LiveKit Community

hmm, are you calling setMicrophoneEnabled multiple times? this log is emitted if it's called multiple times while we're attempting to publish

nope I dont call it on room connected, only on button click

does it happen on the first time you click it?

is there anyway to check the publish track? I try with `getTracks()` but return with empty

yes, even I try to call immediately after room connected, still got this issue

Can you try the sample app and see if it repros there?

<@U01NZKTNE1H> I am testing on sample project, after connect it always excute failed to connect block

are there any errors emitted in the logs?

<@U01NZKTNE1H> I think I found my issue with room instance, I forgot to add room to useEffect dependency, so it keep old instance

good to hear, did that fix all your issues?

<@U01NZKTNE1H> still face one more issue.
first open screen is okay, but when I navigate back and open again, got same error with duplicate track

I call room .disconnect but it not clear all track

Hi there!

LiveKit offers server-side moderation APIs which will allow you to manage permissions for participants in the room. This might help:

<https://docs.livekit.io/guides/server-api|https://docs.livekit.io/guides/server-api> 
<https://docs.livekit.io/guides/server-api/#participantpermission|https://docs.livekit.io/guides/server-api/#participantpermission>

hey <@U03R3QXPYQP>, you can use server APIs to <https://docs.livekit.io/guides/server-api/#updateparticipant|UpdateParticipant API> to change permission of the user that's already in the room.

then on the client side, you can listen to the <https://docs.livekit.io/client-sdk-js/enums/ParticipantEvent.html#ParticipantPermissionsChanged|PermissionChanged> event on the client side to determine if you should be publishing tracks.

If all the participant permission is set to false, the WebRTC connection will be established, but can I set the participant permission that neither voice nor data channel can be received?
<https://docs.livekit.io/guides/server-api/#participantpermission>

Also, is it possible to set Participant Permission, which does not receive voice or data channels by default at the time of participation?

sure, but do you not want participants to receive others' tracks?

Yes, I don’t want to give any permissions at first.
Participant can join the room when the owner gives permission to the participants.

After that, Participant can also hear other Participant in the room (including the owner)

sure, if they don't have subscribe permissions, then no tracks will be sent down

Really?
Can’t receive everyone’s tracks including participants before permission?
Is it just a WebRTC connection established? Can’t do anything with that alone?

we enforce permissions on a per-participant basis. <https://docs.livekit.io/guides/access-tokens/#video-grant|see this table> on the permissions you could grant to each participant. if you are not giving them any permissions, then they would not be able to publish nor subscribe tracks to the room

Is it possible to disallow all permissions on <https://docs.livekit.io/guides/access-tokens/#video-grant|this table> (ie all false)?
Also, it says Video grant,  but is it the same for audio only?

~What if roomJoin is false?~
Is it necessary to specify only roomJoin because there must be at least one?

I’m sorry that I didn’t understand :bow:

none of these permissions are mutually exclusive.. if you don't specify roomJoin, then that participant will not be able to join the room.

Hey <@U03K5G2V5RT>, would you mind clarifying for me what the “stream screen” you’re referring to is? Do you mean watermarking the actual video streams?

Hey <@U01KZ0SCDKL>, Yes, we have integrated livekit in our broadcast system and we would like to show a watermark of our logo on the actual video stream 

I understand, thanks for clarifying. LiveKit doesn’t decode any video streams on the server side, thus the easiest way is to add your logo/branding on the client side, in your application.

I see, thanks for your time <@U01KZ0SCDKL> 

There are other options, but they’re also a lot more work. For example, you could somehow hook into LiveKit’s egress system to capture decoded streams, add your watermark, then re-encode and stream back out. It would be more resource-intensive and probably introduce some latency (I’m not the expert on this), but it _is_ possible.

Hi Julius, which type of egress are you trying to start?

<@U01KZ61GV5Z>

I use `"livekit-server-sdk": "^1.0.1"` with the following command:

```startTrackCompositeEgress(
  roomName,
  output,
  audioTrackId,
  videoTrackId,
  {
    width: 1920,
    height: 1080,
    depth: 24,
    framerate: 30,
    audioCodec: AudioCodec.AAC,
    audioBitrate: 128,
    audioFrequency: 48_000,
    videoCodec: VideoCodec.H264_MAIN,
    videoBitrate: 4500,
  }
)```

<@U02EU2ZPTQU>

Can you post a small snippet how to *unpublish* all of a participant's video / audio tracks with `livekit-client` / `@livekit/react-core / @livekit/components` please ?
Default behaviour in LiveKit react components is to mute, but I'd like to remove the tracks instead.

you can do this by accessing the `localParticipant` directly (you would normally do that via the `room` instance).

