TURN Server Configuration for WebRTC 2.0

TURN (Traversal Using Relays around NAT) servers are essential when direct peer-to-peer WebRTC connections cannot be established due to restrictive network conditions (e.g., firewalls or symmetric NATs). This article explains how to configure the default TURN server in the WebRTC 2.0 signaling server, how to override it using SDK properties or registry settings, and how web clients receive this configuration automatically.

TURN Server setup

The default TURN server is defined in the signaling server's configuration file:

\signaling2.0\config\production.json

before starting the TURN image, you have to change the TURN server configuration via docker-compose.xml file 

Specify the TURN user, password, and your current TURN server public IP address.

docker-compose.yml

- USERNAME=Your_user
- PASSWORD=Your_password
- REMOTE_IP=Your_current_server_public_ip

Signaling server TURN configuration

Go to your signaling server Docker image and specify your TURN server in the signaling server configuration. Edit the signaling2.0/signaling2.0/config/production.json. You need to find the "turn_server": and specify the like 

"turn_server" : 
[
"user1@password1:your_turn_server1:3478",
"user2@password2:your_turn_server2:3478",
"user3@password2:your_turn_server3:3478"
]

For example:

signaling2.0/signaling2.0/config/production.json

{
  "isDev": false,
  "server": {
    "port": 8888,
    "secure": false,
    "key": "config/sslcerts/key.pem",
    "cert": "config/sslcerts/cert.pem",
    "password": null,
    "pingTimeout": 10000,
    "pingInterval": 4800,
    "loglevel": 1
  },
  "metadata": {
    "turn_server": "turnUser@turnPassword:turnservername:3478",
    "special_rooms": {
      "gateways": { "name": "VTGateways", "role": "gateway" },
      "auth": { "name": "Auth", "role": "auth" }
    }
  },
  "auth": {
    "vt_auth_user": "default",
    "vt_auth_pass": "default"
  }
}

The format should follow this pattern: username@password:ip:port

STUN Server Implicitly Included

When you specify a TURN server, you do not need to define a separate STUN server. The specified TURN server will automatically serve as both a STUN and TURN server, as this is standard behavior for TURN infrastructure.

Overriding the TURN Server via SDK

If a TURN server is defined through IMFProps or registry settings, this configuration will override the value set in the signaling serverproduction.json

TURN Server Propagation to Web Clients

When a TURN server is configured in the signalling server's production.json, it is automatically sent to all web clients during the signalling handshake. Web-based applications that connect via the public API endpoint will receive the TURN configuration from the signalling server and apply it as part of their ICE candidate gathering process.

This behaviour ensures that all TURN/STUN settings are consistent and do not require hardcoding in your web client code.

TURN Server Docker Image

Medialooks provides a Docker-based TURN server image for your convenience. 

Turn server WebRTC 2.0 docker image download

Build and run the TURN docker image

docker compose build docker compose up -d 

Once deployed, you can reference this TURN server in your production.json as described earlier.