How to deploy FastAPI with Gunicorn and Uvicorn workers in production

Gunicorn with Uvicorn Workers

Uvicorn alone is a single process. In production, Gunicorn acts as a process manager, spawning multiple Uvicorn workers to use all CPU cores and enable zero-downtime restarts.

Install and run

pip install gunicorn uvicorn[standard]

gunicorn app.main:app \
    --workers 4 \
    --worker-class uvicorn.workers.UvicornWorker \
    --bind 0.0.0.0:8000 \
    --timeout 60 \
    --keep-alive 5 \
    --access-logfile -

Worker count formula: (2 × CPU cores) + 1. For a 2-core server, use 5 workers. Too many workers increases memory usage without improving throughput.

Nginx reverse proxy (minimal config)

server {
    listen 80;
    location / {
        proxy_pass http://127.0.0.1:8000;
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
    }
}

Nginx handles SSL termination, static file serving, and connection buffering — offloading work from Python processes. Never expose Uvicorn or Gunicorn directly to the internet without a reverse proxy.

Graceful reload

kill -HUP $(cat gunicorn.pid)

HUP signal triggers a graceful reload — workers finish in-flight requests before restarting with new code. No dropped connections during deployments.