gst-plugins-rs/net/webrtc/README.md

18 KiB

webrtcsink and webrtcsrc

All-batteries included GStreamer WebRTC producer and consumer, that try their best to do The Right Thing™.

It also provides a flexible and all-purposes WebRTC signalling server (gst-webrtc-signalling-server) and a Javascript API (gstwebrtc-api) to produce and consume compatible WebRTC streams from a web browser.

Use case

The webrtcbin element in GStreamer is extremely flexible and powerful, but using it can be a difficult exercise. When all you want to do is serve a fixed set of streams to any number of consumers, webrtcsink (which wraps webrtcbin internally) can be a useful alternative.

Features

webrtcsink implements the following features:

  • Built-in signaller: when using the default signalling server, this element will perform signalling without requiring application interaction. This makes it usable directly from gst-launch.

  • Application-provided signalling: webrtcsink can be instantiated by an application with a custom signaller. That signaller must be a GObject, and must implement the Signallable interface as defined here. The default signaller can be used as an example.

    An example is also available to use as a boilerplate for implementing and using a custom signaller.

  • Sandboxed consumers: when a consumer is added, its encoder / payloader / webrtcbin elements run in a separately managed pipeline. This provides a certain level of sandboxing, as opposed to having those elements running inside the element itself.

    It is important to note that at this moment, encoding is not shared between consumers. While this is not on the roadmap at the moment, nothing in the design prevents implementing this optimization.

  • Congestion control: the element leverages transport-wide congestion control feedback messages in order to adapt the bitrate of individual consumers' video encoders to the available bandwidth.

  • Configuration: the level of user control over the element is slowly expanding, consult gst-inspect-1.0 for more information on the available properties and signals.

  • Packet loss mitigation: webrtcsink now supports sending protection packets for Forward Error Correction, modulating the amount as a function of the available bandwidth, and can honor retransmission requests. Both features can be disabled via properties.

It is important to note that full control over the individual elements used by webrtcsink is not on the roadmap, as it will act as a black box in that respect, for example webrtcsink wants to reserve control over the bitrate for congestion control.

A signal is now available however for the application to provide the initial configuration for the encoders webrtcsink instantiates.

If more granular control is required, applications should use webrtcbin directly, webrtcsink will focus on trying to just do the right thing, although it might expose more interfaces to guide and tune the heuristics it employs.

Building

Make sure to install the development packages for some codec libraries beforehand, such as libx264, libvpx and libopusenc, exact names depend on your distribution.

cargo build

Usage

Open three terminals. In the first one, run the signalling server:

cd signalling
WEBRTCSINK_SIGNALLING_SERVER_LOG=debug cargo run --bin gst-webrtc-signalling-server

In the second one, run a web browser client (can produce and consume streams):

cd gstwebrtc-api
npm install
npm start

In the third one, run a webrtcsink producer from a GStreamer pipeline:

export GST_PLUGIN_PATH=<path-to-gst-plugins-rs>/target/debug:$GST_PLUGIN_PATH
gst-launch-1.0 webrtcsink name=ws meta="meta,name=gst-stream" videotestsrc ! ws. audiotestsrc ! ws.

The webrtcsink produced stream will appear in the former web page (automatically opened at https://localhost:9090) under the name "gst-stream", if you click on it you should see a test video stream and hear a test tone.

You can also produce WebRTC streams from the web browser and consume them with a GStreamer pipeline. Click on the "Start Capture" button and copy the "Client ID" value.

Then open a new terminal and run:

export GST_PLUGIN_PATH=<path-to-gst-plugins-rs>/target/debug:$GST_PLUGIN_PATH
gst-launch-1.0 playbin uri=gstwebrtc://127.0.0.1:8443?peer-id=[Client ID]

Replacing the "peer-id" value with the previously copied "Client ID" value. You should see the playbin element opening a window and showing you the content produced by the web page.

Configuration

The webrtcsink element itself can be configured through its properties, see gst-inspect-1.0 webrtcsink for more information about that, in addition the default signaller also exposes properties for configuring it, in particular setting the signalling server address, those properties can be accessed through the gst::ChildProxy interface, for example with gst-launch:

gst-launch-1.0 webrtcsink signaller::uri="ws://127.0.0.1:8443" ..

