LiveKit Community

that's right! those are the primary difference.

might not be a bad idea to post the error message as text also

Hi! <@U02Q46USG4D> thanks for the reply I’ve posted all the errors in  text format if you click on the link 

<https://livekit-users.slack.com/archives/C01KVTJH6BX/p1659985708489769|https://livekit-users.slack.com/archives/C01KVTJH6BX/p1659985708489769>
Please click on the link

we are investigating, looks like there's an issue with HLS egress in the last release. fix coming shortly!

Heya Daniel, have you two made any progress on this? Are you running lk-server in kubernetes or on VMs? I have referenced this connectivity doc from linode a few times and found it pretty helpful: <https://www.linode.com/docs/guides/troubleshooting-basic-connection-issues/>
Do you have logs from both redis and lk-server around the disconnection times that might give a better indication of what is happening?

No, we haven't figured it out yet. They are running in vm's.

Thanks for that doc link, will look into that more.

I think the challenge that we are having is really debugging where the connection problem is.

Osman tried several different implementations of redis. Here's what he tried:

1- I installed the external Redis server myself
2- I installed the external Redis server via Linode's marketplace
3- I set up the external Redis server with WAN ip
4- I set up the external Redis server with LAN ip
5- I tried upgrading the external Redis server to the latest version
6- I set up the external Redis server via <http://CloudClusters.io|CloudClusters.io>

So it doesn't seem to work properly no matter where redis is installed (unless it is installed on the same vm as livekit in a single instance deployment).

Thanks for the doc link we'll look into that, maybe it can give us some hints.

<@U027YH2JNB1> can you post some more details here on it?

Yes, it was as you said, <@U03S81VRSHF> I'll collect more log data and come back here. Thanks for the link

Afaik the redis package has some reconnection logic around pubsub, so I would think it should reconnect. Debug logs will definitely be helpful, you can set that level here for livekit server: <https://github.com/livekit/livekit/blob/master/config-sample.yaml#L98>

livekit-livekit-1  | 2022-08-12T17:16:45.667Z	ERROR	livekit	rtc/signalhandler.go:24	could not handle answer	{"room": "myroom", "roomID": "RM_ob8zuMzuQUZj", "participant": "user01", "pID": "PA_5HSavBiXexAJ", "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"}

Errors like this one aren't wholly unexpected, here's a thread with this same error that talks a bit more about that: <https://github.com/livekit/livekit/issues/602>
I would also expect to see something regarding redis in the log if it is in fact the culprit?

Yes, I found this page and read it. it doesn't mention the same problem as ours.

The reason we think the reason is redis, the only difference is that redis is external in single-node and multi-node setup with the same configuration.

so we are not sure if we have faulty redis

Well <@U03SP6DMWUS> you also even tried a cloud redis too rather than our own install right? That seemed to indicate to me that it wasn't an issue with the redis installation. Correct me if I am wrong about that.

yes I have connected redis server via <http://CloudClusters.io|CloudClusters.io>.

Any chance you could zip up the log and give me the timestamp of the issues?

we have access to these 2 nodes via a loadbalancer on the linode. With loadbalancer, nodes are connected via LAN IP addresses.

loadbalancer configuration
443 HTTPS -&gt; 7880

I couldn't find a configuration suggestion regarding the loadbalancer in the livekit documentation.

image.png

sometimes with no problem, sometimes i get a failed result for video/audio

On a failed test I would check <chrome://webrtc-internals/> and see if you can figure out a specific port range that is failing. I would double check <https://docs.livekit.io/deploy/ports-firewall/> as well and ensure you have that connectivity between nodes.

first of all thank you for your help.
but any port is not closed. firewall or iptables is not active. Since linode is not like aws, I don't need to open any ports via console.

The latest situation is as follows.

