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=sfuenvironment 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
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
Alternatively, Docker can be used to run Peer Calls.
See go.mod for more information
- TypeScript (since peer-calls
See package.json for more information.
Installation & Running
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
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
||csv||Enables or disables logging for certain modules||
||string||Base URL of the application|
||string||IP to listen to||
||int||Port to listen to||
||string||Path to TLS PEM certificate. If set will enable TLS|
||string||Path to TLS PEM cert key. If set will enable TLS|
||string||Hostname of Redis server|
||int||Port of Redis server|
||string||Prefix for Redis keys. Suggestion:
||csv||List of interfaces to use for ICE candidates, uses all available when empty|
||csv||List of ICE Server URLs|
||string||Can be empty or
||string||Secret for coturn|
||string||Username for coturn|
||string||Access token for prometheus
The default ICE servers in use are:
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.
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"
/metrics URL will not be accessible without an access token set. The access token can be provided by either:
Bearer mytoken, or
- Providing the access token as a query string:
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_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
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
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:
Client-side logs can be configured via
localStorage.log=1enables logging of Redux actions and state changes
localStorage.debug=peercalls,peercalls:*enables all other client-side logging
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
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:
In Firefox, it might be useful to use
about:webrtc to debug connection issues. In Chrome, use
When experiencing connection issues, the first thing to try is to have all peers to use the same browser.
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
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
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
sudo systemctl enable coturn sudo systemctl start coturn
See Contributing section.
If you encounter a bug, please open a new issue!
If you wish to support future development of Peer Calls, you can donate here: