LiveKit Community

Yes, you should be able to use `listEgress` and `stopEgress` on our Egress client (which runs on your backend server), see docs here: <https://docs.livekit.io/server/egress/#listegress>

also, if you need, you can request an increase to the egress limit on your project settings page

this is how self hosted or for cloud solution?

it gives me error on the request for ListEgress

ok I maanage to get egress lists, and I have only 1 egress, but in yout dashboard it show me 10

noting here for the record: Olexiy’s project had exceeded the <https://docs.livekit.io/cloud/quotas-and-limits/|monthly 60 minutes of transcoding that comes with the free plan>; our UI wasn’t super clear about that (we’ve filed a ticket to improve this!)

<@U051KT6B0RZ> ListEgress API only shows them for up to 3 days after they end, and is more useful for looking at active egress or checking if an egress failed. The dashboard will show all past egress

Bumping this... has anyone hosted a livekit server on GCP/AWS who would be willing to chat about how you identified what kind of instance to deploy?

Hello, there's a few things to consider when choosing between open source and cloud.
1. Identify how many users and what type of use-case you have. We have some load-testing docs which should help you here: <https://docs.livekit.io/oss/benchmark/#load-tester|docs>. Note that these were run on a single 16 vCPU gcp VM.
2. How geographically diverse is your workload. If you only have users in one region with smaller workloads, you might be fine to run a single VM. If you've got users in many places, you might need to run multi-VM/kubernetes in different regions backed by redis to ensure good connection speed for your users. Depending on the answer here you might see increased compute and devops cost for open source.
3. Scale. If you're starting small and only have a few users currently, open source can be a good option, but cloud also offers 50GB free per month: <https://docs.livekit.io/cloud/quotas-and-limits/|docs>. Also keep in mind that cloud was designed to be a drop-in config change for users wanting to switch between open source and cloud.
Last we have a cloud calculator that can help estimate your monthly cloud costs that you can use to compare to running open source: <https://www.livekit.io/pricing/calculator|calc>
Hope this helps!

Thanks, this is helpful! Our application will be mostly 1:1 video meetings, with multiple running at the same time. So I'm thinking the "Large meeting" application in the load-testing docs will be more applicable than the "Livestreaming" one, but I'm guessing I'll need to also take into account additional CPU overhead for each individual meeting?

I would think the 'Conferencing' use case would be most applicable for the calculator; with 2 people on screen at a time. Multiple rooms won't add much to CPU overhead, and with open source a room can only be hosted on a single node, so with small 2 person rooms you can scale horizontally as needed.

Oh yep, sorry: to clarify, I was talking about the loadtest page. I've checked out the calculator with the Conferencing option and the costs work well for us as we start building. I'm trying to figure out if there's a point where it becomes more cost effective for us to self-host as we scale up; also have a meeting w/ the sales team tomorrow which should help. Thank you!

You would want to host your own rtmp server

What versions of LiveKit and Egress are you using?
StopEgress returning an `ENDING` status is expected but it sounds like the final update is getting lost

Hello,
Thank you for your response.
I am using 1.3.5 version of LiveKit and 1.6.0 of Egress.
Status isn't such a big problem, but the webhook is. :sweat_smile:

Do you egress logs have any errors, and does your LiveKit server have any logs that say `egress updated/finished` ? And are you using `egress: use_psrpc: true`  in your LiveKit config? 

The status updates are sent from Egress to LiveKit server, which then sends webhooks based on the status of the update. I'm wondering if the updates aren't being received

