Deployment

There are popular servers written in Python that contain ASGI applications and serve HTTP. These servers stand alone when they run; you can proxy to them from your web server.

Uvicorn

Uvicorn is an ASGI server based on uvloop and httptools, with an emphasis on speed.

Running a project in Uvicorn

When Uvicorn is installed, a uvicorn command is available which runs ASGI applications. Uvicorn needs to be called with the location of a module containing an ASGI application object, followed by what the application is called (separated by a colon).

For a typical project, invoking Uvicorn would look like:

$ uvicorn your_project:app

This will start one process listening on 127.0.0.1:8000. It requires that your project be on the Python path.

In development mode, you can add --reload to cause the server to reload any time a file is changed on disk.

Running Uvicorn with Gunicorn

Gunicorn is a robust web server that implements process monitoring and automatic restarts. This can be useful when running Uvicorn in a production environment.

To install Uvicorn and Gunicorn, use the following:

$ pip install gunicorn uvicorn

Then start Gunicorn using the Uvicorn worker class like this:

$ gunicorn your_project:app -k uvicorn.workers.UvicornWorker

Hypercorn

Hypercorn is an ASGI server that supports HTTP/1, HTTP/2, and HTTP/3 with an emphasis on protocol support.

Running a project in Hypercorn

When Hypercorn is installed, a hypercorn command is available which runs ASGI applications. Hypercorn needs to be called with the location of a module containing an ASGI application object, followed by what the application is called (separated by a colon).

For a typical project, invoking Hypercorn would look like:

$ hypercorn your_project:app

This will start one process listening on 127.0.0.1:8000. It requires that your project be on the Python path.

---

If you want to deploy your application to another ASGI server, look up the documentation about how to use a ASGI application with it.