SaaS Dev8 min read · 20 January 2026
Deploying FastAPI to Railway: The Production Checklist
Every mistake I made deploying ContentForge AI's backend to Railway — and the exact configuration, environment variables, health checks, and monitoring setup that finally made it rock-solid.
FastAPIRailwayDevOpsPythonProduction
Railway Is Great — Until It Isn't
Railway makes deployment easy. Push to GitHub, it builds and deploys. But "deployed" is not the same as "production-ready." Here's what I learned the hard way.
The Procfile
Always use a Procfile to define your start command:
web: uvicorn main:app --host 0.0.0.0 --port $PORT --workers 2
Note $PORT — Railway injects this. Never hardcode 8000.
Health Check Endpoint
Railway needs to know your app is alive:
python
@app.get("/health")
async def health():
return {
"status": "healthy",
"timestamp": datetime.utcnow().isoformat(),
"version": os.getenv("APP_VERSION", "1.0.0")
}
Set this in Railway Settings → Health Check Path → /health
Database Connections
Never create a new DB connection per request. Use a connection pool:
python
from sqlalchemy.ext.asyncio import create_async_engine, AsyncSession
from sqlalchemy.orm import sessionmaker
engine = create_async_engine(
DATABASE_URL,
pool_size=5, # Max 5 connections (Railway free = 10 limit)
max_overflow=10,
pool_pre_ping=True, # Test connection before use
pool_recycle=3600, # Recycle after 1 hour
)
AsyncSessionLocal = sessionmaker(engine, class_=AsyncSession)
async def get_db():
async with AsyncSessionLocal() as session:
yield session
Environment Variables Checklist
bash
DATABASE_URL=postgresql+asyncpg://... # Use asyncpg for async
SECRET_KEY=<64-char random string>
ENVIRONMENT=production
ALLOWED_ORIGINS=https://yourfrontend.vercel.app
REDIS_URL=redis://... # For rate limiting/caching
SENTRY_DSN=https://... # Error tracking
Rate Limiting with Upstash Redis
python
from upstash_ratelimit import Ratelimit, FixedWindow
from upstash_redis import Redis
ratelimit = Ratelimit(
redis=Redis.from_env(),
limiter=FixedWindow(max_requests=20, window=60),
)
@app.middleware("http")
async def rate_limit_middleware(request: Request, call_next):
identifier = request.client.host
result = ratelimit.limit(identifier)
if not result.allowed:
return JSONResponse(
status_code=429,
content={"error": "Rate limit exceeded"}
)
return await call_next(request)
Sentry Error Tracking
python
import sentry_sdk
from sentry_sdk.integrations.fastapi import FastApiIntegration
sentry_sdk.init(
dsn=os.getenv("SENTRY_DSN"),
integrations=[FastApiIntegration()],
traces_sample_rate=0.1, # 10% of requests
environment=os.getenv("ENVIRONMENT", "development"),
)
With these in place, ContentForge AI's Railway backend has maintained 99.7% uptime over 3 months.
