LiveKit Community

This <https://livekit-users.slack.com/archives/C01KVTJH6BX/p1682016022044869|thread> should get you up-to-date on what's happening in this regard.

<@U0558RZ2L20> please only post messages once in a single channel. :pray::skin-tone-4:  

I posted in general initially but later realized there is a dedicated channel for cloud

How are you getting the participant that is being passed into this context? If it's via ParticipantLoop, it would include the local participant as well as remote

```const CameraRoom = () =&gt; {
    const participants = useRemoteParticipants();

    return (
      &lt;ParticipantLoop participants={participants}&gt;
        &lt;Video /&gt;
      &lt;/ParticipantLoop&gt;
    );
  };```
I am only passing in remote participants :thinking_face:

ah I see.. that is odd. are you seeing this with the latest version (v1.0) of components?

CC: <@U02EU2ZPTQU> <@U03RM7G8Q6A>

ah `"@livekit/components-react": "^0.8.2"`  let me update

Thanks for the confirmation, I’m looking into it now.

<@U045UBXUYTW> the bug is on our side. Should be fixed with this: <https://github.com/livekit/components-js/pull/471>
Can you confirm that setting the volume still works for you, even though you see a warning?

Should I still write:
```&lt;AudioTrack source={RemoteAudioTrack.Source.Microphone}...```
or is
```&lt;AudioTrack source={Track.Source.Microphone}...```
the same thing?

same thing, we usually use `Track.Source.Microphone`

Somewhat related since it seems like that was just a false console log, I am able to lower and raise the volume but I can't mute it by setting to 0. Is this the expected behavior?

Definitely not expected. Can you share the code snippet to your implementation? In JS `0` can also be interpreted as `false`, maybe it’s a logic error (I speak from experience :slightly_smiling_face: )? If that’s not the case, I’ll take a look tomorrow.

I have a slider that changes sets the volume.

I have also printed the value of {volume} and its `0` but i still can make out the audio even though its faint.

```const [volume, setVolume] = useState&lt;number&gt;(0.5);

...

&lt;AudioTrack source={Track.Source.Microphone} volume={volume} /&gt;```

are you sure that this is the only &lt;AudioTrack/&gt; instance you have of that participant in the app ?
you can also verify by checking the volume level of the `&lt;audio/&gt;` html element in your tree and check whether that actually says 0.

<@U045UBXUYTW> have you solved the problem with the sound still audible or does it need further investigation from our side?

Sorry I’ll try what Lukas recommends when I am back next week. I am on vacation this week. I’ll follow up though for sure!

I just realised why you would still hear the audio volume with faint volumes prior to <@U03RM7G8Q6A>’s fix above:
before that we were checking `if(volume)` which - for the value `0` is falsy in JS :face_palm:
but the fix above should take care of this!

Oh hmm I think I am waiting for a release do have ocupes lates change so that might be why.

Ok back from vacation, I updated to 1.0.2 and it now appears to be properly muting the track. Thanks!

that error message doesn't sound like it's Egress. can you confirm what image you are deploying?

hi, have you changed anything on your end? especially Electron version and/or SDK version?

The audio track silence warning is likely indicating the hardware device acquired could not produce valid audio samples.

Thanks for your reply!

I haven’t changed the sdk and electron versions!

The livekit server is built on an ecs container.
The server load has not reached max when we checked.

It only occurs under certain users,
The bug was suddenly fixed at a certain point in time, so the cause of the problem is still unknown,

I guess we have to investigate the possible causes one by one.

This seems more like a client/hardware issue to me. if you can get a repro, please file a bug report.

Was the user using Windows? Windows tends to have restrictive media device limitations where only a single app could acquire the hardware. We've seen errors similar to these before when the user had another app that's using the hardware.

Hi david, I’m working on same development team with <@U055MN785FC> and am invesitigating this issue, too.

&gt; Was the user using Windows?
No, the user was using OSX and Apple Silicon CPU. This issue is still being investigated, however, the issue is not reproducible and the conditions for reproduction are unknown, therefore it is taking a long time to produce a report. We will send you a bug report when we could reproduce the problem.