Enable 'navigation' a.k.a user interactivity with the content

webrtcsink implements the GstNavigation interface which allows interacting with the content, for example move with your mouse, entering keys with the keyboard, etc... On top of that a WebRTCDataChannel based protocol has been implemented and can be activated with the enable-data-channel-navigation=true property allowing a client to send GstNavigation events using the WebRTC data channel.

The gstwebrtc-api and webrtcsrc implement the protocol as well and they can be used as a client to control a remote sever.

You can easily test this feature using the wpesrc element with the following pipeline that will start a server that allows you to navigate the GStreamer documentation:

gst-launch-1.0 wpesrc location=https://gstreamer.freedesktop.org/documentation/ ! queue ! webrtcsink enable-data-channel-navigation=true meta="meta,name=web-stream"

You can control it inside the video running within your web browser (at https://127.0.0.1:9090 if you followed previous steps in that readme) or with the following GSteamer pipeline as a client:

gst-launch-1.0 webrtcsrc signaller::producer-peer-id=<webrtcsink-peer-id> enable-data-channel-navigation=true ! videoconvert ! autovideosink

Sending HTTP headers

During the initial signalling server handshake, you have the option to transmit HTTP headers, which can be utilized, for instance, for authentication purposes or sticky sessions:

gst-launch-1.0 webrtcsink signaller::uri="ws://127.0.0.1:8443" signaller::headers="headers,foo=bar,cookie=\"session=1234567890; foo=bar\""

Testing congestion control

For the purpose of testing congestion in a reproducible manner, a simple tool has been used, it has been used on Linux exclusively but it is also documented as usable on MacOS too. Client web browser has to be launched on a separate machine on the LAN to test for congestion, although specific configurations may allow to run it on the same machine.

Testing procedure was:

  • identify the server machine network interface (e.g. with ifconfig on Linux)

  • identify the client machine IP address (e.g. with ifconfig on Linux)

  • start the various services as explained in the Usage section (use GST_DEBUG=webrtcsink:7 to get detailed logs about congestion control)

  • start playback in the client browser

  • Run a comcast command on the server machine, for instance:

    $HOME/go/bin/comcast --device=$SERVER_INTERFACE --target-bw 3000 --target-addr=$CLIENT_IP --target-port=1:65535 --target-proto=udp
    
  • Observe the bitrate sharply decreasing, playback should slow down briefly then catch back up

  • Remove the bandwidth limitation, and observe the bitrate eventually increasing back to a maximum:

    $HOME/go/bin/comcast --device=$SERVER_INTERFACE --stop
    

For comparison, the congestion control property can be set to "disabled" on webrtcsink, then the above procedure applied again, the expected result is for playback to simply crawl down to a halt until the bandwidth limitation is lifted:

gst-launch-1.0 webrtcsink congestion-control=disabled

Monitoring tool

An example of client/server application for monitoring per-consumer stats can be found here.

License

All the rust code in this repository is licensed under the Mozilla Public License Version 2.0.

Code in gstwebrtc-api is also licensed under the Mozilla Public License Version 2.0.

Using the AWS KVS signaller

AWS_ACCESS_KEY_ID="XXX" AWS_SECRET_ACCESS_KEY="XXX" gst-launch-1.0 videotestsrc pattern=ball ! video/x-raw, width=1280, height=720 ! videoconvert ! textoverlay text="Hello from GStreamer!" ! videoconvert ! awskvswebrtcsink name=ws signaller::channel-name="XXX"

Using the WHIP Signaller

WHIP Client

WHIP Client Signaller uses BaseWebRTCSink

Testing the whip client as the signaller can be done by setting up janus and https://github.com/meetecho/simple-whip-server/.

