Deployment and Observability

The VoiceEngine backend server is written in Node.js 20+ and utilizes the native C++ mediasoup library to power high-concurrency, low-latency WebRTC streams.

---

Production Deployment with Docker Compose

We recommend running the voice server via Docker Compose:

version: '3.8'

services:
  voice-server:
    image: ghcr.io/cypherlink-studios/voiceengine-server:latest
    container_name: voiceengine-sfu
    restart: unless-stopped
    network_mode: "host" # Highly recommended for Linux WebRTC performance
    environment:
      NODE_ENV: production
      HTTP_PORT: 3000
      MEDIASOUP_MIN_PORT: 40000
      MEDIASOUP_MAX_PORT: 49999
      ANNOUNCED_IP: "203.0.113.10" # Public IP of your dedicated server
      PLUGIN_SECRET: "your-hmac-secret-key"
      JWT_SECRET: "your-jwt-secret-key"
      REDIS_URL: "redis://localhost:6379"
    volumes:
      - ./media:/app/media
      - ./cache:/app/cache

Caution

The UDP port range configured under MEDIASOUP_MIN_PORT and MEDIASOUP_MAX_PORT (e.g. 40000-49999/udp) must be open in your firewall (UFW / iptables) and cloud hosting security groups (AWS, Cloudflare Spectrum, Hetzner).

---

Nginx Reverse Proxy Setup

Proxy both HTTPS web assets and WebSocket handshakes to port 3000:

server {
    server_name voice.yourdomain.com;

    location / {
        proxy_pass http://127.0.0.1:3000;
        proxy_http_version 1.1;
        proxy_set_header Upgrade $http_upgrade;
        proxy_set_header Connection "upgrade";
        proxy_set_header Host $host;
        proxy_set_header X-Real-IP $remote_addr;
        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
        proxy_set_header X-Forwarded-Proto $scheme;
    }

    listen 443 ssl http2;
    ssl_certificate /etc/letsencrypt/live/voice.yourdomain.com/fullchain.pem;
    ssl_certificate_key /etc/letsencrypt/live/voice.yourdomain.com/privkey.pem;
}

---

Prometheus & Grafana Observability

VoiceEngine provides a standard Prometheus metrics endpoint at /metrics:

Exposed Metrics

Metric	Type	Description
voiceengine_active_producers	Gauge	Count of active microphone tracks currently transmitting.
voiceengine_active_consumers	Gauge	Count of active receiving audio streams on the SFU.
voiceengine_webrtc_packet_loss	Counter	Cumulative WebRTC audio packets lost over the network.
voiceengine_audio_jitter_seconds	Histogram	Statistical spread of transit time variation between audio packets.
voiceengine_media_cache_bytes	Gauge	Total storage occupied by cached audio and media tracks.

Prometheus Job Configuration (prometheus.yml)

scrape_configs:
  - job_name: 'voiceengine'
    scrape_interval: 5s
    static_configs:
      - targets: ['127.0.0.1:3000']

A turnkey Grafana dashboard is bundled in monitoring/grafana-dashboard.json providing real-time graphs for SFU bitrate, concurrent users, and voice stream health.