<https://livekit.socialglow.com/>
<https://livekit.socialglow.com/|1st person is successfully connecting to the room. The 2nd person is trying to connect to the room from another computer connected to the same LAN network. (I tried this situation on 2 phones, the same situation occurs. ie over WAN)>

<https://livekit.socialglow.com/|When the 2nd person tries to connect, the chrome console prints the following error. I get this error every time. And the same thing happens every time.>

<https://livekit.socialglow.com/|livekit-client.js:14650 Uncaught (in promise) DOMException: Failed to execute 'setLocalDescription' on 'RTCPeerConnection': Failed to set local answer sdp: Called in wrong state: stable>
<https://livekit.socialglow.com/|await (async)>
<https://livekit.socialglow.com/|handleSignalResponse @ livekit-client.js:13369>

<https://livekit.socialglow.com/|Whereas, there is no problem with the live3 single node, whose livekit.yaml files are almost identical. All participants successfully connect to the room.>

<https://livekit.socialglow.com/|There are 2 differences between multi-node and single-node.>
<https://livekit.socialglow.com/|1- multi node is using loadbalancer>
<https://livekit.socialglow.com/|2- multi node uses external redis>

<https://livekit.socialglow.com/|Attached are also console logs that I got from chrome.>

node1-livekit.socialglow.com-1660338618645.log

node2-livekit.socialglow.com-1660338603793.log

-even if a person is successfully connected to the room, after a while (after 45 seconds) it loses its connection by itself.
- I always see the same error when the second person tries to join the room while there is one person in the room.

'RTCPeerConnection': Failed to set local answer sdp: Called in wrong state: stable await (async)

node1- webapp logs
successfully connected to myroom in 3246ms 45.33.13.129:58517
participant user01 connected
connection quality changed user01 excellent
participant user02 connected undefined
participant PA_TtSV8sSBLZoK disconnected
participant user02 connected undefined
participant PA_FvBPtR5429if disconnected
participant user02 connected undefined
participant PA_TtSV8sSBLZoK disconnected
participant user02 connected undefined
participant PA_BgfK4PuoUtbd disconnected
participant user02 connected undefined
participant PA_EbsKaHXvNae6 disconnected
participant user02 connected undefined
participant PA_BgfK4PuoUtbd disconnected
participant user02 connected undefined
participant PA_EbsKaHXvNae6 disconnected
Reconnecting to room
Successfully reconnected. server 45.33.13.129:58517
disconnected from room {
  "reason": 6
}

node2- webapp logs
successfully connected to myroom in 2468ms 45.33.25.190:52071
participant user01 connected
participant user02 connected
disconnected from room {
  "reason": 0
}
could not connect: could not connect PeerConnection after timeout

I have been poking at this a little bit, but no luck so far. I will try and take a longer look next week. Let me know if you figure anything out, looks like our docs could use some help on a multi-VM setup :smiley:

Awesome, thanks for the help on this <@U02Q4J8KS85>! We're really hoping we can get LiveKit working for us :slightly_smiling_face:

We look forward to your detailed multi-node setup document.  <@U02Q4J8KS85> thanks

Heya <@U03S81VRSHF> and <@U03SP6DMWUS> dug into this today and have a solution for you. Livekit is expecting a unique hostname to be set per VM and uses it as a key in the distributed setup to send requests to the correct node: <https://github.com/livekit/protocol/blob/main/utils/id.go#L37>
However, linode does not set a unique hostname for it's vms, everything i've booted is named 'localhost'. Here's how you can set unique hostnames in linode: <https://www.linode.com/community/questions/18546/how-to-change-hostname>

Awesome, glad it worked! We've had some discussions on how to solve the duplicate hostname issue as well.

One more note on deployment. Since linode's nodebalancers only allow you to add VMs in the same region, I'm not 100% sure it will be super useful to run multiple VMs per region. Livekit currently runs rooms on a single node, so even with large rooms traffic will get routed to the same node. If you're expecting lots of different rooms with traffic coming from the same region; then the multi-node in a single region might work ok. However, I'd recommend doing a multi-region VM setup to allow traffic to connect to the closest server.

Very good suggestion. Thanks again. However, I don't think it can be very efficient as you say, since people are directed on a room/server basis. However, regardless of which server the room is opened on, I trust you very much to allow users to connect to the closest server. you will surely succeed!

In our use case we are anticipating a large number of small rooms with one person publishing a stream and the others only watching. We expect less than 10 people to be watching per stream. We're trying to build a one to many live stream similar to a facebook live or youtube live. Thanks so much for the help getting past this issue!

