LiveKit Community

unfortunately, jitpack has some issues and older artifacts got deleted. please update livekit to the latest version, which will have a working artifact dependency, or manually exclude the `com.github.webrtc-sdk:android`  dependency and depend on `io.github.webrtc-sdk:android:104.5112.07`  instead

more information here: <https://livekit-users.slack.com/archives/C01KZ61HZRR/p1678845047639649?thread_ts=1677614895.016259&amp;cid=C01KZ61HZRR>

I don't know which channel to turn to for help.sorry

image.png

I believe you can enter any string as the secretName. cert manager has some magic that will generate it for you.

after that, you'd want to point to `<wss://mydomain>`, it'll be running on port 443

<@U01KZ61GV5Z> Any suggestion is highly appreciated

You can set the resolution and bitrate in the publish options. You can also change the default publish options in the RoomOptions passed when creating a Room object.

docs here: <https://docs.livekit.io/client/publish/#advanced-track-management>

are you using `{exact: deviceId}` as a constraint?

that should force firefox to use the one specified :slightly_smiling_face:

Found out that we had wrong constraints set actually. Thanks for the help anyway! :pray:

I am trying to record via cli and get "clock problem." error.
```2023-03-22T08:09:17.045Z	ERROR	egress	pipeline/watch.go:88	GStreamer error: clock problem.	{"nodeID": "NE_Ex3DEeCgrBE3", "handlerID": "EGH_DuGcijufhQ6U", "clusterID": "", "egressID": "EG_BQSndNpV8XcN", "element": "GstMP4Mux", "error": "Impossible to configure latency: max 0:00:01.000000000 &lt; min 0:00:01.033333334. Add queues or other buffering elements."}
<http://github.com/livekit/egress/pkg/pipeline.(*Pipeline).handleMessageWarning|github.com/livekit/egress/pkg/pipeline.(*Pipeline).handleMessageWarning>
	/workspace/pkg/pipeline/watch.go:88
<http://github.com/livekit/egress/pkg/pipeline.(*Pipeline).messageWatch|github.com/livekit/egress/pkg/pipeline.(*Pipeline).messageWatch>
	/workspace/pkg/pipeline/watch.go:43
<http://github.com/tinyzimmer/go-gst/gst.goBusFunc|github.com/tinyzimmer/go-gst/gst.goBusFunc>
	/root/go/pkg/mod/github.com/livekit/go-gst@v0.2.34-0.20230210170313-8fc9f59623d4/gst/cgo_exports.go:128
_cgoexp_eb73c21f527d_goBusFunc
	_cgo_gotypes.go:17388
runtime.cgocallbackg1
	/usr/lib/go-1.18/src/runtime/cgocall.go:314
runtime.cgocallbackg
	/usr/lib/go-1.18/src/runtime/cgocall.go:233
runtime.cgocallback
	/usr/lib/go-1.18/src/runtime/asm_amd64.s:971
runtime.systemstack_switch
	/usr/lib/go-1.18/src/runtime/asm_amd64.s:436
runtime.cgocall
	/usr/lib/go-1.18/src/runtime/cgocall.go:167
<http://github.com/tinyzimmer/go-glib/glib._Cfunc_g_main_loop_run|github.com/tinyzimmer/go-glib/glib._Cfunc_g_main_loop_run>
	_cgo_gotypes.go:2822
<http://github.com/tinyzimmer/go-glib/glib.(*MainLoop).Run.func1|github.com/tinyzimmer/go-glib/glib.(*MainLoop).Run.func1>
	/root/go/pkg/mod/github.com/livekit/go-glib@v0.0.0-20230223001336-834490045522/glib/gmainloop.go:53
<http://github.com/tinyzimmer/go-glib/glib.(*MainLoop).Run|github.com/tinyzimmer/go-glib/glib.(*MainLoop).Run>
	/root/go/pkg/mod/github.com/livekit/go-glib@v0.0.0-20230223001336-834490045522/glib/gmainloop.go:53
<http://github.com/livekit/egress/pkg/pipeline.(*Pipeline).Run|github.com/livekit/egress/pkg/pipeline.(*Pipeline).Run>
	/workspace/pkg/pipeline/pipeline.go:205
<http://github.com/livekit/egress/pkg/service.(*HandlerV0).Run.func2|github.com/livekit/egress/pkg/service.(*HandlerV0).Run.func2>
	/workspace/pkg/service/handler_deprecated.go:85
2023-03-22T08:09:17.114Z	WARN	egress	service/handler_deprecated.go:145	egress failed	{"nodeID": "NE_Ex3DEeCgrBE3", "handlerID": "EGH_DuGcijufhQ6U", "clusterID": "", "egressID": "EG_BQSndNpV8XcN", "egressID": "EG_BQSndNpV8XcN", "request_type": "room_composite", "output_type": "file", "error": "GStreamer error: clock problem."}```

Have you checked out <https://docs.livekit.io/server/egress/autoegress/|auto egress>?

<@U041KCGQCGZ> can you post the Egress version, as well as the RoomCompositeRequest you used?

Unrelated, I recommend adding `egress: use_psrpc: true` in your LiveKit server config to upgrade the internal communication system (<https://github.com/livekit/livekit/blob/master/pkg/config/config.go#L253>)

<@U01UB0LHTTQ> I installed Egress yesterday via helm. I think it's version is latest.

<@U01KZ61GV5Z> I am using flutter client, I get token from backend and connecting to room. I don't know how to use CreateRoom via curl.

you can use createRoom API from your backend too.

<@U01KZ61GV5Z> that makes sense. I will try. How can I save videos to local on kubernetes? FTP option would be nice.

<@U01UB0LHTTQ> my request.json :
```{
    "room_name": "788",
    "layout": "grid",
    "file": {
        "file_type": "MP4",
        "filepath": "livekit-demo/my-room-test.mp4",
        "s3": {
            "access_key": "myawskey",
            "secret": "aws-access-secret",
            "region": "aws-region",
            "bucket": "my-bucket"
        }
    }
}```
livekit-cli start-room-composite-egress --request request.json

hah.. I don't think we'll want to add FTP :sweat_smile:

a S3-compatible bucket is probably the best way to go here.

