Deploy
Node.js and Python apps
Run Node.js and Python web apps as managed services behind the web server, with Git deploys.
Node.js and Python apps run as managed systemd services on a local port; the web server in front handles the domain, HTTPS and static files.
1. Install the runtime
In the server's web stack, Install Node.js (22, 24 or 26) or Install Python (3.12, 3.13 or 3.14).
2. Add the site
Add site with Runs on: Node.js or Python, the version and the domain. Zefily assigns a free local port to the app.
3. Deploy the code
On the site's Deploy pane, set up a Git deploy with the Node.js or Python preset:
| Node.js | Python | |
|---|---|---|
| Build commands | npm ci, npm run build --if-present |
uv sync --frozen --no-dev |
| Start command | npm start |
uv run gunicorn --bind 127.0.0.1:$PORT app:app |
The Start command runs in the live release as the site user. Your app must listen on 127.0.0.1:$PORT; the environment also carries HOST=127.0.0.1, WEB_CONCURRENCY (worker processes sized for the server's cores and memory) and NODE_ENV=production for Node.js. Python apps use uv with the managed Python version; for FastAPI or another ASGI app use for example uv run uvicorn main:app --host 127.0.0.1 --port $PORT --workers $WEB_CONCURRENCY.
The app restarts on every deploy; if it does not answer after a deploy, the site switches back to the previous release.
Environment and database
Put configuration in the site's .env; connect a database to get DATABASE_URL. The app's output is on the Logs pane under App process.
Background workers
Run queue consumers, schedulers or bots as daemons of the site's user, with their working directory in the site's current folder.
Updated October 18, 2026