merlinran/peer-calls

Group peer to peer video calls for everyone written in Go and TypeScript

Peer Calls v4

WebRTC peer to peer calls for everyone. See it live in action at peercalls.com.

The server has been completely rewriten in Go and all the original functionality works. An optional implementation of a Selective Forwarding Unit (SFU) is available to make Peer Calls consume less bandwith for user video uploads. This wouldn't haven been possible without the awesome pion/webrtc library.

The config file format is still YAML, but is different than what was in v3. The v3 source code is available in version-3 branch. Version 4 will no longer be published on NPM since the server is no longer written in NodeJS.

What's New in v4

• Core rewritten in Golang.
• Selective Forwarding Unit. Can be enabled using NETWORK_TYPE=sfu environment variable. The peercalls.com instance has this enabled.
• Ability to change video and audio devices without reconnecting.
• Improved toolbar layout. Can be toggled by clicking or tapping.
• Multiple videos are now shown in a full-size grid and each can be minimized.
• Video cropping can be turned off.
• Improved file sending. Users are now able to send files larger than 64 or 256 KB (depends on the browser).
• Device names are correctly populated in the dropdown list.
• Improved desktop sharing.
• Copy invite link to clipboard. Will show as share icon on devices that support it.
• Fix: Toolbar icons render correctly on iOS 12 devices.
• Fix: Video autoplays.
• Fix: Toolbar is no longer visible until call is joined
• Fix: Add warning when using an unsupported browser
• Fix: Add warning when JavaScript is disabled

TODO for Selective Forwarding Unit

• Support dynamic adding and removing of streams
• Support RTCP packet Picture Loss Indicator (PLI)
• Support RTCP packet Receiver Estimated Maximum Bitrate (REMB)
• Add handling of other RTCP packets besides NACK, PLI and REMB
• Add JitterBuffer (experimental, currently without congestion control)
• Support multiple Peer Calls nodes when using SFU
• Add support for passive ICE TCP candidates
• End-to-End Encryption (E2EE) using Insertable Streams. See #142.

Requirements

• Node.js 8 or Node.js 12
• Go 1.14

Alternatively, Docker can be used to run Peer Calls.

Stack

Backend

• Golang
• pion/webrtc
• github.com/go-chi/chi
• nhooyr.io/websocket

See go.mod for more information

Frontend

• React
• Redux
• TypeScript (since peer-calls v2.1.0)

See package.json for more information.

Installation & Running

Download Release

Head to Releases and download a precompiled version. Currently the binaries for the following systems are built automatically:

• linux amd64
• linux arm
• darwin (macOS) amd64
• windows amd64

Deploying onto Kubernetes

The root of this repository contains a kustomization.yaml, allowing anyone to patch the manifests found within the deploy/ directory. To deploy the manifests without applying any patches, pass the URL to kubectl:

kubectl apply -k github.com/peer-calls/peer-calls

Using Docker

Use the peercalls/peercalls image from Docker Hub:

docker run --rm -it -p 3000:3000 peercalls/peercalls:latest

Building from Source

git clone https://github.com/peer-calls/peer-calls.git
cd peer-calls
npm install

# for production
npm run build
npm run build:go:linux

# for development
npm run start

Building Docker Image

git clone https://github.com/peer-calls/peer-calls
cd peer-calls
docker build -t peer-calls .
docker run --rm -it -p 3000:3000 peer-calls