I have too much free space on my server. so I didn't want to pay AWS :smile:

Hi guys, I also ran into this issue and struggled for a few days. I tried egress 1.5.6 and 1.6.0, same issue.  Looking forward to the fix. thanks. here is my egress RoomComposite request
```{
    "room_name": "my-first-room",
    "layout": "speaker-dark",
    "file": {
        "filepath": "livekit/",
        "s3": {
            "access_key": "access_key",
            "secret": "secret",
            "region": "us-west-1",
            "bucket": "mybucket"
        }
    }
}```

just released 1.6.1 to docker and helm, let me know if you still see any issues!

Hi <@U01UB0LHTTQ> thanks for the update. however it is the same issue once using the 1.6.1

<@U01UB0LHTTQ> same error. I have add egress:use_psrpc:true in my helm config file but I am still getting this warning "Using deprecated egress client. Upgrade livekit-server to v1.3.5+ and add egress:use_psrpc:true to your livekit config".

Could it be that the Livekit cannot read this value in the config file?

which version of livekit-server are you running? do you have this field set?: <https://github.com/livekit/livekit/blob/master/config-sample.yaml#L218>

rtcse.yaml

I’m unable to reproduce, but my hunch is that the audio and/or video input is a bit slower on some machines which pushes it over the minimum latency. I’m releasing v1.6.2 (takes about an hour), which should fix it if that’s the case

released on docker and helm, let me know if that fixes it for you

there my configs:
js:
```roomOptions:{
    // adaptiveStream: true,
    adaptiveStream: {
        pauseVideoInBackground: false,
        pixelDensity: 'screen',
    },
    publishDefaults: {
        simulcast: true,
        videoSimulcastLayers: [
            new VideoPreset(160, 90, 30_000, 1),
            new VideoPreset(384, 216, 240_000, 25)
        ],
        videoEncoding: {
            maxBitrate: '600_000',
            maxFramerate: 25,
        },
        screenShareSimulcastLayers: [
            ScreenSharePresets.h720fps5
        ]
    },
    dynacast: true,
    videoCaptureDefaults: {
        resolution: VideoPresets.h540.resolution
    },
},```
and server config:
port: 7880
redis:
  address: localhost:6379

rtc:
  port_range_start: 10000
  port_range_end: 19999
  tcp_port: 7881
  use_external_ip: false
  allow_tcp_fallback: true
  use_ice_lite: true
  congestion_control:
     enabled: true
     allow_pause: true
  pli_throttle:
     low_quality: 500ms
     mid_quality: 1s
     high_quality: 2s
keys:
  APIbmc: *****************
room:
  empty_timeout: 300
  enabled_codecs:
    - mime: audio/opus
    - mime: video/vp8
    - mime: video/vp9
    - mime: video/h264
  #   # the API key to use in order to sign the message
  api_key: 'APIbmc'
  urls:
    - <https://example.com/api/*****/conf_webhook>

turn:
  enabled: true
  udp_port: 3478
  tls_port: 5349
  relay_range_start: 20000
  relay_range_end: 29999
  external_tls: true
  domain: <http://turn.example.com|turn.example.com>

that event is sent during congestion control events where there is insufficient bandwidth to receive the subscribed video. (server will try to send in a lower bitrate stream first, but then eventually pause the video if it still couldn't deliver it).

you can try setting `allow_pause: false`, which will prevent stream pausing. but the downside is that you'll likely experience video freezes/audio breaking out

<@U01KZ61GV5Z> One more question, "server will try to send in a lower bitrate stream first" - server try to send only problem stream or lower bitrate will be for all streams?

I'm not sure what you mean by problem stream.. but the challenge on the server side is if we know the user can only receive 500kbps, but there are 2mbps of video and audio to send to them, we have to make a decision on which streams to send. our algorithm will try to reduce quality of all videos to be under 500kbps, but if it still doesn't fit, it'll start pausing videos (when `allow_pause` is true)

Thanks for your explanation. Problem stream in my mind was the most heaviest stream in bitrate

yup, it'll start with the stream with the heaviest bandwidth usage first

Can i control the time beetwen attends to unpause stream? Or use some methods for manual attemp unpause?

We started getting reports of this today, I believe it’s already fixed on the main branch. Will confirm and release a new version tomorrow

Hopefully it will be fixed by tomorrow. Thank you for your kind response. <@U01UB0LHTTQ>

Just pushed v1.6.1 to docker and helm, let me know if you see the issue again!

I downloaded the current 1.61 version source code, built it as a dockerfile for egress and redistributed it, but I got the same error as below. I'm attaching the dockerfile I used just in case.


FROM ubuntu:22.04 AS template

WORKDIR /workspace

# install go
RUN apt-get update &amp;&amp; apt-get install -y yarnpkg

# copy templates
COPY template-default/ .

# build
RUN yarnpkg install
RUN yarnpkg lint
RUN yarnpkg build

# template 관련해서 multistage 0 build

FROM livekit/gstreamer:1.22.0-dev AS livekit

ARG TARGETPLATFORM

WORKDIR /workspace

# install go
RUN apt-get update &amp;&amp; apt-get install -y golang

# download go modules
COPY go.mod .
COPY go.sum .
RUN go mod download

# copy source
COPY cmd/ cmd/
COPY pkg/ pkg/
COPY version/ version/

# copy templates
COPY --from=template workspace/build/ cmd/server/templates/
# delete .map files
RUN find cmd/server/templates/ -name *.map | xargs rm

# build
RUN if [ "$TARGETPLATFORM" = "linux/arm64" ]; then GOARCH=arm64; else GOARCH=amd64; fi &amp;&amp; \
    CGO_ENABLED=1 GOOS=linux GOARCH=${GOARCH} GO111MODULE=on go build -a -o egress ./cmd/server

FROM livekit/gstreamer:1.22.0-prod

ARG TARGETPLATFORM

# install deps
RUN apt-get update &amp;&amp; \
    apt-get install -y \
    curl \
    fonts-noto \
    gnupg \
    pulseaudio \
    unzip \
    wget \
    xvfb \
    gstreamer1.0-plugins-base-

# install chrome
COPY --from=livekit/chrome-installer:latest /chrome-installer /chrome-installer
RUN /chrome-installer/install-chrome "$TARGETPLATFORM" &amp;&amp; \
    rm -rf /chrome-installer

# setup
RUN adduser root pulse-access &amp;&amp; \
    apt-get clean &amp;&amp; \
    rm -rf /var/lib/apt/lists/* &amp;&amp; \
    mkdir -pv ~/.cache/xdgr

# update env
ENV PATH=${PATH}:/chrome
ENV XDG_RUNTIME_DIR=$PATH:~/.cache/xdgr

# egress
COPY --from=livekit /workspace/egress /bin/

# run
COPY build/egress/entrypoint.sh /
ENTRYPOINT ["/entrypoint.sh"]

<@U01UB0LHTTQ>

I’m unable to reproduce, I think there is a difference with the system you are running it on which makes one of the inputs ~500ms slower

I’m releasing 1.6.2 now, which increases the queues by that amount and should work for you

problem solved, thank you for your help sincerely <@U01UB0LHTTQ>

Yup, typically creating a new room would be the simplest way. After the break out they could rejoin back to the main room.

Our layouts are fullscreen and don't add any filtering or manipulation - the `speaker` layout will show the current speaker only, and the `grid` layout will show all participants in a grid. If these don't fit your needs, a custom layout would be the right approach

you need a custom template for this. it could be a slight modification of your existing app

Thank you for your confirmation team :+1:

Hi <@U01UB0LHTTQ> <@U01KZ61GV5Z>,

We tried <https://github.com/livekit/egress/tree/main/template-default> template and integrate our custom components.
The thing is, adding our own custom components  (eg: whiteboard, chat, footer menu, emoji / raise hand, audio / video testing module, network stats capturing, streaming etc. ) to custom template add more complexity to the code logic to maintain those.

• is there any better approach  to address our requirement that we are not seeing keeping the code complexity to the minimum. May be inject `iframe url`  inside React root component level  so user can login to custom conference bridge ui and RoomComposite will handle meeting start, meeting end,etc. (hybrid version of Web recording + RoomComposite egress) 
• `you need a custom template for this. it could be a slight modification of your existing app` -  Can you share insights on this approach on how to convert own custom ui application into a ustom template without adding more compleity to exsisting custom react components that are not available by default livekit.
 If there aren't any, our next approach is to evaluate our requirement with web egress (<https://docs.livekit.io/server/egress/web/>) .Highly appreciate your insights in regard to above.

Any feedback on above request team ? Really appreciate if you can share you guys insights if possible

Web and Room composite are very similar, with the difference being that room composite handles joining the room, and starting and stopping the egress. You do not need to start with template-default, instead you can start with your existing app and make some small changes using <https://github.com/livekit/egress/tree/main/template-sdk>

Noted with Thanks. Will try the approach and confirm.

A follow up question: Once we create our own app with EgressHelper are we able to pass custom egress template with roomCreate options dynamically, instead of hardcoding the template path domain in egress service config yaml (template_base in <https://github.com/livekit/egress#config>).

yeah, the example <https://docs.livekit.io/server/egress/room-composite/#starting-a-roomcomposite|here> shows that you can pass that in for each request

checkout the Meet project on GitHub and see if you can match your setup to it. that would be a good starting point.

<http://github.com/livekit/meet|github.com/livekit/meet>


Thanks you! I will share this with the frontend developer.

<@U050F7L35AP> you can read more about tokens here:
<https://docs.livekit.io/concepts/authentication/>

You should _not_ store your API/secret pair on the client or expose it to your end-users. In order for other users to join the same room, you need to generate an access token for each one. The simplest way to do this by adding one of our <https://github.com/orgs/livekit/repositories?q=server-sdk&amp;type=all&amp;language=&amp;sort=|server SDKs> to your backend and making a request from Flutter to an endpoint which returns a token. You can then use that token from your client to connect to the room.

<@U01KZ0SCDKL> Actually I won't be storing the key pair in production. Right now, I am just testing it. So, as you said, anyone who has a valid token can join any room on the server right ?

are you testing locally or a self-hosted deployment?

I have the server up and am able to generate tokens with the CLI

what’s the CLI command you’re using to generate a token? can you paste it here?

I tried `livekit-cli create-token --create --list --room test --admin`

do you have a file with your API key and secret? (don’t share that here)

Like `LIVEKIT_KEYS="&lt;API_KEY&gt;: &lt;SECRET_KEY&gt;"`

`livekit-cli create-token --join --room test --admin --identity participant1 --api-key &lt;API_KEY&gt; --api-secret &lt;SECRET_KEY&gt;`

i’m not sure why you were able to generate a token with passing `identity`

Yes, at least the tcp connection. Now need to get UDP working

but if you remove `identity`, it doesn’t work?

I think before I tried `livekit-cli create-token --admin`

This also worked without identity `livekit-cli create-token --create --list --room test --admin`

hm…so what about without --create or --list?

the very fact that i’m asking this means we need to make this better and less confusing

Yeah perhaps. On the test page it would be good to have a line showing the correct incantation

<@U02CEMTDG2E> also added:
&gt; livekit-server I mean

I’m trying to get this deployment working on <http://fly.io|fly.io>, where UDP support is a bit unusual, in that you have to bind to a specific address

I seem to have gotten it working now by specific a single muxed UDP port :+1:

Is it also possible to enable TURN behind a TLS-terminating load balancer?

LiveKit Server binds to all available interfaces (`0.0.0.0`) and there isn’t a configuration option to change that. I’m not familiar with <http://fly.io|fly.io>, but I believe <@U02C10ZEP8R> has gotten it working there.

External TURN TLS termination should be there too: <https://github.com/livekit/livekit/pull/168>

It’s working now, just wondering about the TURN server

<https://github.com/livekit/livekit/blob/master/config-sample.yaml#L200>

I never got TURN working on fly, but that was some time ago. I also eventually stopped using fly for my services because of reliability issues.

Yes, we talked back in the day about it. Things are better now, so I’m trying to set it up, and see about what features would be needed for the multi-region setup to work, now that seems to be more developed in Livekit

Cool. If it works out well, let me know. I'd be happy to try again.

Hey <@U049R9F507Q> we are investigating a couple of reports but have not found anything conclusive yet. are you still experiencing the issue?

I wrote this note just after confirming another incident, around 20 minutes ago. I know of at least 4 incidents over the past few hours. I’m sure there are others.

looks like we were seeing higher error rates from one of our servers, that server had been removed from the fleet as of ~20 mins ago

I’ll continue to monitor the situation and will let you know if we’re still seeing it. Thank you!

hi jordon!
the `&lt;ParticipantTile&gt;` component already has logic for a placeholder integrated. In case you want to construct your own, it could be helpful to look at the logic in there.

that said, the latest changes (unreleased) should make your use case a lot easier with the components as well.

One way of doing this for example would be
```  const cameraTracks = useTracks([{Track.Source.Camera, withPlaceholder: true}]);

return (
   &lt;GridLayout tracks={cameraTracks}&gt;
          &lt;TrackContext.Consumer&gt;
            {(track) =&gt;
              track &amp;&amp; (
                &lt;div className="my-tile"&gt;
                  {isTrackReference(track) ? &lt;VideoTrack {...track} /&gt; :           
                                 &lt;p&gt;Camera placeholder&lt;/p&gt;
                  }
                &lt;/div&gt;
           &lt;/TrackContext.Consumer&gt;
   &lt;/GridLayout&gt;
);```
(code is untested)

IMG-20230323-WA0000.jpg

sort of. you can disable simulcast on the publishing side for screen shares by specifying   `publishDefaults: { screenShareSimulcastLayers: [] }`
the empty array means there will be no simulcast layers instantiated

FYI : `publishDefaults: { screenShareSimulcastLayers: [] }`  did the trick for us.

ss.PNG

<@U041KCGQCGZ> I’m not the expert here, but I’m not sure if the S3 uploader was designed to also work with S3-compatible storage services.

cc <@U01UB0LHTTQ> 

<@U041KCGQCGZ> can you share the errors that you do see?

"RequestError: send request failed\ncaused by: Put \"<http://mybucket.myipaddress:9000/895-2023-03-29T075351.mp4\>": dial tcp: lookup mybucket.myipaddress on 10.152.183.10:53: no such host"

it is normal because mybucket.ipaddress is not make sense.

If I try this way, I get the following error: "InvalidParameter: 1 validation error(s) found.\n- minimum field size of 1, PutObjectInput.Bucket.\n"

I believe minio requires a certain region to be set. It might be us-east-1, but I will have to double check that

I defined eu-central-1 as the region in the MinIO settings as I specified in the codes above.

It would be great if I could get a code sample for minio.

<@U01KZ61GV5Z> <@U01UB0LHTTQ> any update? it is a bit urgent for me.

<@U041KCGQCGZ> can you file an issue in the Egress repo? We are happy to help to look into it. However, to set expectation on the timelines, the team is juggling a lot and cannot make any guarantees as to when MinIO compatibility will make it in.

Use `ForcePathStyle: true` with the S3Upload
<https://github.com/livekit/protocol/pull/223>

<@U04862SMJU8> how are you obtaining the 360 video?

Any thoughts/ideas here <@U022KB72NFJ> <@U02EU2ZPTQU>?

What is the resolution of your 360 video?
You can create a new <https://developer.mozilla.org/en-US/docs/Web/API/HTMLCanvasElement/captureStream|MediaStreamTrack for your canva>s, and publish a <https://github.com/livekit/client-sdk-js#publish-tracks-from-any-source|custom video track through livekit>.

I’m not sure about the rendering part, you may need to copy the received video into the canvas

<https://threejs.org/docs/#api/en/textures/VideoTexture>

Thanks guys. On Australian time zones so bit of a delay in my response.

360 video is streamed via insta 360 pro camera. 4k resolution.

Will try your suggestions!

<@U01KZ0SCDKL> have you faced this issue?

which platform are you seeing this behavior?

```[MediaDevice{deviceId: 0, label: bottom, kind: audioinput},
MediaDevice{deviceId: 1, label: back, kind: audioinput}, MediaDevice{deviceId: 2, label: , kind: audioinput}]```
These seem to be `audioinput`  devices, so the deviceId can’t be used for `room?.selectAudioOutputDeviceId` , the kind should  be `audiooutput` .
you can try `var audioOutputs = await Hardware.instace.audioOutput();`  to get the output list.

No `room?.selectAudioOutputDeviceId` This is `room?.selectAudioInputDeviceId`

<@U04DED3V5PT> you don’t need to explicitly mention Duan, he will respond to this when he has time. Thanks for your understanding. :pray::skin-tone-4:  

Jin, I think what you are saying is:
• in the list, it's stating deviceId of 2
• when you call `room?.selectedAudioInputDeviceId`, it's returning `headset` instead of 2 as expected ?

am I understanding correctly? if so that seems to be a bug

And one more, there is no label for headset.

okay, this should be a bug, if you don’t mind please open an issue in one of flutter-webrtc or client-sdk-flutter,

OK. Let me create and share the link here

<https://github.com/livekit/client-sdk-flutter/issues/262>
Here is the ticket.

<@U045P17Q86N> if I understand you correctly, you are wondering where you can use the LK WebSocket connection more generally for sending/receiving arbitrary data messages? Is there are particular reason you want to do that? There are many general WebSocket solutions out there. As you pointed out, there is a built-in way to send data messages between connected participants via LK SDKs. Why does this not work for your use case?

for example, I use websockets on my platform to send updated data (currently use django channels which to be honest, I hate - or maybe I am just too dumb to use it right)
So that's my use case. I am running two websockets practically, and I'd rather stick to one
If not possible, will stick with two and revisit my channels implementation perhaps or just try something else

Gotcha, thanks for the additional detail! Today, the LK websocket is used for signaling and can’t be co-opted as a general-purpose websocket. For now it’s probably best to stick with two and revisit your channels implementation when it’s convenient. If you’re ever looking for how to use LK data messages, you can find that <https://docs.livekit.io/client/data-messages/|here>.

have been using them for incall chat, works fine

you saved my bacon, would've had to write so much stuff :smile:

really appreciate the kind words Cris :pray:

hi <@U050G79E42K>
thanks for the report, currently the only way to hook into these errors is to listen to `RoomEvent.MediaDevicesError` on the room itself.
We’ll return them in the `onMediaDeviceFailure` callback on `&lt;LiveKitRoom/&gt;` .

edit: PR at <https://github.com/livekit/components-js/pull/398/files>

Please thread your messages. :pray::skin-tone-4:  

crapthings wrote:
&gt; docker run --rm -it -v$PWD:/output livekit/generate
&gt; i use this to get caddy.yml
&gt; I don’t know how to fill this file