  • Set up a janus instance with the videoroom plugin configured to expose a room with ID 1234 (configuration in janus.plugin.videoroom.jcfg)

  • Open the <janus/share/janus/demos/videoroomtest.html> web page, click start and join the room

  • Set up the simple whip server as explained in its README

  • Navigate to http://localhost:7080/, create an endpoint named room1234 pointing to the Janus room with ID 1234

  • Finally, send a stream to the endpoint with:

gst-launch-1.0 -e uridecodebin uri=file:///home/meh/path/to/video/file ! \
  videoconvert ! video/x-raw ! queue ! \
  whipwebrtcsink name=ws signaller::whip-endpoint="http://127.0.0.1:7080/whip/endpoint/room1234"

You should see a second video displayed in the videoroomtest web page.

WHIP Server

WHIP Server Signaller uses BaseWebRTCSrc

The WHIP Server as the signaller can be tested in two ways.

Note: The initial version of whipserversrc does not check any auth or encryption. Host application using whipserversrc behind an HTTP(s) proxy to enforce the auth and encryption between the WHIP client and server

1. Using the Gstreamer element whipwebrtcsink

a. In one tab of the terminal start the WHIP server using the below command

RUST_BACKTRACE=full GST_DEBUG=webrtc*:6 GST_PLUGIN_PATH=target/x86_64-unknown-linux-gnu/debug:$GST_PLUGIN_PATH gst-launch-1.0 whipserversrc signaller::host-addr=http://127.0.0.1:8190 stun-server="stun://stun.l.google.com:19302" turn-servers="\<\"turns://user1:pass1@turn.serverone.com:7806\", \"turn://user2:pass2@turn.servertwo.com:7809\"\>" ! videoconvert ! autovideosink

b. In the second tab start the WHIP Client by sending a test video as shown in the below command

RUST_BACKTRACE=full GST_DEBUG=webrtc*:6 GST_PLUGIN_PATH=target/x86_64-unknown-linux-gnu/debug:$GST_PLUGIN_PATH gst-launch-1.0 videotestsrc ! videoconvert ! video/x-raw ! queue ! \
  whipwebrtcsink name=ws signaller::whip-endpoint="http://127.0.0.1:8190/whip/endpoint"

2. Using Meetecho's simple-whip-client

Set up the simple whip client using using the instructions present in https://github.com/meetecho/simple-whip-client#readme

a. In one tab of the terminal start the WHIP server using the below command

RUST_BACKTRACE=full GST_DEBUG=webrtc*:6 GST_PLUGIN_PATH=target/x86_64-unknown-linux-gnu/debug:$GST_PLUGIN_PATH gst-launch-1.0 whipserversrc signaller::host-addr=http://127.0.0.1:8190 stun-server="stun://stun.l.google.com:19302" turn-servers="\<\"turns://user1:pass1@turn.serverone.com:7806\", \"turn://user2:pass2@turn.servertwo.com:7809\"\>" name=ws ! videoconvert ! autovideosink ws. ! audioconvert ! autoaudiosink

b. In the second tab start the simple-whip-client as shown in the below command

./whip-client --url http://127.0.0.1:8190/whip/endpoint \
        -A "audiotestsrc is-live=true wave=red-noise ! audioconvert ! audioresample ! queue ! opusenc ! rtpopuspay pt=100 ssrc=1 ! queue ! application/x-rtp,media=audio,encoding-name=OPUS,payload=100" \
        -V "videotestsrc is-live=true pattern=ball ! videoconvert ! queue ! vp8enc deadline=1 ! rtpvp8pay pt=96 ssrc=2 ! queue ! application/x-rtp,media=video,encoding-name=VP8,payload=96" \
        -S stun://stun.l.google.com:19302 \
        -l 7 \
        -n true

Terminating the client will close the session and the client should receive 200 (OK) as the response to the DELETE request

Using the LiveKit Signaller

Testing the LiveKit signaller can be done by setting up LiveKit and creating a room.

You can connect either by given the API key and secret:

gst-launch-1.0 -e uridecodebin uri=file:///home/meh/path/to/video/file ! \
  videoconvert ! video/x-raw ! queue ! \
  livekitwebrtcsink signaller::ws-url=ws://127.0.0.1:7880 signaller::api-key=devkey signaller::secret-key=secret signaller::room-name=testroom

Or by using a separately created authentication token

gst-launch-1.0 -e uridecodebin uri=file:///home/meh/path/to/video/file ! \
  videoconvert ! video/x-raw ! queue ! \
  livekitwebrtcsink signaller::ws-url=ws://127.0.0.1:7880 signaller::auth-token=mygeneratedtoken signaller::room-name=testroom

You should see a second video displayed in the videoroomtest web page.