```const localTracks = room.localParticipant.getTracks().reduce&lt;LocalTrack[]&gt;((acc, pub) =&gt; {
      if (pub.track) {
        acc.push(pub.track as LocalTrack);
      }
      return acc;
    }, []);
room.localParticipant.unpublishTracks(localTracks); ```
We do recommend using the mute approach though, especially if you are planning on publishing/unpublishing fairly often.

<@U02EU2ZPTQU>
If I follow your recommendation I would need an event like `unmute` and `mute` on *webhook* side.
AFAIK I can't observe these actions with webhooks.

would you want to share more about your usecase? you are right, that no mute events are available via webhooks currently, but would be interesting to know what you are trying to achieve.

<@U02EU2ZPTQU>

It's about getting a stable setup for Egress recording as long as Egress seems picky.

I use these Docker images:
• `livekit-server v1.1`
• `egress v1.0.5`
and this js library server side `"livekit-server-sdk": "^1.0.1"` with the following command:

```startTrackCompositeEgress(
  roomName,
  output,
  audioTrackId,
  videoTrackId,
  {
    width: 1920,
    height: 1080,
    depth: 24,
    framerate: 30,
    audioCodec: AudioCodec.AAC,
    audioBitrate: 128,
    audioFrequency: 48_000,
    videoCodec: VideoCodec.H264_MAIN,
    videoBitrate: 4500,
  }
)```

Issue:
If I start recording with a muted media track (audio and/or video), Egress keeps saying `EGRESS_STARTING`
but with unmuted tracks it starts recording with `EGRESS_ACTIVE`

My *workaround* would be to ensure that the participants only offer unmuted tracks.
So the moment one participant would mute the track, the webhook side stops the recording. But as long as there is no event for webhooks like this, I enforce the audio and video button in the react app to remove the tracks instead of only muting them. AFAIK this triggers events a  webhook can listen to, namely `track_published` and `track_unpublished`

for a track composite, is your goal to stop recording if both audio/video are muted? if only one of them is muted, track composite would still record the other track.

There isn't a way to indicate remote participant status. However, the client SDKs have built-in reconnection when it's interrupted. We would not send a disconnected message until the retry attempts have all failed.

Okay thanks David. Do you know how long the client SDKs attempt reconnection before giving up?

yup, we fire <https://docs.livekit.io/guides/room/events/#events|Reconnecting &amp; Reconnected> when clients are retrying internally

sequence is described <https://docs.livekit.io/guides/room/connect/#network-changes-and-reconnection|here>

hi VJ, glad you've found us! :slightly_smiling_face:

Multi-node deployment is supported currently via Kubernetes. You can also just deploy separate VMs pointing to a shared redis instance. It's possible to place a load balancer in front of the VMs and they will load balance between each other automatically.

Wow :open_mouth: , That's straight forward and yet so sophisticated!!

you could call `room.disconnect` when your component/page unmounts

which component are you using <@U03C8KQDJMA>? Perhaps our components should do this automatically. wdyt <@U02EU2ZPTQU>?

a modified version of `@livekit/react-components'` with starting point like this: <https://github.com/livekit/livekit-react/blob/v1.0.0/packages/components/src/LiveKitRoom.tsx>

Yeah, it could make sense to disconnect when the `Room` component unmounts!

this has been merged: <https://github.com/livekit/livekit-react/pull/79>

Screenshot 2022-08-02 at 16.16.26.png

looks like you’re sending an empty object - isn’t there a name required?

what does your token permissions look like? CreateRoom requires <https://docs.livekit.io/guides/access-tokens/#video-grant|roomCreate> permissions in the grant.

`livekit-cli create-token \` 
  `--api-key devkey --api-secret secret \`
  `--join --room my-first-room --identity user1 \`
  `--valid-for 24h`

it looks like this. how do i use the grant with twirp?

you've created a join token... what you'd want is
```livekit-cli create-token \ 
  --api-key devkey --api-secret secret \
  --create --valid-for 24h```

hi team! one question, I am doing a benchmark test with livekit-cli and something happened.

