Skip to content

Palfrey

Palfrey

Palfrey is a clean-room, high-performance Python ASGI server with source-traceable parity mapping.

Test Suite Package version Supported Python versions


Documentation: https://palfrey.dymmond.com 📚

Source Code: https://github.com/dymmond/palfrey

The official supported version is always the latest released.


Palfrey is a clean-room ASGI server focused on three things:

  • behavior you can reason about
  • deployment controls you can operate safely
  • performance you can reproduce and verify

Protocol runtime modes include HTTP/1.1 backends plus opt-in HTTP/2 (--http h2) and HTTP/3 (--http h3) paths.

This documentation is written for both technical and non-technical readers.

  • Engineers can use the protocol details, option tables, and runbooks.
  • Product, support, and operations teams can use the plain-language summaries and checklists.

What Palfrey Does

At runtime, Palfrey sits between clients and your ASGI application.

  1. accepts TCP or UNIX socket connections
  2. parses protocol bytes into ASGI events
  3. calls your app with scope, receive, send
  4. writes responses back to clients
  5. manages process behavior (reload, workers, graceful shutdown)

Who Should Start Where

If you are new to ASGI

  1. Installation
  2. Quickstart
  3. Terms and Mental Models
  4. Server Behavior

If you operate production services

  1. Deployment
  2. Workers
  3. Observability
  4. Troubleshooting
  5. Release Process

First 60 Seconds

Create main.py:

from __future__ import annotations


async def app(scope, receive, send):
    """Return a plain-text greeting for HTTP requests."""
    if scope["type"] != "http":
        return

    body = b"Hello from Palfrey"
    await send(
        {
            "type": "http.response.start",
            "status": 200,
            "headers": [
                (b"content-type", b"text/plain; charset=utf-8"),
                (b"content-length", str(len(body)).encode("ascii")),
            ],
        }
    )
    await send({"type": "http.response.body", "body": body})

Run Palfrey:

palfrey main:app --host 127.0.0.1 --port 8000

Check it:

curl http://127.0.0.1:8000

Gunicorn + Palfrey worker:

gunicorn main:app -k palfrey.workers.PalfreyWorker -w 4 -b 0.0.0.0:8000

Documentation Structure

Getting Started

  • install, verify, and run your first app
  • move from a minimal app to real startup patterns

Concepts

  • what ASGI is, and how Palfrey applies it
  • how HTTP, WebSocket, and lifespan flows behave
  • how server internals affect user-visible outcomes

Reference

  • full CLI and config surface
  • protocol and logging behavior
  • env var model and common errors

Guides

  • migration, security hardening, production rollout
  • practical troubleshooting and FAQ

Operations

  • deployment shapes, workers, reload model
  • capacity planning, observability, benchmark method
  • platform-specific notes and release process

Plain-Language Summary

If your application is the business logic, Palfrey is the runtime control layer around it. A good runtime control layer gives teams:

  • predictable startup and shutdown
  • fewer surprises under traffic spikes
  • clearer incident response paths
  • safer, repeatable deployments