I have this error on LiveKit Logs :
```livekit-livekit-1  | 2023-04-04T19:43:31.698741771Z 2023-04-04T19:43:31.698Z	ERROR	livekit	service/ioinfo.go:166	failed to update egress info	{"error": "could not update egress info: CROSSSLOT Keys in request don't hash to the same slot", "errorVerbose": "CROSSSLOT Keys in request don't hash to the same slot\ncould not update egress info\<http://ngithub.com/livekit/livekit-server/pkg/service.(*RedisStore).UpdateEgress|ngithub.com/livekit/livekit-server/pkg/service.(*RedisStore).UpdateEgress>\n\t/workspace/pkg/service/redisstore.go:415\ngithub.com/livekit/livekit-server/pkg/service.(*IOInfoService).UpdateEgressInfo\n\t/workspace/pkg/service/ioinfo.go:82\ngithub.com/livekit/livekit-server/pkg/service.(*IOInfoService).egressWorkerDeprecated.func1\n\t/workspace/pkg/service/ioinfo.go:164\nruntime.goexit\n\t/usr/local/go/src/runtime/asm_amd64.s:1598"}```
and Egress logs has nothing bad

By the way we'll update the LiveKit version to the latest (1.4.0) and we'll enable `use_psrpc` anyway

Interesting, this is a bug with using a redis cluster. I’ve created a ticket here <https://github.com/livekit/livekit/issues/1578> and we’ll try to have it fixed for the next version

Update: After updating my LiveKit version to 1.4.0 and enabling the configurations to `use_psrpc` and `signal_relay`  we still get the same error on both LiveKit &amp; Egress servers.
This is the error from Egress server :
```livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400138278Z 2023-04-05T19:32:24.400Z	ERROR	egress	service/handler.go:211	failed to send update	{"nodeID": "NE_Dd3a54Ks3bZm", "handlerID": "EGH_ied2TAxWwFvM", "clusterID": "", "egressID": "EG_5TW2iaRegUUL", "error": "could not update egress info: CROSSSLOT Keys in request don't hash to the same slot"}
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400154318Z <http://github.com/livekit/egress/pkg/service.sendUpdate|github.com/livekit/egress/pkg/service.sendUpdate>
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400159540Z 	/workspace/pkg/service/handler.go:211
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400162871Z <http://github.com/livekit/egress/pkg/service.(*Handler).sendUpdate|github.com/livekit/egress/pkg/service.(*Handler).sendUpdate>
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400166050Z 	/workspace/pkg/service/handler.go:183
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400169534Z <http://github.com/livekit/egress/pkg/service.(*Handler).Run|github.com/livekit/egress/pkg/service.(*Handler).Run>
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400172838Z 	/workspace/pkg/service/handler.go:101
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400176329Z main.runHandler
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400179469Z 	/workspace/cmd/server/main.go:215
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400191531Z <http://github.com/urfave/cli/v2.(*Command).Run|github.com/urfave/cli/v2.(*Command).Run>
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400194498Z 	/root/go/pkg/mod/github.com/urfave/cli/v2@v2.25.0/command.go:273
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400197516Z <http://github.com/urfave/cli/v2.(*Command).Run|github.com/urfave/cli/v2.(*Command).Run>
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400200311Z 	/root/go/pkg/mod/github.com/urfave/cli/v2@v2.25.0/command.go:266
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400203843Z <http://github.com/urfave/cli/v2.(*App).RunContext|github.com/urfave/cli/v2.(*App).RunContext>
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400207158Z 	/root/go/pkg/mod/github.com/urfave/cli/v2@v2.25.0/app.go:332
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400216837Z <http://github.com/urfave/cli/v2.(*App).Run|github.com/urfave/cli/v2.(*App).Run>
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400219895Z 	/root/go/pkg/mod/github.com/urfave/cli/v2@v2.25.0/app.go:309
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400223146Z main.main
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400226295Z 	/workspace/cmd/server/main.go:71
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400229728Z runtime.main
livekit-egress-livekit-egress-1  | 2023-04-05T19:32:24.400232695Z 	/usr/lib/go-1.18/src/runtime/proc.go:250```
However we now receive the webhook.
<@U01UB0LHTTQ>

<@U050J8SDA5D> what version of Egress and livekit-server are you running?

can you include the request.json that you've sent as well?

<@U01KZ61GV5Z>
thanks for the reply!

how can i  see the version both of Egress and livekit-server?
is there any cli command or something else?

Egress is running on Digital Ocean k8s cluster
and
Livekit-server is running on Docker hosted on VM

Docker image is latest one i suppose.
```REPOSITORY               TAG       IMAGE ID       CREATED        SIZE
livekit/livekit-server   latest    a56c926b5d44   5 weeks ago    42.2MB```
request.json is below.
```{
  "room_name": "egress-test01",
  "layout": "speaker-dark",
  "file": {
    "filepath": "my-room-test.mp4",
    "s3": {
      "access_key": "xxxxx",
      "secret": "xxxx",
      "region": "ap-northeast-3",
      "endpoint": "<https://xxxx-livekit-egress.s3.ap-northeast-3.amazonaws.com>"
    }
  }
}```

your request under s3 is missing `bucket` . see example <https://docs.livekit.io/server/egress/room-composite/#stream-to-hls|here>

<@U01KZ61GV5Z>
sorry, i misunderstood the required parms...
thanks a lot again!

image.png

It would be nice if the completed ones were not shown. It looks like it will return a very long json after a while. <@U01KZ61GV5Z>

you can use `--active` to list the only active ones.. currently it returns everything in the last few days

The product page states it has 6 cores and 12 threads, does that match the machine that you have?

Is this running on a cloud somewhere? If you could link us to the instance type, that'd be helpful. CC: <@U03H6MNQ9G9>

yes it matches.
It is <https://www.hetzner.com/dedicated-rootserver/ax41-nvme|Dedicated Root Server>.

can you run `lscpu` and attach the output here?

```Architecture:                    x86_64
CPU op-mode(s):                  32-bit, 64-bit
Byte Order:                      Little Endian
Address sizes:                   43 bits physical, 48 bits virtual
CPU(s):                          12
On-line CPU(s) list:             0-11
Thread(s) per core:              2
Core(s) per socket:              6
Socket(s):                       1
NUMA node(s):                    1
Vendor ID:                       AuthenticAMD
CPU family:                      23
Model:                           113
Model name:                      AMD Ryzen 5 3600 6-Core Processor
Stepping:                        0
Frequency boost:                 enabled
CPU MHz:                         2201.064
CPU max MHz:                     3600.0000
CPU min MHz:                     2200.0000
BogoMIPS:                        7200.16
Virtualization:                  AMD-V
L1d cache:                       192 KiB
L1i cache:                       192 KiB
L2 cache:                        3 MiB
L3 cache:                        32 MiB
NUMA node0 CPU(s):               0-11```

What is the content of either `/sys/fs/cgroup/cpu.max` or `/sys/fs/cgroup/cpu,cpuacct/cpuacct.usage_percpu`, depending on the kernel version

```cat /sys/fs/cgroup/cpu,cpuacct/cpuacct.usage_percpu

17152890430665 17125510743844 17253791459997 17637899768908 17605844271262 17567205759884 17083243343493 17018759744664 17048314156725 17423645465750 17368707803920 17370324750433 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0```

```cat: /sys/fs/cgroup/cpu.max: No such file or directory```

Thanks for the report. We have a workaround that should ship in the next release.

I enabled that setting and the publisher's video stopped coming. I haven't tested it but I think some events are starting to not trigger, TrackPublishedEvent for example.
When I disabled it everything was fine.

Thank you for the reports. I can reproduce these as well when Redis is used. We'll look into it.

<@U041KCGQCGZ> and <@U02LCP104CQ> thanks for reporting this. This issue is fixed in the latest release - <https://github.com/livekit/livekit/releases/tag/v1.4.1>

<@U04VDH778AV> I tried `signal_relay`  but still the publisher's video doesn't coming. I think there is one more problem that has been overlooked.

I think you can switch audio input/output, video input device by the hardware API during the calling.

Two devices that are too close will cause their respective echo cancellation to fail, and the sound will be cross-passed and amplified, but you can try to reduce the output or input gain.

Thank you for replying. About the in-call and media channel, I don't think that the hardware API is relevant. It gives me which devices are available. However, I want to know how to use the media channel for the call. Right now, the communication is done through the in-call channel of a phone, which is typically used when making voice calls. I want it to be done using the media channel, which is used when watching videos or listening to music using a phone.

I remember when I was using Agora, there was an option to choose among audio profiles or scenarios to achieve this (wheter to use in-call or media volume). Does LiveKit offer an equivalent feature?

this seems like a local issue with your app and appium integration, not sure where we fit into this

So <@U01UB0LHTTQ> probably knows the answer to this

This was a bug, fixed in <https://github.com/livekit/livekit/pull/1547>

cool thanks for the info, I will update my livekit and check it out

Unfortunately we don’t have UI for that yet; DM me your project and I’ll do it for you manually

There is my config for room:
```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
    },
},```


Which documentation are you looking at? Could you share a link?

Thanks for the report! This is our component docs on <http://docs.livekit.io|docs.livekit.io>. I'll fix it now

Hey Matt!

You are right about LiveKit Cloud being a turnkey service. Though tokens is not something we can easily abstract away for you.

The main reason is an access token encapsulate identity and permissions (who the user is, and what they are allowed to do). They are created using the API key/secret that you can create within your project.

The reason why we recommend putting token generation on the backend is to avoid having to embed api secret in the client code itself (where end user may see it embedded in the code). If the end user gets the api secret, they are able to perform operations that you may not be intending them to perform.

To get a "token server" running you could have it running in a serverless function. The only library that you'd need is the `AccessToken` class.

similarly, the room management features are also available only on the server side.. However, we could look into making them work on the client too. with an access token, a client *should* be able to make these API calls

Ok, thanks! So then in order to avoid a single point of failure, would a serverless function be the best choice as this is the most resilient / resistant to attack?

I am willing to learn what I need to - just need some help making a choice about the specific technologies that should be used.

yeah, that should probably be the easiest if your application doesn't already have a backend.

Heroku would work fine too honestly. because they are able to scale up by spinning up more instances

A few other services I've heard of for serverless functions are AWS Lambda, Vercel, and Serverless (literally called "Serverless"). Do you have any insight about those and how well they would work for a LiveKit token/room management server? Sounds like AWS Lambda is fairly complicated to use but Vercel abstracts-away a lot of it?

<@U04DUL22NF5> Vercel is great, Firebase, Supabase, Cloudflare, Serverless, Lambda, <https://sst.dev/|this> might also work well as an abstraction for Lambda

Great, thank you! :slightly_smiling_face: If you had to narrow it in, are there any preferred, or "go to" options that the LiveKit team typically chooses? (regardless of server vs serverless)

we usually spin up our own backends :see_no_evil:

~which frontend stack are you using <@U04DUL22NF5>?~ Oh, i see now it’s Unity WebGL

how are you serving the Unity WebGL page/player?

It's served as a static site via an Amazon AWS S3 storage bucket, which is pointed to by a full-page iFrame on our site.

here’s a quick example of an endpoint that just generates tokens based on what you pass in:
<https://replit.com/join/lbmfjqgfxe-russdsa>

note: this is not for production usage, but it’s fairly straightforward to move this to a serverless function hosted by some provider

separately we’re discussing how to make this whole thing easier, at least while in dev mode :slightly_smiling_face:

Also if you were curious about our platform, we have some demos available on our site at <http://simulacra.io|simulacra.io>. My goal with LiveKit is to first integrate it for screen share onto a 3D in-world screen. Then eventually migrate voice chat from Agora to LiveKit.

That's correct Matt. You don't actually need to use any of the room management APIs. By default rooms are automatically created when the first participant joins.

If you eventually build moderation capabilities that requires you to, for example, remove a participant from the room, or to close the room entirely. You could use the server APIs for that.

by default there is no limit to maxParticipants, and emptyTimeout is 5m

Great, thank you <@U01KZ61GV5Z>! :slightly_smiling_face:

yah. it's <https://github.com/livekit/livekit/blob/master/config-sample.yaml#L138|here>