&gt; This seems more like a client/hardware issue to me
I guess the problem is not a client/hardware issue because that problem was happening even if the user restarted their machine.

no problem, we'd be interested in learning what the root cause is. there are a good number of users running on Electon with livekit.

publishData would be the appropriate way to do this

The thing is when I sent a small image file that was converted to List&lt;int&gt;, it took too long time and threw TimeOutException. I'm sure there is a limit or something. I wonder how much it can handle. <@U01NZKTNE1H>

LiveKit isn't designed to send files.. while you *can* use data channels for this, it's not as easy to use as just having participants to pull a file from an URL

how large was the image that you are trying to send?

it'd be better if you can chunk it up into pieces of &lt;4k or so, instead of sending as a large payload.

Yeah I'm sure it's bigger than 4k so I changed the strategy to sending the url :grin:

A self-hosted deployment of LiveKit is HIPAA compliant to the extent that the application you’re providing end users is HIPAA compliant.

LiveKit Cloud is not HIPAA compliant today, but yes, we do have plans for it to be. Cloud is SOC2 Type II, and very soon GDPR-compliant. HIPAA will come next.

Great, thanks <@U01KZ0SCDKL>. Is there an ETA on Cloud HIPAA-compliance?

We don’t have one yet, but I’m happy to share one when we do. Are you working on a healthcare project? Feel free to DM me or <@U02BH3ZHJ0Y> if you’re open to sharing more about your use case.

Do you and <@U02BH3ZHJ0Y> expect LK Cloud to be HIPAA-compliant with the year?

I’m not making any promises, but I do expect it to be, yes.

Not yet, but we’ve been meaning to OSS it for a while. I’ll revive that convo with the team! In the meantime, perhaps you can DM me your suggested changes?

tomorrow? lol :stuck_out_tongue_closed_eyes:

image.png

In this 19h+ session the only participant left ~8m after joining, and the session is still running. Our webhook terminates the connection if there are no participants other than our camera left.

In this 9h session, the last participant left at 7:53pm however the room ended at 10:47 almost 3 hours later.

that seems bizarre, it should be fired instantaneously and retried if it couldn't reach the websocket endpoint.. can you DM me the session in question? we'll take a look at the server logs

Sure thing I’ll send it over in a bit. Thanks 

for sure! we recommend running within containers for the standard deploy.

on 2, definitely! It's open source and you are welcome to use any piece of it however you want :slightly_smiling_face:

Welcome <@U057CC82HK2>, good to have you here!
I will let others speak if there is a better way to do this with the Go SDK, but another way to get video into a LiveKit room is via ingress: <https://docs.livekit.io/server/ingress/>

you could use our new Rust SDK as well, it'll help you connect to a room as a participant, grab tracks and decode them for you for processing

The ingress could be a good idea <@U03RM7G8Q6A>. Would it be possible to run the service on a local network and offline?

<@U057CC82HK2> yes, you can host the ingress service yourself. Please check the GitHub repo and docs for more information:
• <https://github.com/livekit/ingress>
• <https://docs.livekit.io/server/ingress/>

ORTC seems great. However, our core protocol has to work with browsers, so it's better for us to stick to the same protocol and ensure all clients implements them the same way.

but yes, there are quite a bit of annoyances having to deal with SDPs

<https://media.tenor.com/CaOZb6NWQRMAAAAC/the-simpsons-bart.gif>

cc <@U02EU2ZPTQU> and <@U02GAC58A79> for a status update here!

hey <@U046CJ4SRGU> , we have entered the last mile, flutter/js already supports E2EE (VP8, H264), but we are still making some fixes for the correct switch to mute after track mute, if you want to experience you can use flutter client sdk 1.3.0, and you will have an encrypted room.

Thanks for the update <@U02GAC58A79>! I don't have much experience with flutter so I will have to wait for the React Native release

Hey Daniel, you are correct. For the subscriber, there is no explicit confirmation of it succeeding. Client's WebRTC stack would fire  `onTrack` callbacks for each track that was received in that offer. Since the offerer decides what to use as TrackIDs, we simply set it to "{participantID}{trackID}" so we can match it to the right participant.