When I set 100 publishers I am getting this error when I try to connect with my browser:

```WebRTC: ICE failed, add a STUN server and see about:webrtc for more details```
If I reduce to 50 publishers it works fine.

Note: the livekit-cli command is executed on a separate instance, not on my machine.

Do you know what could be the cause?

image.png

Hey <@U03RJUWRMMJ>, may I ask you how you are generating the Token?

I'm using the test generated token right now

specifically here is the whole configuration set up. It seems to have gone smoothly? I'm not sure where I've gone wrong

Id be curious if you could generate another token and test with that.

So I started from the start once more, and got it all working just about 13 minutes ago. Posting positive results across the board. Not sure where I went wrong the first time around but I'm happy it's all done!

yeah, definitely! which client are you publishing with? You can set your <https://docs.livekit.io/guides/room/publish/#advanced-track-management|own publish settings here>

So specifically this is my setup; I'm on two seperate VMs on the same Oracle Cloud network. One is running Foundry Virtual Tabletop software, the other is running Livekit. We just use audio, we don't use video at our table and I'm using the Livekit server for the Foundry Virtual Tabletop to achieve this, using this module:

<https://foundryvtt.com/packages/avclient-livekit>

(In fact I know I'm one of many using this very module!)

To explain fully I'm very very new to all this sort of thing (working with Linux and Cloud servers) and I wouldn't call myself a developer by any stretch of the imagination. So I'm wondering do I need to use the Server SKL in some way because I don't think this is something I can adjust on the Client SKL side as I think that's the module?

Mainly its getting used for like long ambient tracks and other sound effects off of youtube, but right now it has that 'phone speaker' sound quality to it which doesn't really compare to my usual setup through Cleanfeed. So yea I'm thinking like 128 kbps is minimum.

So I found the Livekit CLI, thought that might be what I needed to change the default settings, but when I try to install it...

Hey Brian, I'm unsure how it works with the Foundry plugin, maybe <@U02C10ZEP8R> would know more. From a livekit perspective, setting the publishing rate is done within the client SDK. The Foundry AVClient would be the one integrating with our SDK; any changes would need to be applied there

I don't currently have audio settings on the client, but I could add that to the settings. Can you open a ticket on the module github repo so I don't forget? :slightly_smiling_face:

Absolutely, I'll open one up right now, thanks for the response!!

that might take a bit for us to implement internally; in the mean time, you can try using <https://github.com/react-native-webrtc/react-native-incall-manager>

hey carlos, would you happen to have logs during the reconnection?

a bunch of them, anything in particular you’re looking for or should I paste them all?

Hey Ricky, these steps should get you up and running on AWS: <https://docs.livekit.io/deploy/vm/>

as long as I install docker on the EC2 Instance ya <@U01KZ61GV5Z>?

the above instructions will generate a cloud init script that installs everything that's needed

CleanShot 2022-08-04 at 11.20.46@2x.png

Hello, <@U01KZ61GV5Z>.
Sorry for disturbing, I’m quite not understand what to do next after this.
Mind to help?

maybe this would help? <http://techflare.blog/beginner-tutorial-cloud-init-in-aws/>

CleanShot 2022-08-04 at 14.58.52@2x.png

I’ve tried to set the DNS to my IP Address of EC2 Instances, also set the inbound rules as mentioned here, but it still can’t accessible. Anything I need to do further <@U01KZ61GV5Z>?

CleanShot 2022-08-04 at 14.59.52@2x.png

it’s also active if I check through the instance

already resolved for this one. thanks a lot for your help <@U01KZ61GV5Z> :pray:  appreciate it a lot

