Quickstart

Run the SFU

The image ships as a single static binary — signaling, forwarding, the collector and the dashboard all start together.

docker run -p 3000:3000 \
  ghcr.io/lumyx-dev/sfu-core:latest

# signaling  ws://127.0.0.1:3000/ws
# dashboard  http://127.0.0.1:3000

No Redis, no separate signaling service, no sidecar to scrape. Building from source instead is covered on Self-hosting.

Connect a client

A client opens a WebSocket to the signaling endpoint and exchanges offer, answer and ICE candidates as JSON. The message names and payloads are specified in the repository README.

const room = await connect("ws://127.0.0.1:3000/ws", token);

room.on("quality", (peer, m) => {
  if (m.packet_loss_ratio > 0.02) {
    room.setSubscription(peer.id, { video: false });
  }
});

Every metric the collector reads off RTCP reaches the client on the same quality event — nothing extra to wire up.

Watch the metrics arrive

Open http://127.0.0.1:3000 once a peer has connected. The dashboard shows the room, its peers, and all six metrics updating live — packet loss, RTT, jitter, NACK ratio, freeze ratio and bitrate — each already armed with a default threshold and a 30s debounce.

lumyx 0.4.1 · MIT
listening on :3000 · dashboard on :3000
metrics collector armed · 6 metrics · 500ms
0 rooms · 0 peers

Next steps

• Metrics reference — what each threshold breach means and how to respond.
• Self-hosting — run it in production, with a persistent config.