Thanks, makes sense! :+1:

What I was wondering if the track ID really set to `{participantID}{trackID}` in the current versions of the SFU (`master`)? - I.e. I’ve checked the code and it seems that the local tracks that the SFU creates and adds to the peer connection (i.e. `DownTrack`) have their ID set to the `livekit.TrackID`, i.e. to the `sid`. The `NewDownTrack()` creates a new `DownTrack`
with the `id` based on the `TrackReceiver` that is passed as an argument to the `NewDownTrack()` . That `TrackReceiver`’s `TrackID` has a type of `livekit.TrackID` (sid).

I tried to track it down to the `MediaTrackReceiver` (that is, the `TrackReceiver` passed to the `NewDownTrack()`), and it’s `ID()` is defined like this:
```func (t *MediaTrackReceiver) ID() livekit.TrackID {
	t.lock.RLock()
	defer t.lock.RUnlock()

	return livekit.TrackID(t.trackInfo.Sid)
}```
Or do I miss something and there is a participant ID that is attached to the `sid` additionally? :slightly_smiling_face:

Quick announcement. We just released livekit-server <https://github.com/livekit/livekit/releases/tag/v1.4.2|v1.4.2>. (helm release will follow tomorrow). In this release, you'll get a preview of our upcoming support for VP9 and AV1 codecs (both with full SVC support). SVC codecs are very cool, giving us a wider range of bitrate adaptation from a single published stream. We'll be writing more about this in the near future.

<https://github.com/livekit/egress#running-locally>

if you made code changes, you can rebuild the egress docker image by running `mage build`

OK, thank you. This means that we will use docker to run service on local.

I see on some android devices problems with sending streams and when I allow only opus and vp8 in config.yml file, all works fine, but now I want also connect IP cameras to room where I wrote small utility to read rtsp stream and pubish to a room, but it's h264 codec and I had to allow it as well. As you based on this I want to tell to this android device to use vp8 codec, or during creating a room set config to only use video/vp8 and audio/opus

is this with the client-sdk-android library?

hmm don’t have the codecs settings exposed at the moment. you can use something like <https://github.com/livekit/client-sdk-android/blob/482a0653099388cfb2028655c84bfb7928b88bc7/video-encode-decode-test/src/main/java/io/livekit/android/videoencodedecode/WhitelistSimulcastVideoEncoderFactory.kt#L12|this> to specific which codecs are allowed. pass this into the `LiveKitOverrides` of `LiveKit.create()`

Screen Recording 2023-05-10 at 4.49.52 PM.mov

your dreams coming true soon <@U04N2ULR17C> 

Looks great! Are you using com.unity.webrtc?

no, this is built on top of our Rust SDK, which wraps libwebrtc directly

it’s here y’all, we just haven’t announced it yet — we’ll do that today :stuck_out_tongue:
<https://github.com/livekit/client-sdk-unity>

Yes. I looked through it. How long can you expect the official version to be out?

is there a roadmap for Android and iOS support?

Thank you <@U01KZ0SCDKL>! This is super exciting

<@U04N2ULR17C> this is the official version, but it’s in a Developer Preview state. There’s a lot of targets to support, so we likely won’t 1.0 this SDK until there’s a significant amount of mileage (users and/or usage).

<@U03RW2KK1DJ> we’re definitely going to support the mobile platforms. I’ll let <@U022KB72NFJ> comment on plans there.

just wait until the skybox shows up in the next video :wink:  

I assume this is in relation to <https://github.com/livekit/client-sdk-flutter/issues/282|this issue>?

I guess we could use the admin api to kick people out of a room when their time expires, but maybe there is an easier way using the ttl of the access token?

Hello <@U057EBR4PHN>, I'm not sure if there's an easier way to do it within LiveKit's server through a Flag, but I don't think I've ever seen that parameter.  With this said, one way you could easily achieve this would be to have a cron job running every minute on a server and as you've said using the Server API, loop through your participants and if one has expired, you can use `removeParticipant` method to remove him from the room. See: <https://docs.livekit.io/server-sdk-js/classes/RoomServiceClient.html#removeParticipant>

Access token expiration only controls the _initial_ join time. Once a participant joins, they would not be "timed out". Removing participants yourself with a server API would be the best option to achieve what you are looking to do.