at the moment, those defaults were just copied over from the underlying library. You can change the preferred device list to suit your own needs, though I'm considering changing the order since a lot of people are preferring speaker first

Hi <@U03S23VFW0L>
If your need is solely to create access tokens, you can look at the file here and adapt it for swift: <https://github.com/livekit/server-sdk-js/blob/main/src/AccessToken.ts>
You don’t need to interact with the livekit-server directly if all you want to do is creating tokens.

If you need to interact with the server directly (e.g. listing rooms) you could put the  logic in a serverless function written in one of the languages the livekit server sdks support (js, go, php, python, ruby) and query that function from your swift code.
On a lower level, the server sdks use twirp (<https://github.com/twitchtv/twirp>). From what I can see twirp also has a swift client, so it could be an option for you to create the API calls with that by looking at how other livekit server sdks make the twirp calls, if the serverless function is not an option.

Hey <@U03S23VFW0L>, I'm align with <@U02EU2ZPTQU>. If you are familiar with coding and such, I would say spinning up a quick serverless endpoint through <https://vercel.com/|versel >should be farily easy to get the JS SDK Server up and running with an endpoint to generate your tokens for your clients.

Because Token creation requires your public api-key &amp; your secret key, I wouldn't try to mimic token creation on client side. You'd be running into security issues.

I get it. I'm woking on generating the token on the server with Vapor. Then calling the server APIs using twirp.

Anyone who finds this in the future, I have created a package <https://github.com/iankoex/LiveKitVapor.git|here>

You shouldn't need to modify anything in S3

I think this is likely due to your Kubernetes pod not having external DNS access.

can you reproduce this with <http://example.livekit.io|example.livekit.io>?

The client sdk will try to reacquire devices when regaining focus, but that should not happen if the track has been muted previously.
Could it be that the mobile browser refreshes the page in your case ?

Nope, it's not refreshing. And yea, I was able to reproduce this on <http://example.livekit.io|example.livekit.io>

Could you file an issue with instructions to reproduce? thanks!

Have raised an issue. <https://github.com/livekit/client-sdk-js/issues/383>

onLeave.png

hi <@U036FGT00TT> do you know which version of livekit-client you are pulling in?

Hey <@U01KZ61GV5Z>, `npm list` gives me this:
_"@livekit/react-components"_: "1.0.0",
_"@livekit/react-core"_: "1.0.0",
_"livekit-client"_: "1.1.9",
_"livekit-react"_: "0.9.2",

can you try setting livekit-client to `1.2.10` ?
also, you won't need the livekit-react dependency, that package is deprecated.  The new package to use should be `@livekit/react-components` , which you've already included.

Just ended up constructing an OrderedDictionary out of the new remoteParticipants and passing that in to the ForEach struct in the end :man-shrugging:

livekit-livekit-1  | 2022-08-07T22:22:32.993Z	INFO	livekit	service/rtcservice.go:213	new client WS connected	{"room": "myroom", "participant": "user002", "remote": false, "connID": "CO_YVB5RejXSuai", "roomID": "RM_Z8rnv3vGjGRA"}
livekit-livekit-1  | 2022-08-07T22:22:32.996Z	INFO	livekit	rtc/room.go:276	new participant joined	{"room": "myroom", "roomID": "RM_Z8rnv3vGjGRA", "pID": "PA_psqkcHV4nHpp", "participant": "user002", "protocol": 8, "options": {"AutoSubscribe":true}}