crapthings wrote:
&gt; this one works
```logging:
  logs:
    default:
      level: INFO
storage:
  "module": "file_system"
  "root": "/data"
apps:
  tls:
    certificates:
      automate:
        - <http://project-yzc.example.com|project-yzc.example.com>
        - <http://project-yzc-live.example.com|project-yzc-live.example.com>
        - <http://project-yzc-turn.example.com|project-yzc-turn.example.com>
  layer4:
    servers:
      main:
        listen: [":443"]
        routes:
          - match:
              - tls:
                  sni:
                    - "<http://project-yzc.example.com|project-yzc.example.com>"
            handle:
              - handler: tls
                connection_policies:
                  - alpn: ["http/1.1"]
              - handler: proxy
                upstreams:
                  - dial: ["localhost:8443"]
          - match:
            - tls:
                sni:
                  - "<http://project-yzc-turn.example.com|project-yzc-turn.example.com>"
            handle:
              - handler: tls
              - handler: proxy
                upstreams:
                  - dial: ["localhost:5349"]
          - match:
              - tls:
                  sni:
                    - "<http://project-yzc-live.example.com|project-yzc-live.example.com>"
            handle:
              - handler: tls
                connection_policies:
                  - alpn: ["http/1.1"]
              - handler: proxy
                upstreams:
                  - dial: ["localhost:7880"]```

livekit.txt

are you connecting with unique identities for the different clients?

which SDK is the "second" client using?

The second is using the server-sdk-go client. The participant identities and participant names are different.

I think I misspoke. The livekit server is logging an error with a REMOTE_DESCRIPTION_RECEIVED event for the initial participant. This happens after the second participant joins the room and subscribes to the first participant's tracks. After the sdp related error, the ICE connection to the initial client is reset.

The error from the above codeblock: `"error":"setting remote description failed: remoteDescription contained media section without mid value",`

The initial participant is using the client-sdk-js, the second participant is using server-sdk-go. This seems to only reproduce on Firefox in some network conditions. I'm using livekit v1.3.5.

Oh that's interesting.. Which participant is the error logged for? Can you attach the associated localDescription and remoteDescription in question?

The participant that this is logged for is the initial participant using client-sdk-js. The log is coming from the rtctransport in livekit server.

The sdp offer that fails is the following.
```m=audio 0 UDP/TLS/RTP/SAVPF 0
c=IN IP4 0.0.0.0
a=rtpmap:0 PCMU/8000
a=inactive
m=video 9 UDP/TLS/RTP/SAVPF 126 127 120 124 121 125 97 98
c=IN IP4 0.0.0.0
a=rtpmap:126 H264/90000
a=rtpmap:127 rtx/90000
a=rtpmap:120 VP8/90000
a=rtpmap:124 rtx/90000
a=rtpmap:121 VP9/90000
a=rtpmap:125 rtx/90000
a=rtpmap:97 H264/90000
a=rtpmap:98 rtx/90000
a=fmtp:126 profile-level-id=42e01f;level-asymmetry-allowed=1;packetization-mode=1
a=fmtp:97 profile-level-id=42e01f;level-asymmetry-allowed=1
a=fmtp:127 apt=126
a=fmtp:120 max-fs=12288;max-fr=60
a=fmtp:124 apt=120
a=fmtp:121 max-fs=12288;max-fr=60
a=fmtp:125 apt=121
a=fmtp:98 apt=97
a=rtcp-fb:126 nack
a=rtcp-fb:126 nack pli
a=rtcp-fb:126 ccm fir
a=rtcp-fb:126 goog-remb
a=rtcp-fb:126 transport-cc
a=rtcp-fb:120 nack
a=rtcp-fb:120 nack pli
a=rtcp-fb:120 ccm fir
a=rtcp-fb:120 goog-remb
a=rtcp-fb:120 transport-cc
a=rtcp-fb:121 nack
a=rtcp-fb:121 nack pli
a=rtcp-fb:121 ccm fir
a=rtcp-fb:121 goog-remb
a=rtcp-fb:121 transport-cc
a=rtcp-fb:97 nack
a=rtcp-fb:97 nack pli
a=rtcp-fb:97 ccm fir
a=rtcp-fb:97 goog-remb
a=rtcp-fb:97 transport-cc
a=extmap:3 urn:ietf:params:rtp-hdrext:sdes:mid
a=extmap:4 <http://www.webrtc.org/experiments/rtp-hdrext/abs-send-time>
a=extmap:5 urn:ietf:params:rtp-hdrext:toffset
a=extmap:6/recvonly <http://www.webrtc.org/experiments/rtp-hdrext/playout-delay>
a=extmap:7 <http://www.ietf.org/id/draft-holmer-rmcat-transport-wide-cc-extensions-01>
a=setup:actpass
a=mid:1
a=msid:- {8bd8a8f7-207b-4b99-bbd2-808efdf92860}
a=sendonly
a=ice-ufrag:10e5da30
a=ice-pwd:e0e19051ba23476a8f9c07d5534dd2d2
a=ssrc:2883058285 cname:{8a113342-10cc-4b93-afb0-f9d6db97413f}
a=ssrc:433786171 cname:{8a113342-10cc-4b93-afb0-f9d6db97413f}
a=ssrc-group:FID 2883058285 433786171
a=rtcp-mux
a=rtcp-rsize
m=application 9 UDP/DTLS/SCTP webrtc-datachannel
c=IN IP4 0.0.0.0
a=setup:actpass
a=mid:2
a=sendrecv
a=ice-ufrag:10e5da30
a=ice-pwd:e0e19051ba23476a8f9c07d5534dd2d2
a=sctp-port:5000
a=max-message-size:1073741823
 parsed:0xc001728480```
I believe the following is the local/remote sdp from the browser's webrtc info.
```Local SDP (Offer)

v=0
o=mozilla...THIS_IS_SDPARTA-99.0 4991156590662245017 1 IN IP4 0.0.0.0
s=-
t=0 0
a=sendrecv
a=fingerprint:sha-256 67:8E:0E:E6:A3:73:39:DA:C5:8F:6F:10:17:05:56:F5:1C:5A:9D:10:B7:6E:EB:0C:30:61:B9:57:63:97:07:1C
a=group:BUNDLE 1 2
a=ice-options:trickle
a=msid-semantic:WMS *
m=audio 0 UDP/TLS/RTP/SAVPF 0
c=IN IP4 0.0.0.0
a=inactive
a=rtpmap:0 PCMU/8000
