LiveKit Community

that's possible you are looking at different `room` instances. Try disabling StrictMode in react to see if that goes away.

FWIW, our react components should work fine with strict mode. Are you using those or is this custom?

Hi <@U01KZ61GV5Z>… Yes, I’m using the `&lt;LiveKitRoom&gt;` component and I’ve debugged it thinking maybe is something from me but also on the connected function that returns the room u’ve been connected `onConnected?: (room: Room) =&gt; void;` manually disabling camera permissions or microphone not working as expected having them both turned off just giving me if I remember right only one error not for both

```What I did for solving this is below:

room.on(RoomEvent.MediaDevicesError, async () =&gt; {
  const browser = getBrowser();
  if (browser !== "Firefox" &amp;&amp; browser !== "Safari") {
    const queryPromises = ["camera", "microphone"].map((name: any) =&gt;
      navigator.permissions.query({
        name: name,
      })
    );
    for await (const status of queryPromises) {
      const perm: any = status;
      if (perm.name === "video_capture" &amp;&amp; perm.state === "denied") {
        setPermissions((prev) =&gt; ({
          ...prev,
          camera: false,
        }));
      }

      if (perm.name === "audio_capture" &amp;&amp; perm.state === "denied") {
        setPermissions((prev) =&gt; ({
          ...prev,
          audio: false,
        }));
      }
    }
    setPermissionsDenied(true);
  }
});```

`PS: For me looks like a problem with re-rendeing in React`

An option for other that may have this problem is working with `RoomEvent_._MediaDevicesError` but seems also to not working as expected.

Now I am aware that Case 2 isn't a good practice and I wouldn't do that it's just interesting that there is a difference in the behavior of the 2 cases.

My expectation in both of these cases would be:
• livekit registers an `end` listener to the track
• if the event is emitted while the track is published
• emit the `mute` event both locally and remotely

From previous experience with users who would create tracks manually, we decided that the safe bet is to leave the mediastreamtrack lifecycle to the integrating user.
This means livekit will do as little as possible to that mediastreamtrack - including not emitting certain events that would indicate that LiveKit is in control of the track. Have you tried if setting the `userProvided` boolean to false helps in your case? <https://docs.livekit.io/client-sdk-js/classes/LocalVideoTrack.html#constructor>

Regarding your observation between revoking permissions and calling track.stop() manually - that’s indeed interesting. Can you open an issue on github about it? As you said, we wouldn’t expect users to call the stop method manually in most cases, but it’s still quite surprising.

&gt; Can you open an issue on github about it?
Will do
&gt; Have you tried if setting the `userProvided` boolean to false helps in your case?
When using `LocalParticipant.publishTrack()` the `userProvided` flag is automatically set to `true`
I can set it to false when calling `replaceTrack` later for a test but generally I don't want to do this as this will try to access UserMedia later on.

&gt; From previous experience with users who would create tracks manually, we decided that the safe bet is to leave the mediastreamtrack lifecycle to the integrating user.
I partially agree, however I do think it's correct to mute the track on the  `end`event because there is no way this track will be valid again

&gt;  I partially agree, however I do think it’s correct to mute the track on the  `end`event because there is no way this track will be valid again
yeah, that sounds reasonable. I’ll take a look whether changing it to additionally mute the track would potentially break things.
For a workaround until then, you could listen to `ParticipantEvent.UpstreamPaused` which will get emitted on `ended` also for user provided tracks.

Yes what I'm seeing now is that when I revoke the camera rights the `localTrackPublication.isUpstreamPaused`  will go to `true` however if I manually stop the track it will stay `false` thanks for the event hint, I didn't see it right away as it's @internal

hmm, is the `TrackEvent.Ended` fired when you manually stop the track? or the `ended` event on the mediastreamtrack itself?

aha, mdn says:
&gt;  Immediately after calling `stop()`, the `readyState` property is set to `ended`. Note that the `ended` event will not be fired in this situation.

yes I just realized the same thing, the event isn't triggered

thats pretty bad / inconsistent behavior I'd say

but in the end this isn't really tragic for me as I'd never call `stop`  on a track that should be published

So I'd say the only change that would be needed is muting the local track when the `ended` event fires and the provider of the track must make sure the track is not `stopped` manually as long as it should be actively streamed.

here's the issue:
<https://github.com/livekit/client-sdk-js/issues/496>

Hi <@U048DMRDREZ> when you get a chance could you try to log in via github again? Please let us know when you do, so we can check the logs :pray:

Which SDK is this? can you share code snippets and/or which API you are calling into?

i’m using the {LiveKitRoom} _from_ ‘@livekit/react-components’;

Hope makes sense the code snippet … I’ve tried to keep just the relevant parts for checking this issue :)
```const [roomParticipants, setRoomParticipants] = useState&lt;
    |
    {
        participants: Map&lt;string,
            RemoteParticipant&gt;;
    } |
    undefined&gt;
    (); 
 

&lt;LiveKitRoom
    url={process.env.REACT_APP_LIVEKIT_WS_URL as string}
    token={token}
    onConnected={room =&gt; {
        setLogLevel('debug');
        onConnected(room);
        room.on(RoomEvent.ParticipantConnected, () =&gt; {
            updateParticipantSize(room);
        });
        room.on(RoomEvent.ParticipantDisconnected, () =&gt;
            updateParticipantSize(room),
        );
        updateParticipantSize(room);
        setRoomCall(room);
    }}
    connectOptions={{}}
    roomOptions={{
        publishDefaults: {
            simulcast: true,
        },
        videoCaptureDefaults: {
            resolution: VideoPresets.h720.resolution,
        },
    }}
    onLeave={() =&gt; history.push('/')}
/&gt;;

Array.from(roomParticipants?.participants ?? []).length &gt; 0 &amp;&amp;
    Array.from(
        room?.participants ?? [],
        ([key, value]: [
            key: any,
            value: RemoteParticipant,
        ]) =&gt; {
            const peer = JSON.parse(
                value.metadata as string,
            );
            if (value.permissions?.hidden) return null;

            return (&lt;
                Participant peer={
                    peer
                }
                value={
                    value
                }
            /&gt;
            );
    },
)



const Particiant = ({value}) =&gt; {
    const {
        isSpeaking,
        microphonePublication
    } = useParticipant(value);

    console.log({
        isSpeaking,
        microphonePublication
    })

}```

the &lt;Participant /&gt; components is rendered for each room paticipant and RemoteParticipant and using the useParticipant() hooked I hoped to achive the behavior seeing a RemoteParticipant toggles the microphone

image.png

<@U01KZ61GV5Z> what i wanna achive is the same behavior you guys have here

damn dummy me :))
i just had to not try to use `microphonePublication` and instead use `value.isMicrophoneEnabled` and works as expected

I will leave a this thread as it is now <@U01KZ61GV5Z>, if is ok for you, if others may come searching for an answer how to do this.

glad you've figured it out! yes of course, let's leave it here for others.

I believe you are looking for room name.
<https://docs.livekit.io/client-sdk-js/classes/Room.html#name>

Thank you for your quick reply :slightly_smiling_face:  But unfortunately I am not looking for the “name” attribute.

I’m interested if there is a way to find out which room or rooms I am currently connected to without having to store them locally.

<@U0491G6RB5F> would you say a little more about your use case/need for this feature?

Here is a bit of background information:
I'm looking to eventually migrate my service <https://www.thing.online/|https://www.thing.online/> to LiveKit in the medium term. "Thing" is a browser based software to run virtual workshops. A room we call meetingspace and you can, among many other features, enter very lightweight conversations with others.

Before a potential migration I want to find out what works and how. In the course of a first test I also tried to connect several rooms/meetingspaces and the question came up if I can get the information with which rooms I am currently connected to without having to remember the created room with "new Room(...)".
The idea to have multiple rooms is also useful for performance/network reasons if you have very full meetings/rooms to divide the area into several "subrooms".

Of course I can save all created room objects in an array or similar. My question is really more out of interest.
Does the LiveKit client know which rooms it is connected to and can I get the information or do I HAVE to save the created objects?

Ah and btw thanks for implementing such a cool and really good product :+1: and of course the nice support here :slightly_smiling_face:

<@U0491G6RB5F> gotcha, I understand this better now. There isn’t support for this (yet) from the client side. I’ll let <@U02EU2ZPTQU> weigh in on that.

From one of our server SDKs, there are a number of room-management APIs:
<https://docs.livekit.io/server/room-management/#roomservice-apis|https://docs.livekit.io/server/room-management/#roomservice-apis>

But, it looks like you can’t filter the List endpoint by participant ID. So not super helpful at the moment for what you want.

Thank you very much for the quick reply <@U01KZ0SCDKL> :slightly_smiling_face:

My question was only about the client, but it is interesting to know that neither the server nor the client knows this.
But maybe <@U02EU2ZPTQU> can bring more light into the darkness.

Heh, to be clear, the server knows it, but doesn’t expose it yet. :upside_down_face: 

Reference from <@U041NNV5U8N> was the one you're looking for: <https://docs.livekit.io/client-sdk-js/classes/Room.html#name>
In livekit room id is treated as name. During create room you'll require to set unique room id which is room name in livekit. For setting up room title or other information you'll require to use metadata. At least that is what we're using in plugNmeet.

I think your client would need to store references to all the rooms that you are connected to. Our client tries to not store global values because it'd be difficult to clean up.

are you doing intensive work in the onFrame callback? doing long tasks in the callback could cause frame drops

Controlling the size of the parent div and setting height="100%" and width="100%" at least prevents overflow, but would be really good to control aspect ratio.

are you including the react-aspect-ratio css? <https://github.com/livekit/livekit-react/tree/master/packages/components|https://github.com/livekit/livekit-react/tree/master/packages/components>

Duh, I did not. Incorrectly assumed the component was importing it within its own dependencies. Thank you, sir.

fixed by reverting to tag 1.4.5 fyi. I think perhaps digital ocean doesn't support the new params added since

ah interesting.. I think we can work around it by not populating these new fields unless they are explicitly set. I'll file an issue for this

<https://github.com/livekit/egress/issues/161>

please thread your logs and config files :pray:
it makes it difficult to navigate your message.

The main error seems to be here in the Egress

```{"level":"debug","ts":1667234127.2074044,"logger":"egress","caller":"pipeline/pipeline.go:350","msg":"[pulsesrc0] LATENCY - Element's latency has changed","nodeID":"NE_Pzf47TKa6S4H","egressID":"EG_E9nwTgMaNmQX"}
{"level":"debug","ts":1667234127.207729,"logger":"egress","caller":"pipeline/pipeline.go:350","msg":"[x264enc0] LATENCY - Element's latency has changed","nodeID":"NE_Pzf47TKa6S4H","egressID":"EG_E9nwTgMaNmQX"}
{"level":"debug","ts":1667234127.2210877,"logger":"egress","caller":"web/source.go:160","msg":"chrome log: END_RECORDING","nodeID":"NE_Pzf47TKa6S4H","egressID":"EG_E9nwTgMaNmQX"}
{"level":"info","ts":1667234127.2211642,"logger":"egress","caller":"service/handler.go:128","msg":"egress updated","nodeID":"NE_Pzf47TKa6S4H","egressID":"EG_E9nwTgMaNmQX","status":"EGRESS_ENDING"}
{"level":"debug","ts":1667234127.222005,"logger":"egress","caller":"pipeline/pipeline.go:453","msg":"sending EOS to pipeline","nodeID":"NE_Pzf47TKa6S4H","egressID":"EG_E9nwTgMaNmQX"}
{"level":"error","ts":1667234157.2221851,"logger":"egress","caller":"pipeline/pipeline.go:455","msg":"pipeline frozen","nodeID":"NE_Pzf47TKa6S4H","egressID":"EG_E9nwTgMaNmQX","stacktrace":"<http://github.com/livekit/egres|github.com/livekit/egres>
s/pkg/pipeline.(*Pipeline).SendEOS.func1.1.1\n\t/workspace/pkg/pipeline/pipeline.go:455"}```

can you share the full request you are making?

I was able to solve it. I had to make a `ClusterIP` service connecting port 7880 on the livekit/server service to the egress service and pointing the web socket url on egress to the DNS of the livekit server. It now connects successfully and I can get a room composit stream sent over rtmp. However currently its just sending a black screen with no audio or video and this is my next challenge.

(I think here is the right place to ask)
Am I getting this correct: If I want to build a chatroom (video/audio) with livekit and my envirenment only has php, I need an additional server for the livekit core server?

not really , but i wouldn’t advise running it like this in production :wink:

<@U0423DP3S6M> why? what do I exactly need? frontend(js...) and php is not enough?

I mean webrtc just needs a TURN server (I could use those from google) and the rest is p2p .. so php is not really needed right? (just for the mediation?)

You can run the core server locally in principle. But probably your web app will become slow. So you do need a separate server.

I use the PHP SDK and it is fine for interacting with LiveKit. But you need the core server for LiveKit. Without the core server you have no LiveKit.

<@U03M86BE4H0> thx. I thought webrtc is p2p, means no server is needed? just a server for turning and signaling (websocket to notify the server when a new client is there to exchange the connection string..) so what is this livekit core server exactly good for?

It is a SFU - it forwards video etc. between the peers. Nowadays you really do not work p2p under the hood.

<@U03M86BE4H0> why is that so? everything I read is about webrtc and p2p.. the whole idea is lost when you have again a middle man ... I am disappointed

Performance and firewalls are the main culprits

isn't it so that even with firewalls you can create a connection string that works well for p2p?

This is it - if you have 10 people, with the SFU they each have a connection to the SFU to upload and download their data, so each person sends the data once. To do p2p, each person would need a connection to every other person, so you are sending 9x the amount of data

I believe the second explanation is the reason you need MCU or SFU.

it will come back down after awhile. the OS doesn't always reclaim memory immediately.

which platform are you having this issue on?

the issue is on the web sdk am using in an Angular project

I do not manage this from my side, file names (and dates) are generated from egress service side,
the problem is only with {time}

The problem is that the character cannot be used with the s3 file name, so I suggested a way to solve the problem.
URL encoding seems to be a way to upload files to s3 including such characters, and other methods include removing the character in question or changing the file name to a value allowed by s3 file name.
*sorry for my poor English.*

I've opened an issue for this. we'll fix this shortly: <https://github.com/livekit/egress/issues/164>

can you share what your network configuration is? where is the client and where is the server?

are you forwarding port 7881 (TCP) correctly on the firewall?

The server is in the external network and the client is in the local area network

The configuration information is as follows:
```port: 80
rtc:
    udp_port: 7882
    tcp_port: 7881
    use_external_ip: false
keys:
    devkey: secret
logging:
    json: true
    level: debug

prometheus_port: 6789

room:
   enabled_codecs:
     - mime: audio/opus
     - mime: video/vp8
     - mime: video/vp9
     - mime: video/h264```


I set up a live kit server locally and it can run normally.
Then modify the use mode of the client：
```ws://（xx.xx.xx.xx）use_ external_ ip/rtc? access_ token=eyJhbGciOiJIUzI1NiIsInR5cCI6IkpXVCJ9.eyJleHAiOjE2Njg2NDU5ODQsImlzcyI6ImRldmtleSIsIm5hbWUiOiJpcGNhbWVyYSIsIm5iZiI6MTY2NjA1Mzk4NCwic3ViIjoiaXBjYW1lcmEiLCJ2aWRlbyI6eyJyb29tIjoiYWRtaW4iLCJyb29tSm9pbiI6dHJ1ZX19.VBEYactr6rzV175mqi2xeG7fy01YjHQ1qdYerOhYalw&amp;sdk=js&amp;version=1.2.9&amp;protocol=8&amp;adaptive_ stream=1```
Tracing the problem found that the candidate type is host (but the server)。
:pray:

local livekit server candidate :
```"candidate":"candidate:431420294 1 tcp 2124414975 192.168.0.106 7881 typ host tcptype passive","sdpMid":"","sdpMLineIndex":0,"usernameFragment":null}
"candidate":"candidate:2756018790 1 udp 2130706431 192.168.0.106 7882 typ host","sdpMid":"","sdpMLineIndex":0,"usernameFragment":null}```
remoter livekit server candidate :
```"candidate":"candidate:2562143777 1 udp 2130706431 xx.23.75.180 55412 typ host","sdpMid":"","sdpMLineIndex":0,"usernameFragment":null}
"candidate":"candidate:625231809 1 tcp 2124414975 xx.23.75.180 7881 typ host tcptype passive","sdpMid":"","sdpMLineIndex":0,"usernameFragment":null}```

not currently. It's configured for all rooms at the moment

Currently, I haven't found an option to have this enabled per track. I'm pretty sure its in room level only.

I’m not sure if this is your exact problem, but you can disable simulcast for screen share only by providing an empty array for the `screenShareSimulcastLayers` in the room options’. default publish options.

To have better sceenshare view, I want the screen share to be static resolution. Is there anyway to set it?