livekit-livekit-1  | 2022-08-07T22:22:33.312Z	ERROR	livekit	rtc/signalhandler.go:24	could not handle answer	{"room": "myroom", "roomID": "RM_Z8rnv3vGjGRA", "participant": "user002", "pID": "PA_psqkcHV4nHpp", "remote": false, "error": "could not set remote description: InvalidModificationError: invalid proposed signaling state transition: stable-&gt;SetRemote(answer)-&gt;stable", "errorVerbose": "InvalidModificationError: invalid proposed signaling state transition: stable-&gt;SetRemote(answer)-&gt;stable\ncould not set remote description\<http://ngithub.com/livekit/livekit-server/pkg/rtc.(*ParticipantImpl).HandleAnswer|ngithub.com/livekit/livekit-server/pkg/rtc.(*ParticipantImpl).HandleAnswer>\n\t/workspace/pkg/rtc/participant.go:663\<http://ngithub.com/livekit/livekit-server/pkg/rtc.HandleParticipantSignal|ngithub.com/livekit/livekit-server/pkg/rtc.HandleParticipantSignal>\n\t/workspace/pkg/rtc/signalhandler.go:23\<http://ngithub.com/livekit/livekit-server/pkg/service.(*RoomManager).rtcSessionWorker|ngithub.com/livekit/livekit-server/pkg/service.(*RoomManager).rtcSessionWorker>\n\t/workspace/pkg/service/roommanager.go:435\nruntime.goexit\n\t/usr/local/go/src/runtime/asm_amd64.s:1581"}
livekit-livekit-1  | <http://github.com/livekit/livekit-server/pkg/rtc.HandleParticipantSignal|github.com/livekit/livekit-server/pkg/rtc.HandleParticipantSignal>
livekit-livekit-1  | 	/workspace/pkg/rtc/signalhandler.go:24
livekit-livekit-1  | <http://github.com/livekit/livekit-server/pkg/service.(*RoomManager).rtcSessionWorker|github.com/livekit/livekit-server/pkg/service.(*RoomManager).rtcSessionWorker>
livekit-livekit-1  | 	/workspace/pkg/service/roommanager.go:435
livekit-livekit-1  | 2022-08-07T22:22:35.312Z	INFO	livekit	rtc/participant.go:710	closing participant	{"room": "myroom", "roomID": "RM_Z8rnv3vGjGRA", "participant": "user002", "pID": "PA_psqkcHV4nHpp", "remote": false, "sendLeave": true, "reason": "RTC_SESSION_FINISH"}
livekit-livekit-1  | 2022-08-07T22:22:35.313Z	INFO	livekit	rtc/room.go:393	closing participant for removal	{"room": "myroom", "roomID": "RM_Z8rnv3vGjGRA", "pID": "PA_psqkcHV4nHpp", "participant": "user002"}
livekit-livekit-1  | 2022-08-07T22:22:35.314Z	INFO	livekit	service/rtcservice.go:232	source closed connection	{"room": "myroom", "participant": "user002", "remote": false, "connID": "CO_YVB5RejXSuai"}
livekit-livekit-1  | 2022-08-07T22:22:35.314Z	INFO	livekit	service/rtcservice.go:259	exit ws read loop for closed connection	{"room": "myroom", "participant": "user002", "remote": false, "connID": "CO_YVB5RejXSuai"}
livekit-livekit-1  | 2022-08-07T22:22:35.314Z	INFO	livekit	service/rtcservice.go:193	server closing WS connection	{"room": "myroom", "participant": "user002", "remote": false, "connID": "CO_YVB5RejXSuai"}

livekit-livekit-1  | 2022-08-07T22:25:48.754Z	ERROR	livekit	rtc/mediatracksubscriptions.go:453	could not write RTCP	{"room": "myroom", "roomID": "RM_fW6aqGfzxEFu", "participant": "user001", "pID": "PA_FQoXcS8iAq4x", "remote": false, "error": "the DTLS transport has not started yet"}
livekit-livekit-1  | <http://github.com/livekit/livekit-server/pkg/rtc.(*MediaTrackSubscriptions).sendDownTrackBindingReports.func1|github.com/livekit/livekit-server/pkg/rtc.(*MediaTrackSubscriptions).sendDownTrackBindingReports.func1>
livekit-livekit-1  | 	/workspace/pkg/rtc/mediatracksubscriptions.go:453
livekit-livekit-1  | 2022-08-07T22:25:48.926Z	ERROR	livekit	rtc/signalhandler.go:24	could not handle answer	{"room": "myroom", "roomID": "RM_fW6aqGfzxEFu", "participant": "user001", "pID": "PA_FQoXcS8iAq4x", "remote": false, "error": "could not set remote description: InvalidModificationError: invalid proposed signaling state transition: stable-&gt;SetRemote(answer)-&gt;stable", "errorVerbose": "InvalidModificationError: invalid proposed signaling state transition: stable-&gt;SetRemote(answer)-&gt;stable\ncould not set remote description\<http://ngithub.com/livekit/livekit-server/pkg/rtc.(*ParticipantImpl).HandleAnswer|ngithub.com/livekit/livekit-server/pkg/rtc.(*ParticipantImpl).HandleAnswer>\n\t/workspace/pkg/rtc/participant.go:663\<http://ngithub.com/livekit/livekit-server/pkg/rtc.HandleParticipantSignal|ngithub.com/livekit/livekit-server/pkg/rtc.HandleParticipantSignal>\n\t/workspace/pkg/rtc/signalhandler.go:23\<http://ngithub.com/livekit/livekit-server/pkg/service.(*RoomManager).rtcSessionWorker|ngithub.com/livekit/livekit-server/pkg/service.(*RoomManager).rtcSessionWorker>\n\t/workspace/pkg/service/roommanager.go:435\nruntime.goexit\n\t/usr/local/go/src/runtime/asm_amd64.s:1581"}