Hey <@U02Q4J8KS85> just letting you know that this fixed the issue for us and we have the cluster running now :slightly_smiling_face:

fixed it. for future reference for any searchers:

<https://www.digitalocean.com/community/questions/bucket-policy-do-spaces?comment=174627>

that functionality currently isn’t implemented on the Android SDK

Hello <@U02T8UUS3J8>, The JS Client SDK, yes allows you to do so. You can test an example with the demo site found here: <https://example.livekit.io/#/>

If you are looking for actual source code, Id say the following would be what you would be after: <https://github.com/livekit/client-sdk-js/blob/267b73078bb72cb948be01dcf7160842374aec9e/src/room/Room.ts#L502>

This is exactly what I was looking for, thank you!

Just tagging <@U01KZ0SCDKL>,  since he would probably be the person for this :slightly_smiling_face:

I did also send an email, feel free to ignore that if this is a better place to chat

Do Livekit core team or associated group provide enterprize support?

For example, I am testing this out and I want to manually specify the IP address so I am setting

```use_external_ip: false```

How do I tell livekit what the ip address is?

Nevermind I found it! <https://github.com/livekit/livekit/blob/master/config-sample.yaml>

glad you found it! If you set `use_external_ip` to true, it'll attempt to discover the external, public address on that machine

you could also override it by setting `--node-ip &lt;ip&gt;`  when starting livekit

Hey Jaspreet, LiveKit can indeed transmit arbitrary data.

<https://docs.livekit.io/guides/room/data/#data-messages|https://docs.livekit.io/guides/room/data/#data-messages>


<@U01KZ0SCDKL> Awesome, thanks for taking out time to reply

I guess this will have additionally benefit of scalability

what if I wanna persist these data streams and when somebody new comes in, I want to them to have access ability to pull the current state

can from server sdk persistance and replay is possible? if yes any examples in open source?

You can certainly hook up your own persistence layer/logic using the server SDK. I've been meaning to put together an example of this for some time. I'll see if I can cook something up this weekend. When you say replay, do you mean being able to run through the sequence of events that occurred in the “room”?

Replay for the whiteboard I wanna persist entire history

offcourse client will be responsible for rendering it

For whiteboard I need to send data in short bust of packets, from scalability point of view is this okay?

Is DataPacket_Kind.RELIABLE going into redis cache?

Gotcha, yeah, persistence and replay is something which, for now, you’d have to implement yourself (using the server SDK).

Redis isn't storing any media or data channel messages transmitted in your room

Are these going to be saved in memory when DataPacket_Kind.RELIABLE?

yes, so DataPacket_Kind.RELIABLE means that all connected clients will recieve this message and that data will be hold in memory and won’t be flushed

if I user unable able to connect for a while they will be droped out of rooom list

cc <@U01KZ61GV5Z> — don't know the answer to this one

RELIABLE delivery only applies to users that are currently connected to the room. If a user is disconnected,  it would not be delivered to them

the simple way to think about reliable / unreliable delivery is UDP vs TCP. Both protocols require clients to be connected at the moment of publishing.

What you are looking for might be persistence and access to shared state. For that, take a look at Participant Metadata. Those are persisted and we guarantee that every participant will see the same copy. In the future this can be extended to fine-grained fields / CRDTs

<https://docs.livekit.io/guides/room/data/#participant-metadata>

<@U01KZ61GV5Z> Thanks for taking time out and reply

So Participant Metadata is persisted in redis?

Is it advisable to save stuff like large svg kind of formats here?

Or would you recommend going for seperate datastore for whiteboard related data structures?

Redis is, I believe, in-memory storage by default. There are persistence options, but my guess is you should have separate, durable storage for your whiteboard data. 

Interesting so if livekit pod goes down, the on going calls are not recoverable?

<https://docs.livekit.io/deploy/distributed/|https://docs.livekit.io/deploy/distributed/>

Can on server side I can handle if pod goes down then restart that running room session as new session on new pod

if the pod goes down, clients will automatically reconnect to the load balancer address, which would place the room on another instance

<@U01KZ61GV5Z> thanks david. So livekit runs redis in persistence mode?

We use Redis as a message bus and DB, but do not require the data to be persistent. all instances periodically update the others via Redis.

So in distributed mode, it’s something like CRDT’s or RAFT to maintain the state across cluster?

LiveKit is certainly extensible. I know of one company I was recently introduced to that builds a whiteboard SDK. I pinged the founder for more info and will let you know.

In plugNmeet we're using excalidraw + websocket. You can have a look as an example: <https://github.com/mynaparrot/plugNmeet-client/tree/main/src/components/whiteboard>

This is also a valid path to take. You could add audio/video comms + presence using LK and build the whiteboard itself as a self-contained module. The tricky part is synchronizing state between the two systems, but not impossible and there are some projects using LK that do exactly <https://theportal.to/|this>.

Hello <@U02T8UUS3J8>, if you take a look at the sample / example file from the client sdk js, there's a working example in there. This said for reference this should be the code reference: <https://github.com/livekit/client-sdk-js/blob/main/example/sample.ts#L283>