m=video 33514 UDP/TLS/RTP/SAVPF 126 127 120 124 121 125 97 98
c=IN IP4 172.22.0.1
a=candidate:0 1 UDP 2121793791 192.168.6.2 34209 typ host
a=candidate:4 1 UDP 2122055935 172.20.0.1 33583 typ host
a=candidate:8 1 UDP 2122252543 172.22.0.1 33514 typ host
a=candidate:12 1 UDP 2122187007 172.19.0.1 52634 typ host
a=candidate:16 1 UDP 2121924863 172.17.0.1 38253 typ host
a=candidate:20 1 UDP 2121990399 172.21.0.1 57504 typ host
a=candidate:24 1 UDP 2122121471 172.18.0.1 45428 typ host
a=candidate:28 1 UDP 2121859327 fd86:2df3:c714:1:c0b:afbe:38fd:a54e 33776 typ host
a=candidate:30 1 TCP 2105065727 192.168.6.2 9 typ host tcptype active
a=candidate:32 1 TCP 2105327871 172.20.0.1 9 typ host tcptype active
a=candidate:34 1 TCP 2105524479 172.22.0.1 9 typ host tcptype active
a=candidate:36 1 TCP 2105458943 172.19.0.1 9 typ host tcptype active
a=candidate:38 1 TCP 2105196799 172.17.0.1 9 typ host tcptype active
a=candidate:40 1 TCP 2105262335 172.21.0.1 9 typ host tcptype active
a=candidate:42 1 TCP 2105393407 172.18.0.1 9 typ host tcptype active
a=candidate:44 1 TCP 2105131263 fd86:2df3:c714:1:c0b:afbe:38fd:a54e 9 typ host tcptype active
a=candidate:0 2 UDP 2121793790 192.168.6.2 35782 typ host
a=candidate:4 2 UDP 2122055934 172.20.0.1 55583 typ host
a=candidate:8 2 UDP 2122252542 172.22.0.1 38338 typ host
a=candidate:12 2 UDP 2122187006 172.19.0.1 40020 typ host
a=candidate:16 2 UDP 2121924862 172.17.0.1 51843 typ host
a=candidate:20 2 UDP 2121990398 172.21.0.1 34472 typ host
a=candidate:24 2 UDP 2122121470 172.18.0.1 41503 typ host
a=candidate:28 2 UDP 2121859326 fd86:2df3:c714:1:c0b:afbe:38fd:a54e 50380 typ host
a=candidate:30 2 TCP 2105065726 192.168.6.2 9 typ host tcptype active
a=candidate:32 2 TCP 2105327870 172.20.0.1 9 typ host tcptype active
a=candidate:34 2 TCP 2105524478 172.22.0.1 9 typ host tcptype active
a=candidate:36 2 TCP 2105458942 172.19.0.1 9 typ host tcptype active
a=candidate:38 2 TCP 2105196798 172.17.0.1 9 typ host tcptype active
a=candidate:40 2 TCP 2105262334 172.21.0.1 9 typ host tcptype active
a=candidate:42 2 TCP 2105393406 172.18.0.1 9 typ host tcptype active
a=candidate:44 2 TCP 2105131262 fd86:2df3:c714:1:c0b:afbe:38fd:a54e 9 typ host tcptype active
a=sendonly
a=extmap:3 urn:ietf:params:rtp-hdrext:sdes:mid
a=extmap:4 <http://www.webrtc.org/experiments/rtp-hdrext/abs-send-time>
a=extmap:5 urn:ietf:params:rtp-hdrext:toffset
a=extmap:6/recvonly <http://www.webrtc.org/experiments/rtp-hdrext/playout-delay>
a=extmap:7 <http://www.ietf.org/id/draft-holmer-rmcat-transport-wide-cc-extensions-01>
a=fmtp:126 profile-level-id=42e01f;level-asymmetry-allowed=1;packetization-mode=1
a=fmtp:97 profile-level-id=42e01f;level-asymmetry-allowed=1
a=fmtp:127 apt=126
a=fmtp:120 max-fs=12288;max-fr=60
a=fmtp:124 apt=120
a=fmtp:121 max-fs=12288;max-fr=60
a=fmtp:125 apt=121
a=fmtp:98 apt=97
a=ice-pwd:621d257843e04e72e78c9855ba01a96d
a=ice-ufrag:6265770a
a=mid:1
a=msid:- {52d72679-bdfe-452c-91d9-7698b2b4589c}
a=rtcp:38338 IN IP4 172.22.0.1
a=rtcp-fb:126 nack
a=rtcp-fb:126 nack pli
a=rtcp-fb:126 ccm fir
a=rtcp-fb:126 goog-remb
a=rtcp-fb:126 transport-cc
a=rtcp-fb:120 nack
a=rtcp-fb:120 nack pli
a=rtcp-fb:120 ccm fir
a=rtcp-fb:120 goog-remb
a=rtcp-fb:120 transport-cc
a=rtcp-fb:121 nack
a=rtcp-fb:121 nack pli
a=rtcp-fb:121 ccm fir
a=rtcp-fb:121 goog-remb
a=rtcp-fb:121 transport-cc
a=rtcp-fb:97 nack
a=rtcp-fb:97 nack pli
a=rtcp-fb:97 ccm fir
a=rtcp-fb:97 goog-remb
a=rtcp-fb:97 transport-cc
a=rtcp-mux
a=rtcp-rsize
a=rtpmap:126 H264/90000
a=rtpmap:127 rtx/90000
a=rtpmap:120 VP8/90000
a=rtpmap:124 rtx/90000
a=rtpmap:121 VP9/90000
a=rtpmap:125 rtx/90000
a=rtpmap:97 H264/90000
a=rtpmap:98 rtx/90000
a=setup:actpass
a=ssrc:3557748696 cname:{eb870658-56ca-4b62-8810-b69b2dd60b2b}
a=ssrc:150923256 cname:{eb870658-56ca-4b62-8810-b69b2dd60b2b}
a=ssrc-group:FID 3557748696 150923256
m=application 52983 UDP/DTLS/SCTP webrtc-datachannel
c=IN IP4 172.22.0.1
a=candidate:0 1 UDP 2121793791 192.168.6.2 43733 typ host
a=candidate:4 1 UDP 2122055935 172.20.0.1 39965 typ host
a=candidate:8 1 UDP 2122252543 172.22.0.1 52983 typ host
a=candidate:12 1 UDP 2122187007 172.19.0.1 38185 typ host
a=candidate:16 1 UDP 2121924863 172.17.0.1 45512 typ host
a=candidate:20 1 UDP 2121990399 172.21.0.1 37559 typ host
a=candidate:24 1 UDP 2122121471 172.18.0.1 44267 typ host
a=candidate:28 1 UDP 2121859327 fd86:2df3:c714:1:c0b:afbe:38fd:a54e 47537 typ host
a=candidate:30 1 TCP 2105065727 192.168.6.2 9 typ host tcptype active
a=candidate:32 1 TCP 2105327871 172.20.0.1 9 typ host tcptype active
a=candidate:34 1 TCP 2105524479 172.22.0.1 9 typ host tcptype active
a=candidate:36 1 TCP 2105458943 172.19.0.1 9 typ host tcptype active
a=candidate:38 1 TCP 2105196799 172.17.0.1 9 typ host tcptype active
a=candidate:40 1 TCP 2105262335 172.21.0.1 9 typ host tcptype active
a=candidate:42 1 TCP 2105393407 172.18.0.1 9 typ host tcptype active
a=candidate:44 1 TCP 2105131263 fd86:2df3:c714:1:c0b:afbe:38fd:a54e 9 typ host tcptype active
a=sendrecv
a=ice-pwd:621d257843e04e72e78c9855ba01a96d
a=ice-ufrag:6265770a
a=mid:2
a=setup:actpass
a=sctp-port:5000
a=max-message-size:1073741823

Remote SDP (Answer)

v=0
o=- 8702163150672924791 1680552103 IN IP4 0.0.0.0
s=-
t=0 0
a=sendrecv
a=fingerprint:sha-256 26:39:AA:F9:DC:E0:39:49:1A:EA:01:96:2A:36:D9:B0:AC:D3:86:3A:34:77:F4:90:24:E6:6A:39:5F:FE:6C:11
a=group:BUNDLE 1 2
m=audio 0 UDP/TLS/RTP/SAVPF 0
c=IN IP4 0.0.0.0
a=candidate:1247215762 1 udp 2130706431 104.171.200.75 53968 typ host
a=candidate:4154749298 1 tcp 2124414975 104.171.200.75 7881 typ host tcptype passive
a=candidate:233762139 1 udp 2130706431 172.17.0.1 53455 typ host
a=candidate:2954730683 1 tcp 2124414975 172.17.0.1 7881 typ host tcptype passive
a=candidate:3528925834 1 udp 2130706431 172.18.0.1 50651 typ host
a=candidate:1873040234 1 tcp 2124414975 172.18.0.1 7881 typ host tcptype passive
a=candidate:3308009161 1 udp 2130706431 172.19.0.1 57219 typ host
a=candidate:2027898665 1 tcp 2124414975 172.19.0.1 7881 typ host tcptype passive
a=candidate:2193111953 1 udp 2130706431 172.20.0.1 53456 typ host
a=candidate:1061898865 1 tcp 2124414975 172.20.0.1 7881 typ host tcptype passive
a=candidate:2512596946 1 udp 2130706431 172.21.0.1 53698 typ host
a=candidate:674258482 1 tcp 2124414975 172.21.0.1 7881 typ host tcptype passive
a=sendrecv
m=video 9 UDP/TLS/RTP/SAVPF 126
c=IN IP4 0.0.0.0
a=recvonly
a=extmap:3 urn:ietf:params:rtp-hdrext:sdes:mid
a=extmap:7 <http://www.ietf.org/id/draft-holmer-rmcat-transport-wide-cc-extensions-01>
a=fmtp:126 profile-level-id=42e01f;level-asymmetry-allowed=1;packetization-mode=1
a=ice-pwd:ufqtmVLQsKaPlPbzCfxNOBqPizWdmQRv
a=ice-ufrag:fJvScAoYCyCJOzfo
a=mid:1
a=rtcp-fb:126 nack
a=rtcp-fb:126 nack pli
a=rtcp-fb:126 ccm fir
a=rtcp-fb:126 goog-remb
a=rtcp-fb:126 transport-cc
a=rtcp-mux
a=rtcp-rsize
a=rtpmap:126 H264/90000
a=setup:active
m=application 9 UDP/DTLS/SCTP webrtc-datachannel
c=IN IP4 0.0.0.0
a=sendrecv
a=ice-pwd:ufqtmVLQsKaPlPbzCfxNOBqPizWdmQRv
a=ice-ufrag:fJvScAoYCyCJOzfo
a=mid:2
a=setup:active
a=sctp-port:5000```

Which version of client-sdk-js are you using?

1.7.1 Let me know if you would like me to try other versions.

Any other thoughts on this issue ? I think I’ll open an issue in client sdk js

Hey Christopher, yes please file an issue for this in JS repo. attaching the offer &amp; log as you did here would be helpful.

Forgot the link, <https://github.com/livekit/client-sdk-js/issues/648>.

Hey, so yeah different platforms handle this differently. But the safest bet is to have the user click a button before you try to do anything with WebAudio. You can try listening for _any_ click on the document but I’ve had issues in the past on iOS.

After that you can resume the audio context <https://developer.mozilla.org/en-US/docs/Web/API/AudioContext/resume> and play any AudioElements that couldn’t play.

Thank you <@U04F14MLUJE>! That resume function was the missing piece.

great! We’re working on simplifying the APIs for this in the context of livekit for the future

Another question <@U04F14MLUJE>. I’m looking at this code block in the spatial audio controller where the web audio nodes are disconnected and set to null: <https://github.com/livekit-examples/spatial-audio/blob/44f7973f49bf8241738b057d0672858a8d8297df/src/controller/SpatialAudioController.tsx#L125-L133>

Is there a reason why you don’t use `track.detach()` instead of manually disconnected+setting to null/undefined? I’m seeing that `track.detach()` should call this function that disconnects the source and gain node, but not the panner node which I suppose would still need to be manually disconnected: <https://github.com/livekit/client-sdk-js/blob/676ecd3f0ee30ba3b1a15d4296ff41084c1baf0a/src/room/track/RemoteAudioTrack.ts#L201-L206>

So unfortunately - because the spatial example must publish a local track (when the local participant plays jukebox in the center), the spatial example doesn’t hit that code path.

&gt; We’re working on simplifying the APIs for this in the context of livekit for the future
When I said this I was actually talking about exactly this! So good eye!

Ah okay, interesting. I’m only dealing with remote tracks so do you think manually dealing with the panner node + calling `track.detach()` will be sufficient for cleaning up audio nodes? Also, does anything need to be done to the HTML Audio Elements themselves when unsubscribing?

in your case you can use `expWebAudioMix`  + `track.setWebAudioPlugins()` . Then you can use track `attach`/`detach`

I’ve tried searching the docs and can’t see any info on `expWebAudioMix`  and `track.setWebAudioPlugins()` . Might be missing them, but do you know where I can learn more?

so I don’t think this behavior is well documents (or documented at all yet) but you already found the cleanup code for it. Here’s the setWebAudioPlugins code: <https://github.com/livekit/client-sdk-js/blob/676ecd3f0ee30ba3b1a15d4296ff41084c1baf0a/src/room/track/RemoteAudioTrack.ts#L159>

Nope, just writing a voice client using vanilla js with the livekit-client and livekit cloud.

ok in that case you’ll need to also call `setAudioContext` <https://github.com/livekit/client-sdk-js/blob/676ecd3f0ee30ba3b1a15d4296ff41084c1baf0a/src/room/track/RemoteAudioTrack.ts#L145>

actually correction, you can set the audio context at the room level

room options has a field `expWebAudioMix`

I’m looking at the code in real time here so bear with me :sweat_smile:

```handleTrackSubscribed(track, publication, participant) {
    if (track.kind === Track.Kind.Audio) {
      const { audioContext } = this.room

      // Create nodes
      const sourceNode = audioContext.createMediaStreamSource(track.mediaStream)
      const pannerNode = audioContext.createPanner()

      // Save references to nodes in track object
      track.pannerNode = pannerNode

      /* 
      Modifying panner node attributes here
      */

      // Connect nodes to destination
      sourceNode.connect(pannerNode).connect(audioContext.destination)

      const audioElement = new Audio()
      audioElement.muted = true
      audioElement.srcObject = track.mediaStream
      audioElement.play()

      track.attachedElements.push(audioElement)

      // Map user id to their track
      this.trackMap[participant.identity] = track
    }
  }

  handleTrackUnsubscribed(track, publication, participant) {
    track.detach()
    delete this.trackMap[participant.identity]
  }```

Here’s how I currently handle the track sub where I set up the audio graph, and the unsub where I detach btw

In case you’re curious or if it helps clarify what I’m achieving

Currently this works but I’m sure it can be more efficient with some of the concepts you’re mentioning

yeah so once you use the set audio plugins api you will call track.attach() and you won’t need to do any of your own audioElement management

and you will also be able to remove `track.attachedElements.push(audioElement)`

Cool! I’m now using `track.setWebAudioPlugins()`, setting `expWebAudioMix` to true in the room constructor, using `track.attach()`  to create an audio element + appending with `document.body.append(audioElement)` , and using `track.detach()` upon unsubbing. It seems to be working well.

Should this cover my bases for properly disconnecting the panner node (tbh I still don’t see how that happens within the `detach()` code) and cleaning up the audio element?

do you have to create the audio element yourself and append it to the document?

ah, it looks like i can just call it without operating on the return value! I was previously following the example here: <https://github.com/livekit/client-sdk-js>

Do you think there’s any further removal process necessary for the audio elements that the nodes are disconnected from? It’s not clear to me that the elements are ever subject to garbage collection. Don’t want this to effect performance when many tracks are running at once.

a full example (including starting of ingress) can be found here <https://github.com/livekit-examples/livestream>

Although the 404 comes in, the connect() method works and a room.state==="disconnected" is returned. Catch block does not trigger. Makes it harder to handle this.

Ok got it -- issue was non-existence of the room.

So the pre-room page has logic that awaits 204 on successful room creation before navigating to the room page and attempting connection. Everything backend was good. BUT, there seems to be a gap or delay in K8 cluster state.

Might qualify as a bug. If Express creates a room and succeeds, that room should be ready to accept connections. This is not the case.

Workaround is to timeout in the client (1000ms right now) before attempting to connect to room.

Of course this only applies to the first participant who joins the room, and to access tokens that do not have the authority to create rooms (instead creating rooms manually.)

Are you saying if `createRoom` succeeds with 200, and you tried to join the room immediately, it would still sometimes return 404?

If that's the case, that'd indeed a bug.

Hello <@U04B8NLF6ER>, I'm rather curious to know how you are trying to achieve this. I am asking because each participant needs to be connected with a unique identity to each room. Are you trying to connect to multiple rooms at the same time with the same connection? If so, I don't think that is a possible thing currently.

We want to use livekit to create a 1-to-many distribution for a large audience. So one livekit server will not be enough to handle the load. So we were thinking to have one intermediate server which receives the input and forwards it to N Edge servers. Basically creating a kind of a WebRTC CDN.

Our though was to create a kind of bridge to connect to room A to get the RTP data. Then connect to a different room with another connection and forward the RTP data to the roomB.

We experimented with the RTMP ingress / egress but this adds a little to much latency for our use case. Or is there any other option to scale livekit in 1-to-many scenarios?

This is a good question and probably more suited for the LiveKit team to answer. Maybe <@U01KZ61GV5Z> would have more info or guidance on this one.

One thing Id like to point out is the LiveKit Cloud does have the multiple server architecture for 1 room in place. Maybe this is something your team could look at using rather than a custom solution. It supports multiple 1000s of users in 1 single room. 1-to-many. Just a suggestion.

It's definitely possible to use server-sdk-go for this. You'd essentially need these forwarders in place that connect to two different rooms, and in `TrackSubscribed` callback, publish it to the other room.

We'll put up example code in the future for something like that.

But I second the suggestion for just using LiveKit Cloud :slightly_smiling_face: It should do exactly what you need without needing any additional code.

Thank you for the input I will explore this and if we get it running we might provide that example for you :slightly_smiling_face:
Although livekit cloud has good pricing, it seems that we have to go the long way. We have developers but no money, sounds contradicting but that's how it is :)

awesome! I'd welcome any examples if you get it to work. feel free to DM me with them and I'll get them posted

Relevant code:
```  def livekit_token_for(person)
    token = LiveKit::AccessToken.new(api_key: Credential.fetch(:livekit, :api_key), api_secret: Credential.fetch(:livekit, :api_secret))
    token.identity = person.id
    token.name = person.name
    token.add_grant(roomJoin: true, room: id)
    token
  end```

Could you share a token rendered by ruby? Might be easier to help &amp; debug.

Are you returning `<http://token.to|token.to>_jwt`?

`eyJhbGciOiJIUzI1NiJ9.eyJuYW1lIjoibWUiLCJtZXRhZGF0YSI6bnVsbCwic2hhMjU2IjpudWxsLCJ2aWRlbyI6eyJyb29tSm9pbiI6dHJ1ZSwicm9vbSI6IjFxMjM1In0sImV4cCI6MTY4MDU0MzU5MCwibmJmIjoxNjgwNTI5MTg1LCJpc3MiOiJBUElWNnM2WHJXNVdidG4iLCJzdWIiOjF9.JGDW0SrbN2Rtw_kX9L5gaoDKILQCTBfQ0ieBI8dtXJA`

Quick question.. is the room you are trying to join already created?

Or would it need an admin grant for that to work?

I'm rather curious to know if your server is configured or not to allow auto room creation

How can I tell? i don’t see an option for that

```token.add_grant(roomCreate: true, roomJoin: true, room: id)```


By changing `t.identity = 1` to `t.identity = "1"`

Tbh, I have never tried an identity being only numerics. Looks like its really expecting a string for the identity, which seems odd.

Yeah, well at least it’s solved :slightly_smiling_face:

I just used the user ID. Should it be a string?

I've never really bothered.. I'm using pretty much the same thing although adding a prefix to my identities for my use case. "p-{id}" per say.

The only important aspect is making sure every identity are different for every users.

ah yeah, identity cannot be set to numeric.. I guess this is a drawback of ruby and duck-typing not playing well.. I think we can add a `to_s` during token generation though.