is your port forwarded/opened the same way? it is a pretty large port range to open/forward.

Do you have a load balancer in front of both of these server instances then?

There is no need to open a port. The servers are not on AWS. Firewall is passive.

Yes, as you said, there is a load balancer in front of it.
Port 443 is forwarded to port 7880.

The connection is not always bad, but mostly bad. When it connects, it disconnects very quickly.

successfully connected to myroom in 3617ms 45.33.13.129:50743 
participant user001 connected  
connection quality changed user001 excellent 
could not connect: could not connect PeerConnection after timeout 
Reconnecting to room 
Successfully reconnected. server 45.33.13.129:50743 
disconnected from room {
  "reason": 6
}

Is there a specific documentation for multi-node installation?

there is with using Kubernetes, we don't yet have docs on multi-node via pure VMs.

Which file do I need to look at so I can know why?

Hi Osman,

One of the current constraints with LiveKit is that a single room must be hosted on a single node. So, the situation you describe with folks in the US, Germany and France all being in the same room, but on different nodes, is not possible at this time.

We do have some interesting updates on this particular topic coming soon though. :blush: 

You are magnificent. I follow your approaches closely and wish you success in your work. Thank you

<@U01KZ0SCDKL> doesn't LiveKit route all users joining the same room to the same node?

yeah, livekit does automatically route all users to the same node behind the scenes. we outline this process in <https://docs.livekit.io/deploy/distributed/#multi-node-routing|our docs>

This is very much dependent on browsers to support. WebRTC *requires* all clients to support VP8 and H.264. For VP9/AV1, it's highly browser dependent.

In the case of an SFU though, supporting new codecs have an additional layer of complexity. (i.e. what happens when half of the users support AV1, and the other half doesn't?)

With that said, we are working on AV1 support in a way that it degrades nicely across all clients. This is coming in the next few months.

Nice, I knew livekit would have a great solution for that :clap:

hey mark, sorry, docs were missing for the iOS broadcast extension. you can follow the instructions here: <https://github.com/flutter-webrtc/flutter-webrtc/wiki/iOS-Screen-Sharing#broadcast-extension-quick-setup>

our example broadcast extension can be found here: <https://github.com/livekit/client-sdk-flutter/tree/main/example/ios/LiveKit%20Broadcast%20Extension>

<@U01NZKTNE1H>
Thank you for your support.
However after everything setup and build successfully, I still cannot manage to make the screen share work.

On my own project, the problem remains, also if I turn off the screen share, the app freezes.

I tried the example, and when I turn on the screen share, the following message appears and the app crashes.
```flutter: 12:07:51: setSourceEnabled(source: TrackSource.screenShareVideo, enabled: true)
[unspecified] container_create_or_lookup_app_group_path_by_app_group_identifier: client is not entitled
* thread #1, queue = 'com.apple.main-thread', stop reason = EXC_BAD_ACCESS (code=1, address=0x0)
    frame #0: 0x00000001dbaf2e90 libsystem_platform.dylib`_platform_strnlen + 16
libsystem_platform.dylib`_platform_strnlen:
-&gt;  0x1dbaf2e90 &lt;+16&gt;: ldr    q0, [x2]
    0x1dbaf2e94 &lt;+20&gt;: adr    x3, #-0x34                ; ___lldb_unnamed_symbol2$$libsystem_platform.dylib
    0x1dbaf2e98 &lt;+24&gt;: ldr    q2, [x3], #0x10
    0x1dbaf2e9c &lt;+28&gt;: and    x4, x0, #0xf
Target 0: (Runner) stopped.```