audio output handling on browsers is generally messy &amp; inconsistent. As Pat pointed out, `switchActiveDevice` will attempt to perform the switch on browsers that support it.

did you configure it to store to cloud storage? that would be the recommended way to use Egress.

If using a local path, it's stored on the instance where Egress is running.

Hey <@U03R8BMAXC1>, is there a particular platform you're looking to do this on?

Go for the backend, swift/ios client. Thanks! 

<@U03TLG84LTD> Can you explain the “Whatsapp call feature” ?
You might need to use CallKit from Apple.

<@U02MAJLGN65>, I have used callkit with livekit. but having one issue specially when the app is in background and the phone is locked. There is no audio exchanged between to devices.

Looks like it changed here:

<https://github.com/livekit/client-sdk-js/blob/e5851ea6c9a8c4be74f60e9ece6f72cdf3560b5b/src/room/participant/Participant.ts#L72>

<https://github.com/livekit/client-sdk-js/blob/e5851ea6c9a8c4be74f60e9ece6f72cdf3560b5b/src/room/participant/Participant.ts#L178>
I think metadata should be set as default too.

hi <@U02LCP104CQ>,
yeah, you are totally right, thanks for the report!

Thanks <@U02EU2ZPTQU> for confirmation. This is affecting  Plug-N-Meet's wait for moderator approval logic here:
<https://github.com/mynaparrot/plugNmeet-client/blob/7a5ee3b05b00e5c9ab3dcc2647ed1e5ba4e00c78/src/helpers/livekit/HandleParticipants.ts#L73>

<@U01KZ0SCDKL> <@U01KZ61GV5Z> Hey guys, quick question on egress, lets say I want to just capture primary video and primary video stream might change based on voice detection. Mind you number of speaker can be like more than 10 at a time of I don’t wanna capture all the streams

<@U03T51VH7JS> when you say “primary video stream might change based on video detection”, do you mean that the participant you’re recording video from may change dynamically during a session?

voice detection* and yea I ethier want to update primary speaker on client or at server (if possible?)

check out this post:
<https://blog.livekit.io/livekit-universal-egress-launch/>

I _think_ what you want is Room composite, with your own custom template.

so yes, it is feasible without changes to the LK APIs, but you may need to provide a custom template.

In nutshell I want to switch between multiple speakers and end up with one video

with custom template? I thought it was like multiple video being captured in headless chrome

think of it more as a version/view of your client application being captured. That version could be exactly the same as your user-facing one, or it could be different, depending on your requirements.

the recorder is a hidden participant in the room

so, if you wanted to just have one video from the current speaker being captured, and then that video element switches based on who’s speaking, that’s a specific layout with some logic on when to switch the stream

you could programmatically mute other participants, other than the current speaker, as well

this can be done with the Go SDK, you can create a Track forwarder that would read from one room and publishing to the other. When publishing, you'd need a <https://github.com/livekit/server-sdk-go/blob/main/sampleprovider.go|SampleProvider> that could feed the publisher samples. That'd be the piece of code you'll need to write.

<@U01KZ0SCDKL> would be great if you can help us with this. Not sure how others are setting it up to work with render.

I'm not sure I’ve chatted with anyone here deploying to Render. Perhaps <@U02BBTRGF98> or <@U02Q4J8KS85> have thoughts.

I'm not familiar with Render, but it doesn't look like they support <https://community.render.com/t/opening-up-a-udp-port-for-incoming-traffic/821|udp traffic> .  I'd recommend using a cloud provider that supports advanced configurations as livekit has a wide range of connectivity requirements: <https://docs.livekit.io/deploy/ports-firewall/>

yes the problem is on deploying the container to render. it fails to start the deamon inside the container.

thank u for the suggestion <@U02Q4J8KS85> 

Right, that is because as you mentioned, we use host networking for docker since there are so many ports forwarded to the container. Even if you didn't use host networking and manually forwarded the ports, you'd run into issues since render doesn't support udp connections. In theory you could do turn/tcp only, but that's not something I would recommend. We do have docs on setups in lots of cloud providers including digital ocean/aws/gcp and some people have had success on linode as well. I'd say your best bet is to go that route.

btw I’m trying Digigal Ocean as an alternative, but we don’t have yet a domain name, would it be possible to activate it directly over the IP address?

You would get CORS issues on the livekit connection tester if you didn't use wss://. I think you'd have the same issue in your own setup. It might work for testing purposes but you'll need the domain name to use secure web sockets.

yes we already tried and e we get CORS issues using ws:// only and the IP

The RN version we are using is 0.67.3, if I download it to 0.55.4, can this be expected to work?

Hi, can you show me exactly which one you are talking about to be sure.

Capture d’écran 2022-08-16 à 20.57.09.png

This feature, which calls `sendUpdateTrackSettings`. I can see it also exists on the iOS engine but it's not public. Any other way to trigger it?

Thanks, I will consider exposing this in the Swift SDK also.

what do you mean by `I close the socket` ?

Hi, does it only happen with 1.0.3 from which version ?
Is it reproducible ?

&gt;  what do you mean by `I close the socket` ?
I create a proxy in my local machine to connect to my livekit server and connect into it like this:

```room.connect("<ws://192.168.1.132:1278>", ...```
And then I kill the proxy so the socket connection is lost

&gt;  Hi, does it only happen with 1.0.3 from which version ?
1.0.1

&gt;  Is it reproducible ?
It is on my app, I could check if it happens in the sample app (or a fresh project)

btw, it only happens if there is another participant in the room. If I am alone the logs are correct

```connection updated to connecting from disconnected(reason: nil)
connection updated to connected from connecting
connection updated to reconnecting from connected
connection updated to disconnected(reason: nil) from reconnecting```

after a few reconnect attempts it should `disconnect`

&gt; it only happens if there is another participant in the room
I will investigate this, thanks :slightly_smiling_face:

I created a fresh app that reproduces the issue

*Steps*

•  Install the app in 2 different devices.
•  connect one of the users to the room through a proxy, we do it this way:
```mitmproxy --listen-host 192.168.1.132 --listen-port 1278 -m "reverse:<https://livekit.ourdomain.com:443>"

# so the url to add in the url field is `<ws://192.168.1.132:1278>````
•  connect the other device into the same room with the normal url
• after a few seconds, kill the proxy
*What happens on 1.0.3? (and NOT in 1.0.1)*
• LiveKit doesn't report disconnected connection state
*Finally*
•  if when the app is stuck in "reconnecting" you tap on "Force Disconnect" the state changes to disconnected but if you tap "connect" again the `room.connect` promise is never resolved, neither the `then` or `catch` are called.


and remember, only happens if both devices are real devices, not simulators.

thanks for inv4estigating :slightly_smiling_face:

Thanks for the details, I will let you know if there is progress.

I can’t reproduce this at the moment.

```Did update connectionState disconnected(reason: nil) -&gt; connecting
Did update connectionState connecting -&gt; connected```
~ Disconnect network ~
```Did update connectionState connected -&gt; reconnecting
Did update connectionState reconnecting -&gt; disconnected(reason: nil)```
I will try your code now.

