Observability

Metrics Endpoints

• Data plane: GET /metrics on :9100 (default)
• Control API: GET /metrics on :3000
• Data plane admin snapshot: GET /admin/runtime on :9100 (requires x-nexis-internal-token)

What to Track

• connect/handshake latency
• join latency
• room message RTT
• websocket errors
• reconnect/resume success rate
• active rooms, members, and room state drift from expected plugin state

Runtime Admin Snapshot

Use the data-plane admin endpoint for live room/session introspection:

curl "http://localhost:9100/admin/runtime?include_state=1" \
  -H "x-nexis-internal-token: <internal-token>"

Query options:

• include_state=1 includes full room state payloads
• room_id=<room-id> filters snapshot to a specific room

Baseline Testing

Use included k6 scenario:

k6 run infra/load/k6-ws.js

Run it regularly after protocol or room-logic changes.

Typed Metrics Snapshot

const samples: MetricSample[] = [
  { name: 'nexis_handshake_total', value: 42 },
  { name: 'nexis_ws_errors_total', value: 0 },
];