Configuration

Environment variables

Variable	Type	Description	Default
PEERCALLS_LOG	csv	Enables or disables logging for certain modules	-sdp,-ws,-nack,-rtp,-rtcp,-pion:*:trace,-pion:*:debug,-pion:*:info,*
PEERCALLS_BASE_URL	string	Base URL of the application
PEERCALLS_BIND_HOST	string	IP to listen to	0.0.0.0
PEERCALLS_BIND_PORT	int	Port to listen to	3000
PEERCALLS_TLS_CERT	string	Path to TLS PEM certificate. If set will enable TLS
PEERCALLS_TLS_KEY	string	Path to TLS PEM cert key. If set will enable TLS
PEERCALLS_STORE_TYPE	string	Can be memory or redis	memory
PEERCALLS_STORE_REDIS_HOST	string	Hostname of Redis server
PEERCALLS_STORE_REDIS_PORT	int	Port of Redis server
PEERCALLS_STORE_REDIS_PREFIX	string	Prefix for Redis keys. Suggestion: peercalls
PEERCALLS_NETWORK_TYPE	string	Can be mesh or sfu. Setting to SFU will make the server the main peer	mesh
PEERCALLS_NETWORK_SFU_INTERFACES	csv	List of interfaces to use for ICE candidates, uses all available when empty
PEERCALLS_NETWORK_SFU_JITTER_BUFFER	bool	Set to true to enable the use of Jitter Buffer	false
PEERCALLS_NETWORK_SFU_PROTOCOLS	csv	Can be udp4, udp6, tcp4 or tcp6	udp4,udp6
PEERCALLS_NETWORK_SFU_TCP_BIND_ADDR	string	ICE TCP bind address. By default listens on all interfaces.
PEERCALLS_NETWORK_SFU_TCP_LISTEN_PORT	int	ICE TCP listen port. By default uses a random port.	0
PEERCALLS_NETWORK_SFU_UDP_PORT_MIN	int	Defines ICE UDP range start to use for UDP host candidates.	0
PEERCALLS_NETWORK_SFU_UDP_PORT_MAX	int	Defines ICE UDP range end to use for UDP host candidates.	0
PEERCALLS_ICE_SERVER_URLS	csv	List of ICE Server URLs
PEERCALLS_ICE_SERVER_AUTH_TYPE	string	Can be empty or secret for coturn static-auth-secret config option.
PEERCALLS_ICE_SERVER_SECRET	string	Secret for coturn
PEERCALLS_ICE_SERVER_USERNAME	string	Username for coturn
PEERCALLS_PROMETHEUS_ACCESS_TOKEN	string	Access token for prometheus /metrics URL

The default ICE servers in use are:

• stun:stun.l.google.com:19302
• stun:global.stun.twilio.com:3478?transport=udp

Only a single ICE server can be defined via environment variables. To define more use a YAML config file. To load a config file, use the -c /path/to/config.yml command line argument.

See config/types.go for configuration types.

Example:

base_url: ''
bind_host: '0.0.0.0'
bind_port: 3005
ice_servers:
 - urls:
   - 'stun:stun.l.google.com:19302'
- urls:
  - 'stun:global.stun.twilio.com:3478?transport=udp'
#- urls:
#  - 'turn:coturn.mydomain.com'
#  auth_type: secret
#  auth_secret:
#    username: "peercalls"
#    secret: "some-static-secret"
# tls:
#   cert: test.pem
#   key: test.key
store:
  type: memory
  # type: redis
  # redis:
  #   host: localhost
  #   port: 6379
  #   prefix: peercalls
network:
  type: mesh
  # type: sfu
  # sfu:
  #   interfaces:
  #   - eth0
prometheus:
  access_token: "mytoken"

Prometheus /metrics URL will not be accessible without an access token set. The access token can be provided by either:

• Setting Authorization header to Bearer mytoken, or
• Providing the access token as a query string: /metrics?access_token=mytoken

To access the server, go to http://localhost:3000.

Accessing From Network

Most browsers will prevent access to user media devices if the application is accessed from the network (not via 127.0.0.1). If you wish to test your mobile devices, you'll have to enable TLS by setting the PEERCALLS_TLS_CERT and PEERCALLS_TLS_KEY environment variables. To generate a self-signed certificate you can use:

openssl req -nodes -x509 -newkey rsa:4096 -keyout key.pem -subj "/C=US/ST=Oregon/L=Portland/O=Company Name/OU=Org/CN=example.com" -out cert.pem -days 365

Replace example.com with your server's hostname.

Multiple Instances and Redis

Redis can be used to allow users connected to different instances to connect. The following needs to be added to config.yaml to enable Redis:

store:
  type: redis
  redis:
    host: redis-host  # redis host
    port: 6379        # redis port
    prefix: peercalls # all instances must use the same prefix

Logging

By default, Peer Calls server will log only basic information. Client-side logging is disabled by default.

