RTCTunnel builds network tunnels over WebRTC.

This is a proof of concept and should NOT be used in production. Errors/Exceptions like disconnects and invalid data are not well handled, and may result in panics.

An overview of the application and how and why it was built is available here: RTCTunnel: Building a WebRTC Proxy with Go.

RTCTunnel can be installed via go install:

go install github.com/rtctunnel/rtctunnel/cmd/rtctunnel@latest

Or downloaded from the releases page (for linux).

RTCTunnel creates a network tunnel over WebRTC between two peers. Those peers are identified by a public key.

To use RTCTunnel first create a config with:

rtctunnel init

You can see info with:

rtctunnel info

Once you've done that on both peers, copy the two public keys and add a route. A route has four components: a local peer, a local port, a remote peer and a remote port. All network connections and data sent to the local port will be forwarded to the remote peer. For example:

export CLIENT_KEY=
export SERVER_KEY=
rtctunnel add-route \
    --local-peer=$CLIENT_KEY \
    --local-port=6379 \
    --remote-peer=$SERVER_KEY \
    --remote-port=6379

The route must be added on both peers and the CLIENT_KEY and SERVER_KEY should be set to the peer keys.

Once the routes are added you can run rtctunnel with:

rtctunnel run

Typically it would be run in the background.

A docker-compose example is available in examples/redis.

Configuration is stored in a yaml file based on github.com/kirsle/configdir:

Linux:   $XDG_CONFIG_HOME or "$HOME/.config"
Windows: %APPDATA% or "C:\\Users\\%USER%\\AppData\\Roaming"
macOS:   $HOME/Library/Application Support

You can see the path by running rtctunnel help.

In addition to the key pair and routes you can set the signal channel in the config:

keypair:
  public: xxx
  private: xxx
routes: []
signalchannel: operator://localhost:8000

The default signal channel is apprtc:// which uses a websocket server for appr.tc.