Have you set up your app group identifier?

For the example project, you'll need to change the app group identifier to your own iirc

Yes, exactly. I changed to a different one

As well as the app group identifier in the info plist?

Sorry, I forgot that.
Now the broadcast is working, But when I disconnect the room, the example app freezed

I found the reason why the example freezed.
Thank you for your help again.

I will try to find out why it did not work on my own project

You mean the below ?
&gt; I found the reason why the example freezed.


Yeah, what caused the example app to freeze?

When the popup shows to select which to broadcast, I chose the broadcast extension of my own project, not the example app’s broadcast extension

Ah, haha. That probably shouldn't freeze anyways, I'll look into that

the timstamp is encoded based on timebase, which is what we are retrieving from the header itself. I believe it. should work for variable FPS streams. Does it not work for you?

the header is defined <https://chromium.googlesource.com/chromium/src/media/+/master/filters/ivf_parser.h|here>.

my situation is that i write the RTP packets to file using ivfwriter, then read later into a filetrack using ivfreader

i’ve found that it plays back too fast. this also happens if i load the ivf file into VLC

the timebase is hardcoded as 1/30, and the ivfwriter simply increments the timestamp by 1 on each RTP write.

so, it seems that the client must be sending frames at less than 30fps (about 10-30% less it seems).

now i thought i could grab the timestamp from the RTP packet, but it seems the ivf timestamp is really more like a frame count that gets multiplied by the timebase - it’s not designed for vfr

so the link you sent confirms that. i guess i can modify ivfwriter to write non-integer timestamps, that might work

so it sounds like on the writer side, it wasn't encoded with TS in mind.

h.264 is trickier since this information isn't encoded. as a result we only handle constant FPS for H.264, and it has to be provided by the user

yep i noticed… does livekit egresss handle all this by using RTP timetstamps?

Egress can only rely on RTP timestamps because it takes WebRTC input -&gt; other output.
We are also working on Ingress that's the other direction.. there we do parse out the incoming presentation TS from containers and translate that to RTP TS.

hi Vincent, currently it's not possible to set max_participants when the room is autocreated.

to work around this, you can disable auto creation and always manually create rooms with the `CreateRoom` API

Image from iOS.MOV

Image from iOS.jpg

Image from iOS.MP4

Hi Khush, it is very hard to follow the issue with video segments. Could you describe the repro steps and paste the errors in text format in the thread?

1&gt; fires an event in unity  using 'react-unity-webgl' and calls a function inside react

2&gt;the function inside react looks something like this

var audioTracks = remoteParticipant.audioTracks;

        audioTracks?.forEach((value, key) =&gt; {
          console.log("next");
          console.log(value);
           value?.setSubscribed(true);}

note: on commenting out the line "
           value?.setSubscribed(true);" everything works fine

3&gt; opens the application on browser everything is fine until i open more than 2 instances of application

4&gt; but if it exceeds more than 3 , *randomly* i get the following error and the entire unity webgl app crashes

next-dev.js?3515:25 An abnormal situation has occurred: the PlayerLoop internal function has been called recursively. Please contact Customer Support with a sample project so that we can reproduce the problem and troubleshoot it.

window.console.error @ next-dev.js?3515:25
_JS_Log_Dump @ <http://webshoot_spacex.framework.js.br:4|webshoot_spacex.framework.js.br:4>
$func10762 @ 063b8b1e:0x1da23a
$func51843 @ 063b8b1e:0xc57319
$func26594 @ 063b8b1e:0x5a54f7
$func13236 @ 063b8b1e:0x258de1
$func770 @ 063b8b1e:0x6dada
$func26509 @ 063b8b1e:0x5a1c1e
$func51761 @ 063b8b1e:0xc54e1b
$vk @ 063b8b1e:0x3fc4f4
(anonymous) @ <http://webshoot_spacex.framework.js.br:4|webshoot_spacex.framework.js.br:4>
(anonymous) @ <http://webshoot_spacex.framework.js.br:4|webshoot_spacex.framework.js.br:4>

tried even dynacast enabled nothing works

<@U01KZ61GV5Z>  <@U01KZ0SCDKL> <@U02BH3ZHJ0Y>  pls need help on this, Is it possible to solve this

This error doesn't look like it's related to livekit.
Sorry, I don't have any knowledge or experience with `react-unity-webgl` .

Maybe subscribing to the track triggers another event that's somehow causing an infinite loop..

Hi! David thank you so much for the reply i think you're right it should be a Unity issue, livekit for js is perfectly good

Thank you ! I’ll check what can be done

Hey Shubham, could you please share a little more about what you're doing? Include code snippets if you can to illustrate the problem.

cc <@U022KB72NFJ> 

Hi <@U01KZ0SCDKL>, It's so great hearing from you, looks like it was a miss from my end in unity. I made the changes and started rebuild. Hope it will work fine now :slightly_smiling_face:

Sounds good <@U03PMN3FVRQ>, we're always around if you need help! :blush: 

After digging, I misconfigured the `UIBackgroundModes` key in my plist. All is good :slightly_smiling_face:

Regarding the PiP integration for iOS 15, does this one work too?

The new PiP API is interesting, i will take a look :slightly_smiling_face:

^ bump very interested in an example :pray: