LiveKit Community

A new set of key &amp; secret didn't help either. Some suggestions would be super. Starting to suspect issue might be related to GCP.

LiveKit supports H.264. I'm not sure how the code snippet you posted relates to that.

encoding/decoding capabilities depends on the device that you are building. Certain android devices don't ship with a H.264 encoder. Otherwise all WebRTC clients can en/decode H.264

please thread your messages, it would make it easier to follow.

The analysis of the core code of the broadcast rtp shows that the v8 encoding format is determined before the rtp, and H264/H265 is not supported.

The broadcast rtp determines whether the extPkt packet is VP8 to the room participant, but there is no H264 determination. I use H264 to send it, and the actual phenomenon is the same. Add the H264 determination (extPkt. Packet. Header. PayloadType==125), and then use d. writeStream WriteRTP (&amp;hdr, payload) can be sent successfully.
<https://github.com/livekit/livekit/blob/master/pkg/sfu/downtrack.go|downtrack-WriteRTP>
```// WriteRTP writes an RTP Packet to the DownTrack
func (d *DownTrack) WriteRTP(extPkt *buffer.ExtPacket, layer int32) error {
	var pool *[]byte
	defer func() {
		if pool != nil {
			PacketFactory.Put(pool)
			pool = nil
		}
	}()

	if !d.bound.Load() {
		return nil
	}

	tp, err := d.forwarder.GetTranslationParams(extPkt, layer)
	if tp.shouldDrop {
		if tp.isDroppingRelevant {
			d.pktsDropped.Inc()
		}
		if err != nil {
			d.logger.Errorw("write rtp packet failed", err)
		}
		return err
	}

	payload := extPkt.Packet.Payload
	if tp.vp8 != nil {
		incomingVP8, _ := extPkt.Payload.(buffer.VP8)
		pool = PacketFactory.Get().(*[]byte)
		payload, err = d.translateVP8PacketTo(extPkt.Packet, &amp;incomingVP8, tp.vp8.Header, pool)
		if err != nil {
			d.pktsDropped.Inc()
			d.logger.Errorw("write rtp packet failed", err)
			return err
		}
	}

	var meta *packetMeta
	if d.sequencer != nil {
		meta = d.sequencer.push(extPkt.Packet.SequenceNumber, tp.rtp.sequenceNumber, tp.rtp.timestamp, int8(layer))
		if meta != nil &amp;&amp; tp.vp8 != nil {
			meta.packVP8(tp.vp8.Header)
		}
	}

	hdr, err := d.getTranslatedRTPHeader(extPkt, tp)
	if err != nil {
		d.pktsDropped.Inc()
		d.logger.Errorw("write rtp packet failed", err)
		return err
	}

	if meta != nil &amp;&amp; d.dependencyDescriptorID != 0 {
		meta.ddBytes = hdr.GetExtension(uint8(d.dependencyDescriptorID))
	}

	_, err = d.writeStream.WriteRTP(hdr, payload)
	if err != nil {
		d.pktsDropped.Inc()
		if <http://errors.Is|errors.Is>(err, io.ErrClosedPipe) {
			writeIOErrors := d.writeIOErrors.Inc()
			if (writeIOErrors % 100) == 1 {
				d.logger.Errorw("write rtp packet failed", err, "count", writeIOErrors)
			}
		} else {
			d.logger.Errorw("write rtp packet failed", err)
		}
		return err
	}

	pktSize := hdr.MarshalSize() + len(payload)
	for _, f := range d.onPacketSent {
		f(d, pktSize)
	}

	if tp.isSwitchingToMaxLayer &amp;&amp; d.onMaxLayerChanged != nil &amp;&amp; d.kind == webrtc.RTPCodecTypeVideo {
		d.onMaxLayerChanged(d, layer)
	}

	if extPkt.KeyFrame || tp.switchingToTargetLayer {
		d.isNACKThrottled.Store(false)
		d.rtpStats.UpdateKeyFrame(1)

		locked, _ := d.forwarder.CheckSync()
		if locked {
			d.stopKeyFrameRequester()
		}

		if !tp.switchingToTargetLayer {
			d.logger.Debugw("forwarding key frame", "layer", layer)
		}
	}

	d.rtpStats.Update(hdr, len(payload), 0, time.Now().UnixNano())
	return nil
}```

the analysis is incorrect. it does support H.264. 

H.265 is a diff codec and is not supported

Hello, do you have time to analyze together (using goland)? My environment is all available. H264 does not support it at present, thank you!

image.png

Use the remote assistance tool of goland software to realize online debugging and analysis

sfubufferFrameMainStream2.h264

use <chrome://webrtc-internals/> The tool can see that the H264 data has been transferred to the subscriber, but the client does not decode successfully.

how is the video encoded? WebRTC spec dictates constrained mainline, and it doesn't support b-frames

I assure you LiveKit itself supports H.264 to WebRTC specs. to verify just open <http://example.livekit.io|example.livekit.io> with Safari and observe both the publisher and subscriber side.

In the VP8 mode used by the browser, the H264 actual test client cannot normally display video, capture and use <chrome://webrtc-internals/> Tool testing can verify

This H264 video stream can be pushed and played normally in P2P mode

You can use the test video I sent you to test (there is no B frame, only I frame and P frame)

there are several ways of validating H.264 support
1. set `enabled_codecs` to H.264 in livekit server config
2. set <https://docs.livekit.io/client-sdk-js/interfaces/TrackPublishOptions.html#videoCodec|videoCodec> to h264 when publishing from client
3. use Safari or iPhone as publishing clients, they both prefer H.264 over vp8 as default codec

livekit-react config：
```
&lt;LiveKitRoom
  url={url}
  token={token}
  onConnected={(room) =&gt; {
    setLogLevel('debug');
    onConnected(room, query);
    room.on(RoomEvent.ParticipantConnected, () =&gt; updateParticipantSize(room));
    room.on(RoomEvent.ParticipantDisconnected, () =&gt; onParticipantDisconnected(room));
    updateParticipantSize(room);
  }}
  roomOptions={{
    adaptiveStream: isSet(query, 'adaptiveStream'),
    dynacast: isSet(query, 'dynacast'),
    videoCaptureDefaults: {
      resolution: VideoPresets.h720.resolution,
    },
    publishDefaults:{
      videoCodec:   'h264'
    }
  }}
  onLeave={onLeave}
/&gt;```
The client still does not display the H264 screen?

livekit-server config：
```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 used C language to design a client (including publisher and subscriber) based on the <https://docs.livekit.io/internals/client-protocol|client-protocol>

web client change `videoCodec` H264 ，other participant In room publish video not display？

are you testing with two browsers? we cannot support your custom C client.

I'm sorry, but I'm not able to spend more time on this. We know H.264 is being used by a lot of LiveKit users and no one has reported issues with it. If you are trying to build something custom, that's fantastic! But we are not able to help with specific issues that you are encountering

function send opus user `d.writeStream.WriteRTP` function and send H264 not use `d.writeStream.WriteRTP`  How is  h264 video stream sended ？tks:pray:

every packet is written via that function, including H.264.

H264/H265 codes were used in the project, and they were tracked and evaluated for one month. can you help solve the problem.

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

or update it ? Also, i want to store list of all connected participants on the livekit server, is there any way to do this ?

if you read carefully:

&gt; With our server SDKs, you can update the metadata from your backend. LiveKit will broadcast metadata changes to all participants in the room. You can receive these updates by listening to the MetadataUpdated callback/event.

But I’m not sure if storing all connected participants in the metadata of localParticipant is a good idea :slightly_smiling_face:

you can use the room metadata to write all the info that are relevant to your product. You can pass there a json with all you want and manage it server side

Will appreciate follow up from core team
<https://github.com/livekit/client-sdk-flutter/issues/132>

Sorry for all the edits, just trying to clean it up as much as possible.

Seeing this in the helm's deployment.yaml:

```          env:
            - name: LIVEKIT_CONFIG
              valueFrom:
                configMapKeyRef:
                  name: {{ include "livekit-server.fullname" . }}
                  key: config.yaml
            {{- if and .Values.livekit.turn.enabled .Values.livekit.turn.tls_port (not .Values.livekit.turn.external_tls) }}
            - name: LIVEKIT_TURN_CERT
              value: /etc/lkcert/tls.crt
            - name: LIVEKIT_TURN_KEY
              value: /etc/lkcert/tls.key```
The tls.crt and tls.key should be inserted by default, right?

Ran the procedure again with a fresh cluster. Didn't do kubectl create secret tls. Yaml:

```replicaCount: 1

# Refer to <https://docs.livekit.io/deploy/kubernetes/> for instructions

livekit:
  rtc:
    use_external_ip: true
  redis:
    address: redis-host:6379
  keys:
    keyAPI: key
  turn:
    enabled: true
    domain: <http://staging.turn.example.com|staging.turn.example.com>
    tls_port: 3478
    secretName: turn-cert

loadBalancer:
  type: do
  # TLS certificate generated automatically with certmanager / letsencrypt
  # secretName is still required. certmanager will place the provisioned
  # certificate in that secret
  clusterIssuer: letsencrypt-prod
  tls:
    - hosts:
        - <http://staging.example.com|staging.example.com>
      secretName: cert

autoscaling:
  enabled: false
  minReplicas: 1
  maxReplicas: 5
  targetCPUUtilizationPercentage: 60
# resources are set assuming a 16 core instance
resources:
  limits:
    cpu: 15000m
    memory: 4096Mi
  requests:
    cpu: 14000m
    memory: 2048Mi```
And got:

```Unable to attach or mount volumes: unmounted volumes=[lkturncert], unattached volumes=[lkturncert kube-api-access-cdtrt]: timed out waiting for the condition

MountVolume.SetUp failed for volume "lkturncert" : secret "turn-cert" not found```

Ok! I think I cleared the turn-cert problem. "kubectl create secret tls" may have worked because I submitted a .pem instead of .crt

I can read the .key and .crt files in the directory!

Need to resolve health checking:

```Liveness probe failed: Get "<http://x.x.x.x:7880/>": dial tcp x.x.x.x:7880: connect: connection refused```
DO seemed to encourage doing this for a load balancer that isn't getting connection for health checks:

```kubectl patch svc livekit-livekit-server-turn -p '{"spec":{"externalTrafficPolicy":"Local"}}'```
That stopped health check warnings, but the container continued to fail and the loadbalancer isn't posting connection. Will keep hitting it.

Haven't seen health check fail in awhile. But, it's still hitting:

```BackOff
Back-off restarting failed container```
Can't find any other events or logs atm.

`"externalTrafficPolicy":"Local"` would mean livekit cannot connect out. that does not work.

if liveness probe is failing, you should see if livekit container is able to run successfully.

Got it, thank you, and its valid to set "externalTrafficPolicy":"Cluster"?

Narrowed it to problem with connecting to Redis.

Would be good to include that as a comment in the do-example.yaml as an option.

Found it from reading the helm PRs.

Without TLS, the kubernetes pod log might read:

unable to connect to redis: EOF

Digital Ocean advises use of redli --tls flag, but the helm chart doesn't seem setup to support redli or connection strings. Therefore, use_tls: true may be necessary, even if the managed redis is configured to trust the k8 cluster.

Local dev client connects to the cluster, lets goooooo :ballot_box_with_check::ballot_box_with_check::ballot_box_with_check:

Thank you <@U01KZ61GV5Z>, I was unsure about the externalTrafficPolicy config. DO and K8 docs made you want to go Local.

Ok, can confirm that at least for my DO deployment, redis username and password is required.

So if you are deploying on digitalocean with TURN, then you can try this config in your values.yaml:

```  redis:
    address: &lt;do-vps-private-host-address&gt;:&lt;do-assigned-redis-port&gt;
    username: &lt;redis-username&gt;
    password: &lt;redis-password&gt;
    use_tls: true```


Hi Faiz, LiveKit cannot run on fargate or other serverless environments. it requires machines with predictable IP and ports.

At the moment, there isn't a quick way to deploy Egress with livekit via VM or ECS. The recommended path is Kubernetes with our official helm charts.

Track egress is not transcoded, therefore cannot be output to RTMP. You should use a TrackComposite egress instead (it could be a composite without an audio track)

Hello <@U01KZ61GV5Z> thanks for the reply.

ECS Fargate does get the public IP address once it's in running state, and automatically attached with Target Group set with the ECS Cluster service. Well it's not actually a serverless, Fargate task will be in a running state until it crashed or manually stopped or scale down based on the ECS Service configuration.

Related with track engress, Can it be send to RTMP without transcoding?

We have existing infra to receive the RTMP and do the realtime transcoding (using nvenc) before output.

<@U0414095DAB> any plans on the roadmap so that livekit kan run on fargate. i was thinking exactly the same for future use, but dont want to use ec2 instance to host k8s

We do not have any plans to support Fargate. You are welcome to try it though. I just don't know enough about it to say if it'll work.Egress should work fine on most platforms.

RTMP is very specific about the types of media it could contain (H.264/aac), so it's not possible to offer it as a generic output type for track egress.

that means StopEgress was called before it ever started recording. If you’re using a custom template, make sure you’re calling `EgressHelper.startRecording();`

I thing something is happening, it occurred again. I stoped the recorder and this happened. Do you have any ideas? or another way to debug this?

is happening on every record I create, we updated to last egress version yesterday. Could be some regression issue?

The only way it can ever show `ABORTED` is if the pipeline is stopped before `EgressHelper.startRecording()` is called. Are you using a custom template?

It sounds like there might be an issue with your template calling the start function
<https://docs.livekit.io/guides/egress/custom-template/>

ok! I will give a look this week. Thanks <@U01UB0LHTTQ> !

[livekit] ping config timeout: 20, interval: 10
[livekit] failed to emit event TrackUnsubscribedEvent(partic…ance of ‘RemoteVideoTrack’) on a disposed emitter
[livekit] failed to emit event TrackUnpublishedEvent(partici…ource: TrackSource.camera)) on a disposed emitter
[livekit] [RemoteTrackPublication&lt;RemoteVideoTrack&gt;#452329978] unnecessary dispose() called.
[livekit] [LocalTrackPublication&lt;LocalVideoTrack&gt;#67175699] unnecessary dispose() called.
[livekit] [LocalTrackPublication&lt;LocalAudioTrack&gt;#200603131] unnecessary dispose() called.
[livekit] ping config timeout: 20, interval: 10
26
[livekit] onDisconnected state:ConnectionState.connected reason:signal
4
[livekit] onDisconnected state:ConnectionState.reconnecting reason:peerConnection
[livekit] failed to emit event TrackUnsubscribedEvent(partic…ance of ‘RemoteVideoTrack’) on a disposed emitter
[livekit] failed to emit event TrackUnpublishedEvent(partici…ource: TrackSource.camera)) on a disposed emitter
[livekit] [RemoteTrackPublication&lt;RemoteVideoTrack&gt;#452329978] unnecessary dispose() called.
[livekit] onDisconnected state:ConnectionState.disconnected reason:signal
[livekit] [LocalTrackPublication&lt;LocalVideoTrack&gt;#126666578] unnecessary dispose() called.
[livekit] [LocalTrackPublication&lt;LocalAudioTrack&gt;#33561891] unnecessary dispose() called.
[livekit] [Engine#736053726] subscriber is null
6
[livekit] Received SignalTrickleEvent but publisher or subscriber was null.
[livekit] [Engine#736053726] subscriber is null
6
[livekit] Received SignalTrickleEvent but publisher or subscriber was null.
[livekit] onDisconnected state:ConnectionState.disconnected reason:signal
[livekit] ping timeout
2
[livekit] [SignalClient#18159710] Could not send message, socket is null
[livekit] ping config timeout: 20, interval: 10

Bad state: No element
[livekit] ping config timeout: 20, interval: 10
[livekit] failed to emit event TrackUnsubscribedEvent(partic…ance of ‘RemoteVideoTrack’) on a disposed emitter
[livekit] failed to emit event TrackUnpublishedEvent(partici…ource: TrackSource.camera)) on a disposed emitter
[livekit] [RemoteTrackPublication&lt;RemoteVideoTrack&gt;#452329978] unnecessary dispose() called.
[livekit] [LocalTrackPublication&lt;LocalVideoTrack&gt;#67175699] unnecessary dispose() called.
[livekit] [LocalTrackPublication&lt;LocalAudioTrack&gt;#200603131] unnecessary dispose() called.
[livekit] ping config timeout: 20, interval: 10
26
[livekit] onDisconnected state:ConnectionState.connected reason:signal
4
[livekit] onDisconnected state:ConnectionState.reconnecting reason:peerConnection
[livekit] failed to emit event TrackUnsubscribedEvent(partic…ance of ‘RemoteVideoTrack’) on a disposed emitter
[livekit] failed to emit event TrackUnpublishedEvent(partici…ource: TrackSource.camera)) on a disposed emitter
[livekit] [RemoteTrackPublication&lt;RemoteVideoTrack&gt;#452329978] unnecessary dispose() called.
[livekit] onDisconnected state:ConnectionState.disconnected reason:signal
[livekit] [LocalTrackPublication&lt;LocalVideoTrack&gt;#126666578] unnecessary dispose() called.
[livekit] [LocalTrackPublication&lt;LocalAudioTrack&gt;#33561891] unnecessary dispose() called.
[livekit] [Engine#736053726] subscriber is null
6
[livekit] Received SignalTrickleEvent but publisher or subscriber was null.
[livekit] [Engine#736053726] subscriber is null
6
[livekit] Received SignalTrickleEvent but publisher or subscriber was null.
[livekit] onDisconnected state:ConnectionState.disconnected reason:signal
[livekit] ping timeout
2
[livekit] [SignalClient#18159710] Could not send message, socket is null
[livekit] ping config timeout: 20, interval: 10
[livekit] [RemoteTrackPublication&lt;RemoteVideoTrack&gt;#452329978] unnecessary dispose() called.
[livekit] [RemoteTrackPublication&lt;RemoteVideoTrack&gt;#132963956] unnecessary dispose() called.
[livekit] [RemoteTrackPublication&lt;RemoteAudioTrack&gt;#298176719] unnecessary dispose() called.
[livekit] [RemoteTrackPublication&lt;RemoteAudioTrack&gt;#505060599] unnecessary dispose() called.
[livekit] [RemoteTrackPublication&lt;RemoteVideoTrack&gt;#204045542] unnecessary dispose() called.
[livekit] [RemoteTrackPublication&lt;RemoteAudioTrack&gt;#385274488] unnecessary dispose() called.
[livekit] [RemoteTrackPublication&lt;RemoteAudioTrack&gt;#500653211] unnecessary dispose() called.
[livekit] [RemoteTrackPublication&lt;RemoteVideoTrack&gt;#468496007] unnecessary dispose() called.
[livekit] [RemoteTrackPublication&lt;RemoteVideoTrack&gt;#329926794] unnecessary dispose() called.

Can you share more details about what you are doing that triggered this?

Currently it doesn't work that way. Everyone connected to the same room must be connected to the same instance.

If you are looking to scale beyond what a single server can handle, take a look at <https://livekit.typeform.com/cloud-beta|Cloud preview>.

If the viewer watches the broadcast on a far server, will the delay of the broadcast increase?

I filled the form, waiting for a response.

maybe slightly, it typically wouldn't exceed 500ms even if on the other side of earth.

This seems like an issue with ports/configuration. I'm not sure k8s would be any different to be honest. the main difference in k8s is being able to handle multi-node / high availability

I have use the config generator from docs

docker run --rm -it -v$PWD:/output livekit/generate

Then copied them to ubuntu 18.x
Sudo .init_script.sh

Open all the port of micro.t3 instance hosted in aws baharain for it being close to me in dubai

The thing i was thinking which I will try is ulimit?

And does it make any difference for tedting 1 room with 2to 3 people?

ubuntu@ip-xxx-xx-xx-xxx:~$ ulimit -Ha
real-time non-blocking time  (microseconds, -R) unlimited
core file size              (blocks, -c) unlimited
data seg size               (kbytes, -d) unlimited
scheduling priority                 (-e) 0
file size                   (blocks, -f) unlimited
pending signals                     (-i) 3821
max locked memory           (kbytes, -l) 123888
max memory size             (kbytes, -m) unlimited
open files                          (-n) 1048576
pipe size                (512 bytes, -p) 8
POSIX message queues         (bytes, -q) 819200
real-time priority                  (-r) 0
stack size                  (kbytes, -s) unlimited
cpu time                   (seconds, -t) unlimited
max user processes                  (-u) 3821
virtual memory              (kbytes, -v) unlimited
file locks                          (-x) unlimited

Is your server located close to where you are? Can you take a screenshot of the `ICE candidate grid` while a connection is ongoing?

Yes, I am in dubai and using baharain instance which is nearby, Will try it on azure also which have local datacenter

what you are describing sounds like something is off. If you can capture a video of what you mean, along with `<chrome://webrtc-internals>` information, that might help us to see what's happening

Thanks mate will do come back with all the info thx

<@U01KZ61GV5Z> check out ivekit.turn	v2@v2.0.8/server.go:184	error when handling datagram: failed to handle Allocate-request from 94.204.235.150:59488: integrity

{“level”:“error”,“ts”:1663067087.5512438,“logger”:“layer4”,“msg”:“handling connection”,“error”:“dial tcp 172.31.45.228:5349: connect: connection timed out”}
{“level”:“error”,“ts”:1663067124.4152918,“logger”:“layer4",“msg”:“handling connection”,“error”:“dial tcp 172.31.45.228:5349: connect: connection timed out”}
{“level”:“error”,“ts”:1663067151.0393233,“logger”:“layer4”,“msg”:“handling connection”,“error”:“dial tcp 172.31.45.228:5349: connect: connection timed out”}
{“level”:“error”,“ts”:1663067794.1113098,“logger”:“layer4",“msg”:“handling connection”,“error”:“dial tcp 172.31.45.228:5349: connect: connection timed out”}

<@U01KZ61GV5Z> <https://github.com/pion/turn/issues/206>

Caddy also throwing connection timeout error in burst, not sure if the issue arising from livekit , or pion or reverse proxy

on azure I will deploy on k8s and use diff reverse proxy

and even configure to use different turn server to test

I am just trying to figure out if its my configuration or internet connection. Caz other platorms work well so might no be my internet connection.

Jaspreet, we are. Here’s the message I shared in <#C01KZ61HZRR|general> a while back if you’re interested:

<https://livekit-users.slack.com/archives/C01KZ61HZRR/p1658779036581289>

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

Make sure you have redis set up properly and in your egress and livekit server configs

Okay let me try,

FYI: LiveKit-server is running fine on local machine without any changes on redis.conf file

Yeah, it's because egress needs to be run with docker and won't connect to a localhost redis address

Livekit server can also run without using redis at all, but the two services communicate through redis

Screenshot 2022-09-17 at 12.26.51 am.png

here is config and running stats of services

Redis:
```bind 127.0.0.1 ::1
protected-mode no
port 6379
timeout 0
tcp-keepalive 300```
egress config:
```log_level: debug
api_key: devkey
api_secret: secret
ws_url: <ws://192.168.65.2:7880>
insecure: true
redis:
  address: 192.168.65.2:6379```
still same error:
```twirp error internal: egress not connected (redis required)```
Not sure, what i am doing wrong.

That error means your livekit server is not using redis

Awesome, Just added the redis-host into livekit-server and it works.

It's not recording the screensharing in the same room, I have started the recording with `"layout": "grid",`
Am i missing something or screenshare will not record?

ok, you are good on your end.. this had been a regression in our recording templates.. will fix

give it a try now :slightly_smiling_face:

the reason i'm asking this, i want to understand the data flow. if i have a 2 livekit servers A and B.  The loadbalancer lets the client connect with server A, but the room is in server B. So livekit directs the client to server B. However will then server B connect back to the client? or will the data flow go via server A? (as that server established the connection with the client and perhaps server B is not aware of the client connection details (like ip etc)) (or is this where redis helps out)

please keep in mind i'm still talking about the situation where the connection in the mentioned image goes from the client directly to the livekit server.

you can manually control the audio output by providing an AudioSwitchHandler to LiveKit like so:
<https://github.com/livekit/client-sdk-android/blob/47bd21b14206a9256efa3d16b333d6da4493db1a/sample-app-common/src/main/java/io/livekit/android/sample/CallViewModel.kt#L33|https://github.com/livekit/client-sdk-android/blob/47bd21b14206a9256efa3d16b333d6d[…]common/src/main/java/io/livekit/android/sample/CallViewModel.kt>

note by default, audio output is routed to the speaker

The new `VideoView.rotationOverride` property should fix/disable the rotation to your preference . This will be available soon for macOS also.

I'll have to add an observer to detect the device's orientation and force the videoView to rotate to the correct position... what I really wanted... was to disable rotation so that the user doesn't see the rotation to be corrected

however... I was just wondering if it was possible to disable the rotation... that solution should solve it but it's not the right thing, that's all

<@U03UD008S7R>, would that be something you'd set at the app level? I think when the app rotates, the video stack would have no choice but to follow that

<@U01KZ61GV5Z> sorry, you can be a little more clear?

On each UIViewController, you could have `shouldAutorotate` return false. Like in <https://www.appsdeveloperblog.com/disable-rotation-of-uiviewcontroller/|this example>

Can you share more about what doesn't work? What happens when you set the ViewController to not rotate? Does the video capture still rotate?

<@U01KZ61GV5Z> yes video capture still rotate

<@U01KZ61GV5Z> you can help? what's your advice about that?

<@U01KZ61GV5Z> I really need your help regarding this, it's causing a lot of bad user experience.

you can add a property to disable camera rotation? or something else?

<@U01KZ61GV5Z> when you can I ask you to respond, I apologize if I'm pressing but it's because it's urgent.

Hi <@U03UD008S7R> It's a bit tricky to control rotation info on the publisher side (because it's built-in to WebRTC), but I'm willing to improve it eventually (with additional improvements as client side recording).

What you can only do now is to  force VideoView's rotation by `rotationOverride` property .

Hi Prasad, can you be more specific about "joining audio"? Are you able to use our <https://example.livekit.io|example webapp> with your server?

Thank you, David! the node-pool was having improper network tags which caused this issue and I have fixed it, it works fine now! :slightly_smiling_face:

these are secure. they ports are listened to only when it's allocated to a particular client. content is transmitted encrypted via DTLS

Hi Team,
one more question is on integrating the livekit reverse proxy like oauth2-proxy. When I tried this integration we are getting the error as `invalid API key:`
Any idea if this integration can really happen?

please avoiding posting dupes. this is a community support channel and we'd do our best to help everyone. that said, we might not know how to help with a particular way that you are choosing to deploy / integrate with third party software

<@U01KZ61GV5Z> thank you for the response

<@U01KZ61GV5Z> What I really wanted to know was that, is it possible to integrate livekit with any SSO tools?

Definitely! LiveKit doesn't handle authentication by itself. Your backend should be issuing tokens with the API key &amp; secret after authenticating users

It should be supported by caddy: <https://caddyserver.com/docs/json/apps/tls/>

YAML and JSON follow similar structures.

hi Sai, please thread your questions. :pray:

My bad sorry!

Great package! Kudos to the efforts!

I'm using flutter as client. And my usecase is collab based podcasts.
So no video usage whatsoever.
How can I listen to audio only.
By default there is only video renderer widget.
If I need to play remote audio how do I do it and if multiple persons speak at the same time. How to play all those tracks in a synchronised way.
Thanks in advance.

thank you!

Audio is played automatically on some platforms, including Flutter! it's mentioned <https://docs.livekit.io/guides/room/receive/#media-playback|here> in the docs.

Thanks for mentioning it again. Yeah I did see it in pub documentation under audio handling. But for some reason I'm not able to hear them!

Example app works perfectly. And isSpeaking reacts perfectly. But still not able to hear the track from the published phone.

Track published in one device is not received by subscribers. The device which is publishing while giving grant in server side I'm giving it as canPublish: true &amp; canSubscribe: true but for all other listeners I'm doing it as  canPublish: false &amp; canSubscribe: true. So this is how I'm generating tokens for the users who are connecting to the room. Then I'm trying to publish audio manually by doing setMicrophoneEnabled(true) and calling publishAudioTrack with created local audio track as param

Now in the publisher side when speaking, I'm getting isSpeaking as true in participants props. But not able to hear the audio thats been published.

Publishing audio from Mi Note 9 and receiving data in iPhone 11

Example app works wonders. Because the video renderer widget is present there,

If i have to manually play audio via some plugin like just_audio, how to do it so? :pray:

Excuse my silly question. Why there is not an audio renderer/player like video renderer?

hey Sai, not a silly question at all. WebRTC libraries on mobile devices automatically decode incoming audio for playback. They don't give the user precise control over it.

If you are having issues, can you upload an example/code snippet how one could reproduce the issue? Using the example app, even if video is not published, audio still plays back just fine in our testing.

Thanks for the explanation. I was able to do it once listener was registered with rebuild triggers.

We have hosted LiveKit server in AWS with c5.xlarge as instance type. Our use-case is audio only like clubhouse. Kindly share an approximate ball park figure of how many rooms we can host simultaneously and how many users can each room accommodate.?

And I'm updating livekit server by this command `docker pull livekit/livekit-server`

It pulls latest image. But is there any other process after this. because when I'm logging in flutter client it shows server version as 1.2.0, where as the latest should be 1.2.1

And how to add metadata so that we can retrieve in participant props.

I must admit that this is a well thought project and stuff just works , it just works out of the box. Amazingly done! Flutter client is rich with features! Thanks again for all your efforts.

<@U03KYURETMF> yup! clients will attempt to reconnect if a connection had been interrupted. Other instances can typically detect the failure within 5s, and will allocate the session to another instance.

it takes json marshalled into bytes
definition: <https://github.com/livekit/protocol/blob/main/livekit_egress.proto#L118>
usage: <https://github.com/livekit/egress/blob/1f3e4b9c6dd7254947c167f1e9a03d6c354e065d/pkg/pipeline/sink/upload.go#L120>

Thanks <@U01UB0LHTTQ>. Will give it a bash.

Screenshot 2022-09-22 at 22.04.49.png

Hunting through the source dead ends at the proto folder.

Hello guys. Please help me with these queries. Thanks in advance :pray: :blush:

in terms of load testing, it'd be good to perform your own. We cannot tell you exactly how it might match your use case.

with docker, you should use a specific tag during deploy to ensure it pulls exactly that version.

re: metadata, please refer to the <https://docs.livekit.io/guides/room/data/#participant-metadata|docs>

Hmm yeah makes sense. I'll simulate it via load testing.

Can you please tell what's that specific tag please. I'm not sure what does that mean. Excuse me for my ignorance.

Great! I'll check out the docs for metadata. Somehow I missed that in the documentation.

Thanks a lot for your time <@U01KZ61GV5Z>

I installed livekit server on my local machine /using this command livekit-server --dev/