I can repro with your code, looking into it…

Ok if i comment out `setMicrophone(enabled: true)` all works well.
It might be some bug related to ~internal clean up~. -&gt; AudioManager internal thread.
Thanks for reporting !

I’m working on this issue in this PR <https://github.com/livekit/client-sdk-swift/pull/121>

do you know if there is a possible workaround? I guess not calling `setMicrophone(enbled:)` is. not an option...

You can try the fix-audiomanager-deadlock branch for now

and also, how long typically takes to include these changes in a release?

I might make more changes and do more testing. I will make a release after that.

if it's happening soon (like 1 week) I guess I can downgrade to 1.0.1 and wait

v1.0.4 is out now :slightly_smiling_face:

Would be happy to know if things are better now .

<@U027YH2JNB1> which version of Egress are you running?

I believe this is a bug fixed in <https://github.com/livekit/egress/pull/66|#66>. and released in `v1.1.0`

I run with docker run -d --rm -e EGRESS_CONFIG_FILE=/out/config.yaml -v /opt/livekit:/out livekit/egress

try running with `livekit/egress:v1.1.0` or
```docker pull livekit/egress```

"Labels": {
                "org.opencontainers.image.created": "2022-08-10T02:53:16.699Z",
                "org.opencontainers.image.description": "Export and record WebRTC sessions and tracks",
                "org.opencontainers.image.licenses": "Apache-2.0",
                "org.opencontainers.image.revision": "93a624c3e888d49ee93d965614e556143d82373a",
                "org.opencontainers.image.source": "<https://github.com/livekit/egress>",
                "org.opencontainers.image.title": "egress",
                "org.opencontainers.image.url": "<https://github.com/livekit/egress>",
                "org.opencontainers.image.version": "v1.1.1"
            }

ok, I think the fix might not have been in the last release. we'll keep you posted

<@U03SP6DMWUS>, a new build is out: <https://github.com/livekit/egress/releases/tag/v1.2.0|v1.2.0>. let me know if this fixes things for you

Great news. I will try. thank you very much for your interest.

I think no one expects to update Kotlin and SDK version when going from 1.1.1 to 1.1.2 :pray:

what issues have you run into when upgrading?

A lot. We use Compose which heavily relies on Kotlin version. To be able to use 1.1.2 we first need to update Kotlin to 1.7 then Compose to 1.2.

This change is also a little bit hidden inside <https://github.com/livekit/client-sdk-android/commit/e6c9d9b7054b304b2d9b87739da139b32a8a5e02> so it’s not possible to notice it by only reading commit messages.

bah, I'll watch out for that in the future

I have also tried the following, but it doesn't work either.
webhook:
   api_key: 'APIHzxDRo2qczNq'
   urls:
     - '<http://localhost:7000>'

hi liuky, how did you install livekit? is it from Docker?

that's the issue. Docker is running on its own IP, so `localhost` to docker is not where you are running your web server.

There are a couple of ways around this:
• run livekit-server without docker
• start docker with `--network host` - this requires linux

which platform are you testing on <@U03B7M4QPRP>?

okay, maybe the issue is related to the local renderer, I’ll check it

Just tested on mac, haven’t reproduced,  as david mentioned, which platform are you using? and the client SDK version is?

latest client sdk from pub.dev , version 1.1.0

what I think is that the issue is coming from room.localParticipant . when it's trying to render the video it's not working  but it's emitting the data to the server so when others are trying they can view it. And I've noticed another thing. When there are no participants except the host , it works perfectly. but when there is another user the video is not working.

Thanks Rey, could you file an issue on GH documenting the above? We are trying to get better at tracking these so they don't slip through the cracks.

<https://github.com/livekit/client-sdk-flutter/issues/154>
I created a related issue and commit a PR to fix the bug of showing frozen camera screen when local track video muted.
<https://github.com/livekit/client-sdk-flutter/pull/155>

As described in the issue, fixed frozen local camera image, when localtrack is muted, the expected camera muted icon should be displayed.

Hello <@U03B7M4QPRP>, I'm not sure if you did get the chance to look at the benchmarking page in the docs, but if you haven't maybe this would help you out. <https://docs.livekit.io/deploy/benchmark/#video-room> With the load-tester tools, you should be able to test it out.