Currently, collection cleanup code is only run when an added event
is received from the server. Where that fails is in scenarios where
a server-side collection turns empty while an affected users is
disconnected - and then reconnects. There's no removed (or updated)
event so no cleanup code is run and you have stale data.
This commit guarantees a stale data check is run whenever a subscription
is established again. The `added` check was also maintained, although
I'm not too sure anymore it's is still needed. That may need to be
revisited.
There are some situations where previously set deviceIds (
local/session storage) may become stale. This causes an unexpected
behavior where audio is temporarily borked until the user clears their
local storage.
This issue has been seen more recently on Safari endpoints when switching
back-and-forth breakout rooms in environments running under iframes.
Also seen randomly on endpoints with virtual input devices.
This centralizes audio gUM calling into a single method that retries the
gUM procedure without pre-set deviceIds only if the initial call fails
due with an OverconstrainedError - hopefully circumventing the issue.
BBBVideoStream 'inactive' event currently does not send the
MediaStream id as the payload for handlers. This can cause theoretical
race conditions due to media stream mismatches in places where
'inactive' handling is done globally (eg video-preview).
This adds the MediaStream id to BBBVideoStream's 'inactive' event
and uses it in video-preview to avoid such race conditions.
Extract the deviceId again from the stream to guarantee consistency
between stream DID vs chosen DID. That's necessary in scenarios where,
eg, there's no default/pre-set deviceId ('') and the browser's
default device has been altered by the user (browser default != system's
default).
There's no rollback procedure in case a device switch fails right now,
nor does the code entrypoints that call the switching procedures wait
for resolution or failure before marking the new device as chosen. That
may cause inconsistent states in a couple of ways:
- No rollback: switch fails, audio is still on but no actual
microphone input is being transmitted
- Not waiting for resolutions: inconsistent chosen devices on failures
Device switching errors are also not surfaced to the end user
This commit:
- Adds device rollback and proper resolution/failure response
awaits to try and make the state a bit more consistent.
- Centralizes the input device switching code to be reused between
different bridges
- Centralizes device ID state management in audio-manager to try and
mantain them a bit more consistent across the board
- Surface device switching failures to the end user
- Guarantee device IDs are set to the session storage on all
appropriate scenarios
RTCRTPSender exposes DSCP marking via `networkPriority` in the encodings
configuration dictionaries. That should allow us to control
QoS priorities for different media streams, eg audio with higher network
priority than video. The only browser that implements that right
now is Chromium.
To use this, the public.app.media.networkPriorities configuration in
settings.yml. Audio, camera and screenshare priorities can be controlled
separately. For further info on the possible values, see:
- https://www.w3.org/TR/webrtc-priority/
- https://datatracker.ietf.org/doc/html/rfc8837#section-5
Firefox doesn't fire the ended evt/onended callback for live
MediaStreamTrack(s). We rely on that event.
Manually emit the ended event which works with the onended callback
when a track is stopped
kurento-utils is unmaintained. It's served us well, but its age
shows. We need to transition to something else if we want to
have better maintainability and include simulcast, multistream, ...
This introduces a simplified/leaner wrapper kit that's almost
API-compatible with what we use right now - so widespread changes
are minimal). It's easier to maintain/read/transition from. This
can be read as an intermediate step to transitioning to
something definitive (ie mediasoup-client).
Under some scenarios, cameras are freezing when the virtual background
code is running due to runPostProcessing(_renderMask) throwing
NS_ERROR_FAILURE - mainly on Firefox - consequently preventing
subsequent TimerWorker ticks from being scheduled.
Cases where I've seen that happen are:
- conferences running under an iframe where the iframe is briefly
stalled for some reason
Address the issue with a try-catch and a log for debugability (it's high
frequency, hence why not error level). We should probably remove the log
entirely once we figure out why the post-processing method is failing.
Use the built-in getLocalStream from the peer wrapper instead (which
relies on getSenders - the proper, spec-compliant way).
Two different issues are addressed:
- RTCPeerConnection.getLocalStreams is a pre-1.0 WebRTC spec which is
deprecated and not recommended.
- Fixed an issue where a switch from full audio to listen only could
cause the latter to be rejected with an error 1004 in rare scenarios.
There could be a race condition where the local getDisplayMedia stream ends
(eg via Chrome`s stop sharing toast) while the broker hasn't finished starting.
That would lead to a scenario where the broker wouldn't emit an end event,
causing screen sharing to be flagged as started with a blank/invalid stream.
- Remove the old listen only bridge (kurento.js), superseded by the equivalent
and equally stable (AS FAR AS LISTEN ONLY IS CONCERNED) sfu-audio-bridge
- Rename FullAudioBridge.js -> sfu-audio-bridge.js
* A more generic name that better represents the capabilities and
the nature of the bridge
* The bridge name identifier in configuration is still the same
('fullaudio')
- Remove the FreeSWITCH listen only fallback
- Temporarily disable the "trickle ICE" pair gathering feature used
in SIP.js (which was always experimental, nonstandard and disabled
by default)
- Updates to settings.yml keys in places where relevant
"default" is not an universally valid default value for deviceIds which was causing issues with Firefox and Safari in some specific scenarios where exact deviceId constraints were being used
Seems to have been introduced by a partial merge commit
There were a bunch of style changes introduced by that partial commit as well; I kept those changes to avoid introducing further conflicts between v2.4-2.5...
- forceRelayOnFirefox: whether TURN/relay usage should be forced to work
around Firefox's lack of support for regular nomination when dealing with
ICE-litee peers (e.g.: mediasoup).
* See: https://bugzilla.mozilla.org/show_bug.cgi?id=1034964
- iOS endpoints are ignored from the trigger because _all_ iOS browsers
are either native WebKit or WKWebView based (so they shouldn't be affected)
When joining breakouts, we now wait for the bridge to be loaded before
automatically start user's audio.
This problems happens only on fullaudio bridge
This commit allows user to join/leave audio using the fullaudio bridge.
This is still under development, but to use this now we must set values of
skipCheck to false, and defaultFullAudioBridge to fullaudio. This
depends on newest version of bbb-webrtc-sfu
ICE lite servers (eg mediasoup) dont need candidates signaled out-of-band; neither does KMS in certain scenarios
Disable their signaling saves us some ticks in bbb-webrtc-sfu and some bandwidth all around
Restored the old behavior when ending breakout rooms while user is in the
breakout audio transfer, which is to the trigger the reconnection to the audio
in the main room.
This behavior could be improved by (instead of reconnecting) transfering user
back to the main room, but this requires some changes in akka-apps/fsesl
which can be treated in a different issue.
Closes#13242