Hi Christopher,
these kind of fixes are very welcome! Feel free to open a PR against the repo!
<@U03RM7G8Q6A> and me are the most active on the components-js repo.

Do we have a way of translating user-visible strings? Or should I just add an English ARIA-label for muted/unmuted?

english for now, but there’s <https://github.com/livekit/components-js/issues/452> so we might want to change that at some point.

In order to reuse the labels across frameworks, I think it would be best to define them in `components-core` . there’s already a setup method for each component that adds a at least a class name as prop, so I think that would probably be the right place for the aria-label as well

The design of LiveKit is everything-through-SFU

Out of curiosity, are you imagining a hybrid app that switches to SFU when there’s 4 or more participants?

In my project, I anticipate having many rooms with only 2-4 people

does “many” mean tens, hundreds, or thousands?

Well of course my project will gain massive adoption upon release, so naturally it will be hundreds of thousands :wink:

of course, how silly of me to ask :stuck_out_tongue:

So if it were me, I’d start by just doing everything through the LiveKit SFU and then implement p2p as an optimization later, after you have hundreds of thousands of users and your server costs are skyrocketing :slightly_smiling_face:

mostly just because our client sdks are easier to work with than raw webrtc stuff

I'm eagerly awaiting that Unity client! If you're looking for any extra testers, please put me on the list :bow:

oh I think actually we are. cc <@U022KB72NFJ>

Hey <@U055P5XKNEL>, the current docs site is not open source (yet). What are you looking to fix? lmk here and I'll make the change

<@U01KZ61GV5Z> Thanks for the reply. I want to contribute in general. It is difficult to understand a lot of stuff.

Also some broken stuff:
Trying to access this URL on Firefox:
<https://docs.livekit.io/getting-started/quickstarts/>
I never have any problems like this one.
No problems on private window but rest of the website works fine. Only at `quickstarts/something` route and children are broken.
First I get this error: `Application error: a client-side exception has occurred (see the browser console for more information).`
 Then after refresh `500 Internal Server Error.`
It maybe because of an extension but why just that route?


Also:
404 error at JavaScript Client Quickstart Guides
<https://docs.livekit.io/getting-started/quickstarts/javascript/>
These are also the same:
 <https://docs.livekit.io/getting-started/quickstarts/unity-web/>
 <https://docs.livekit.io/getting-started/quickstarts/flutter/>
Tried with FIrefox(Version 113.0.5672.93 (Official Build) (64-bit)) and Chrome(Version 113.0.5672.93 (Official Build) (64-bit)) on Windows 11 Pro(22H2, OS: 22621.1702)

we don't have a way to do this today from the java SDK.

you'd have to create a token with a grant field:
`sha256` set to a sha256 hash of the payload. See <https://github.com/livekit/protocol/blob/main/webhook/url_notifier.go#L103|here> for our implementation from webhook notifier

aw gotcha, was hoping there was an easy way to generate the token with the CLI or something

Looks cool, would like to try it out :slightly_smiling_face: Do you use the Swift SDK ?

I think you can still enable red, but I’m not sure for dtx, google speech to text requires a continuous stream.

You can try to still enable it, I think Kitt will autorecover if Google closes the stream

Hello <@U048LRAPUJJ>, would you be able to provide more context around your implementation, such as which Client SDK, etc?

Am using the JavaScript SDK in an angular application

is this using LK Cloud or you're running your own LiveKit Server?

Any Console Logs which could provide some insights as why this would be happening?

I just log some message when disconnecting event is fired .

Can you turn debug logs on in JS SDK and attach logs of when it did this last? It'll also be helpful to see the session link to the dashboard as well

IMG-20230518-WA0001.jpg

" RM_etdD4EPomCWH " is a session I'd for one of the sessions auto disconnect happened.

The image is screen shot of livekit logs

<@U048LRAPUJJ>, please export the entire log file from Chrome including timestamps. I'm unable to see anything useful with limited context.