Streaming from LiveKit using the livekitwebrtcsrc element

First, publish a stream to the room using the following command:

gst-launch-1.0 livekitwebrtcsink name=sink \
    signaller::ws-url=ws://127.0.0.1:7880 \
    signaller::api-key=devkey \
    signaller::secret-key=secret \
    signaller::room-name=testroom \
    signaller::identity=gst-producer \
    signaller::participant-name=gst-producer \
    video-caps='video/x-h264' \
  videotestsrc is-live=1 \
  ! video/x-raw,width=640,height=360,framerate=15/1 \
  ! timeoverlay ! videoconvert ! queue ! sink.

Then play back the published stream:

gst-launch-1.0 livekitwebrtcsrc \
    name=src \
    signaller::ws-url=ws://127.0.0.1:7880 \
    signaller::api-key=devkey \
    signaller::secret-key=secret \
    signaller::room-name=testroom \
    signaller::identity=gst-consumer \
    signaller::participant-name=gst-consumer \
    signaller::producer-peer-id=gst-producer \
    video-codecs='<H264>' \
  src. ! queue ! videoconvert ! autovideosink

Auto-subscribe with livekitwebrtcsrc element

With the LiveKit source element, you can also subscribe to all the peers in your room, simply by not specifying any value for signaller::producer-peer-id. Unwanted peers can also be ignored by supplying an array of peer IDs to signaller::excluded-producer-peer-ids. Importantly, it is also necessary to add sinks for all the streams in the room that the source element has subscribed to.

First, publish a few streams using different connections:

gst-launch-1.0 \
  livekitwebrtcsink name=sinka \
    signaller::ws-url=ws://127.0.0.1:7880 \
    signaller::api-key=devkey \
    signaller::secret-key=secret \
    signaller::room-name=testroom \
    signaller::identity=gst-producer-a \
    signaller::participant-name=gst-producer-a \
    video-caps='video/x-vp8' \
  livekitwebrtcsink name=sinkb \
    signaller::ws-url=ws://127.0.0.1:7880 \
    signaller::api-key=devkey \
    signaller::secret-key=secret \
    signaller::room-name=testroom \
    signaller::identity=gst-producer-b \
    signaller::participant-name=gst-producer-b \
    video-caps='video/x-vp8' \
  livekitwebrtcsink name=sinkc \
    signaller::ws-url=ws://127.0.0.1:7880 \
    signaller::api-key=devkey \
    signaller::secret-key=secret \
    signaller::room-name=testroom \
    signaller::identity=gst-producer-c \
    signaller::participant-name=gst-producer-c \
    video-caps='video/x-vp8' \
  videotestsrc is-live=1 \
  ! video/x-raw,width=640,height=360,framerate=15/1 \
  ! timeoverlay ! videoconvert ! queue ! sinka. \
  videotestsrc pattern=ball is-live=1 \
  ! video/x-raw,width=320,height=180,framerate=15/1 \
  ! timeoverlay ! videoconvert ! queue ! sinkb.
  videotestsrc is-live=1 \
  ! video/x-raw,width=320,height=180,framerate=15/1 \
  ! timeoverlay ! videoconvert ! queue ! sinkc.

Then watch only streams A and B by excluding peer C:

gst-launch-1.0 livekitwebrtcsrc \
  name=src \
  signaller::ws-url=ws://127.0.0.1:7880 \
  signaller::api-key=devkey \
  signaller::secret-key=secret \
  signaller::room-name=testroom \
  signaller::identity=gst-consumer \
  signaller::participant-name=gst-consumer \
  signaller::excluded-producer-peer-ids='<gst-producer-c>' \
  src. ! queue ! videoconvert ! autovideosink
  src. ! queue ! videoconvert ! autovideosink