Self-Hosting the Relay

The relay is a lightweight FastAPI application that you can run anywhere.

Requirements

• Python 3.11+
• uv (recommended) or pip

Running locally

cd relay
uv run python -m mykube_relay

The relay starts on http://0.0.0.0:8000 by default.

Configuration

All configuration is via environment variables:

Variable	Default	Description
SESSION_TTL_WAITING	120	Seconds before an unpaired session expires
SESSION_TTL_PAIRED	3600	Seconds before a paired session expires
MAX_PAIR_ATTEMPTS	5	Max wrong pairing attempts before session is destroyed
HOST	0.0.0.0	Listen address
PORT	8000	Listen port

Docker

FROM python:3.11-slim

WORKDIR /app
COPY relay/ .

RUN pip install uv && uv sync

EXPOSE 8000
CMD ["uv", "run", "python", "-m", "mykube_relay"]

docker build -t mykube-relay .
docker run -p 8000:8000 mykube-relay

Using your relay

Point both server and client to your relay URL:

mykube server --relay-url https://relay.example.com
mykube client --relay-url https://relay.example.com

Deployment considerations

• TLS: Put the relay behind a reverse proxy (nginx, Caddy, etc.) with TLS. WebSocket connections require wss:// in production.
• Memory: Sessions are stored in-memory. The relay uses minimal resources but will lose all sessions on restart.
• Scaling: The relay is single-instance by design (in-memory session store). For high availability, consider sticky sessions or a shared session backend.
• Monitoring: The relay logs session creation, pairing, and WebSocket connections to stdout.