Railway)
DATABASE_URL	—	PostgreSQL URL in prod (injected by Railway)
USER_NAME / PASSWORD	auto-generated	API docs login — prompted during Railway onboarding; if left empty, generated and printed once in the startup logs
SECRET_KEY	auto-generated	Session signing key — set a stable value to keep sessions across restarts

Memory Footprint

Railway bills by memory per minute, so the image is tuned for a low idle footprint — ~62MB idle (measured via cgroup in Docker), down from ~150MB:

• The container execs the venv Python directly — no resident uv run wrapper process
• MALLOC_ARENA_MAX=2 + MALLOC_TRIM_THRESHOLD_ cap glibc arena bloat
• Bytecode is precompiled at build time (UV_COMPILE_BYTECODE=1) for faster, leaner cold starts
• Multi-stage build ships only the venv and src/ — no build tools in the final image

Tip: if your traffic is bursty, enable Railway's App Sleeping on the service — compute cost drops to ~$0 while idle, and this template's single-process design is compatible with it.

Learn More

• FastAPI Documentation
• Uvicorn Documentation
• uv Documentation
• FastAPI Tips