Server-side logs can be configured via the PEERCALLS_LOG environment variable. Setting it to * will enable all server-side logging:

• PEERCALLS_LOG=*

Client-side logs can be configured via localStorage.DEBUG and localStorage.LOG variables:

• Setting localStorage.log=1 enables logging of Redux actions and state changes
• Setting localStorage.debug=peercalls,peercalls:* enables all other client-side logging

Development

Below are some common scripts used for development:

npm start              build all resources and start the server.
npm run build          build all client-side resources.
npm run start:server   start the server
npm run js:watch       build and watch resources
npm test               run all client-side tests.
go test ./...          run all server tests
npm run ci             run all linting, tests and build the client-side

Browser Support

Tested on Firefox and Chrome, including mobile versions. Also works on Safari and iOS since version 11. Does not work on Microsoft Edge because they do not support DataChannels yet.

For more details, see here:

• http://caniuse.com/#feat=rtcpeerconnection
• http://caniuse.com/#search=getUserMedia

In Firefox, it might be useful to use about:webrtc to debug connection issues. In Chrome, use about:webrtc-internals.

When experiencing connection issues, the first thing to try is to have all peers to use the same browser.

Epheremal UDP Ports for ICE

The UDP port range can be defined for opening epheremal ports. These ports will be used for generating UDP host ICE candidates. It is recommended to enable these UDP ports when ICE TCP is enabled, because the priority of TCP host candidates will be higher than srflx/prflx candidates, as such TCP will be used even though UDP connectivity might be possible.

ICE TCP

Peer Calls supports ICE over TCP as described in RFC6544. Currently only passive ICE candidates are supported. This means that users whose ISPs or corporate firewalls block UDP packets can use TCP to connect to the SFU. In most scenarios, this removes the need to use a TURN server, but this functionality is currently experimental and is not enabled by default.

Add the tcp4 and tcp6 to your PEERCALLS_NETWORK_SFU_PROTOCOLS to enable support for ICE TCP:

PEERCALLS_NETWORK_TYPE=sfu PEERCALLS_NETWORK_SFU_PROTOCOLS=`udp4,udp6,tcp4,tcp6` peer-calls

To test this functionality, udp4 and udp6 network types should be omitted:

PEERCALLS_NETWORK_TYPE=sfu PEERCALLS_NETWORK_SFU_PROTOCOLS=`tcp4,tcp6` peer-calls

Please note that in production the PEERCALLS_NETWORK_SFU_TCP_LISTEN_PORT should be specified and external TCP access allowed through the server firewall.

TURN Server

When a direct connection cannot be established, it might be help to use a TURN server. The peercalls.com instance is configured to use a TURN server and it can be used for testing. However, the server bandwidth there is not unlimited.

Here are the steps to install a TURN server on Ubuntu/Debian Linux:

sudo apt install coturn

Use the following configuration as a template for /etc/turnserver.conf:

lt-cred-mech
use-auth-secret
static-auth-secret=p4ssw0rd
realm=example.com
total-quota=300
cert=/etc/letsencrypt/live/rtc.example.com/fullchain.pem
pkey=/etc/letsencrypt/live/rtc.example.com/privkey.pem
log-file=/dev/stdout
no-multicast-peers
proc-user=turnserver
proc-group=turnserver

Change the p4ssw0rd, realm and paths to server certificates.

Use the following configuration for Peer Calls:

iceServers:
- urls:
  - 'turn:rtc.example.com'
  auth_type: secret
  auth_secret:
    username: 'example'
    secret: 'p4ssw0rd'

Finally, enable and start the coturn service:

sudo systemctl enable coturn
sudo systemctl start coturn

Contributing

See Contributing section.

If you encounter a bug, please open a new issue!

Support

The development of Peer Calls is sponsored by rondomoon. If you'd like enterprise on-site support or become a sponsor, please contact hello@rondomoon.com.

If you wish to support future development of Peer Calls, you can donate here:

Thank you ❤️

License

Apache 2.0