Just trying out the example from the docs:
```&lt;LiveKitRoom token="my-token" serverUrl="<wss://my-livekit-server>" connect={true} &gt;
  &lt;VideoConference /&gt;
&lt;/LiveKitRoom&gt;```

Which package are you using <@U02CEMTDG2E>?

Ok it looks like components-js! cc <@U02EU2ZPTQU> 

components-js ’s doc still in alpha
i’m watching source code right now

that’s unexpected! the meet app uses the the exact same pattern as the one you are referencing.
When running one of the examples (E.g. `minimal` example) - which also uses that pattern, do you see the same error ?
also the `next.config.js` in our `example` folder has some overrides to make source maps work in nextjs. having those source maps would be helpful in determining which function `ve` is supposed to be

Right, I’m actually doing something unusual, embedding into a Rails app directly using `htm`<https://github.com/developit/htm>

Probably will switch back to just using next

ah, I remember the thread now!
that sounds likely to be connected, yes! still, if you’d want to try to enable source maps within your rails app, that might give a hint to where/why things are breaking unexpectedly

Where can I find more resources and clear sample code about js-client?

Getting the volume of each track is still in the feature list in flutter/native in our roadmap, maybe it will be implemented in the next 1-2 months,

But there is an alternative, use getStats to get the audio level for track.

Thanks for letting me know, I'll look forward to it and use my crappy hand-crafted solution while waiting for it.

hi!
you could either use `useMediaTrack` or `useMediaTrackByName` to get a reference to a single track.
Or, alternatively filter `useTracks` to narrow things down
`useTracks([Track.Source.Camera]).filter((ref) =&gt; ref.participant.identity === myHostName)`

The rendering of either one of those tracks can be done with the `&lt;VideoTrack /&gt;` component

Nevermind I found it described here. <https://github.com/livekit/client-sdk-js#audio-playback>

Though, I wonder, is this usually required on browsers like Desktop Chrome?

It *can* be required on Chrome. depending on if the user has interacted with the page or not.

OK. So each new page load might require it

Just wondering if I can have a ‘Join’ button in the preJoin room that will do the same thing as ‘start audio’

yup, if there isn't a page reload, that click should do it.

<@U01KZ0SCDKL> thank you for your answer,

I took the example of livekit for react-native and I have to continue on that and I start to change the code  or there is another method recommended by you ?

cc: <@U02EU2ZPTQU>. There is code in the original thread.

what bitrate did it ultimately send? which version of Safari (is it mobile/desktop?)

I think there are two competing `videoEncoding`settings. The `preset` that you define further down also has a `encoding` section, which most likely overrides the one above.
Can you test by specifying your own custom preset with the desired encoding parameters?

<@U01KZ61GV5Z> This is the last version of safari on desctop. The real bitrate was 1200+ kbts

<@U02EU2ZPTQU> I will test if you change my config and give it back.  I didn’t get the idea and I’m afraid I’ll start changing the wrong thing

ah, I misread your config! what I said doesn’t make sense, sorry!

One more things, i am speaking about the Hight quality, if i resize video container to smaller dimension simulcast layer works

from the previous thread
&gt; `maxBitrate: '600_000',`
it looks like you’re setting maxBitrate as a string, while it should be a number

<@U02EU2ZPTQU> Yep! It's works. I'm shocked. This option has been around for a very long time. And I took it somewhere from an example or code. I'm very sorry for worrying

you'd want to set the timer in `TrackSubscribed` to measure downstream tracks.

it is for when the room is created and no participants join

Thanks, that’s what I suspected. So any chance to modify that 30s timeout if someone joins and then disconnects? To allow that user to switch the browser, login to app and join the stream again ?

currently there isn't a way to adjust that timeout. Do you want to file an issue <https://github.com/livekit/livekit/|here>? If there's enough interest from folks, it's an option that can be added

Ok, thank you for info. I’ve created a ticket (<https://github.com/livekit/livekit/issues/1776>)

do you have a code sample of that reproduces this? does this happen with the react-native example app?

```const [room] = useState(
    () =&gt;
      new Room({
        publishDefaults: { simulcast: false },
        adaptiveStream: true,
      }),
  )
  const { participants } = useRoom(room)

  useEffect(() =&gt; {
    let connect = async () =&gt; {
      await AudioSession.startAudioSession()
      await room.connect(VIDEOCHAT_URL, state.myToken, {})
      await room.localParticipant.enableCameraAndMicrophone()
      setConnected(true)
    }
    if (state.myToken) {
      connect()
    }

    return () =&gt; {
      onExitChat
    }
  }, [state.myToken, room, onExitChat])```

is the same room object being used both times it connects to the server? or different room each time?

could you verify whether it is different? should be as simple as doing something like:
```let connect = async () =&gt; {
  console.log("room check: ", room == global.room)
  global.room = room;
  // ...
}```

could you also turn on debug logs with:
```import { LogLevel, setLogLevel } from 'livekit-client';

setLogLevel(LogLevel.debug);```
and post the logs of what happens when you go through the repro steps?

livekit.log

so, I think it's because we are creating new Room object in react native.

hmm upon reconnect, can you see participants if you check the room object directly with `room.participants`?

I logged room.participants and I can see only me. I don't see other participant.

to be clear, the website sees the react-native user join, just the react-native user can’t see the website user?

can you verify if `room.sid` is the same between connects?

yep, website can see react native user when rejoin, but react native user doesn't see any user when rejoin.

what’s the livekit-client &amp; livekit/react-native versions being used?

```"livekit-client": "^1.5.0",
    "livekit-react-native": "<https://github.com/livekit/client-sdk-react-native>",```

can you post what’s in your package-lock.json/yarn.lock file for those?

so, when rejoin, it creates token again, but room name is same.

writing up some instructions for something to try, but in the meanwhile can you see if using livekit-client 1.8.0 changes anything?

1. `yarn add <https://github.com/livekit/client-sdk-js#v1.8.0>`
2. `cd node_modules/livekit-client`
3. `yarn`
4. Add `console.log("response:", resp);` to SignalClient line 275. This will let us see the messages sent from the server directly.
<https://github.com/livekit/client-sdk-js/blob/v1.8.0/src/api/SignalClient.ts#L275>
5. `yarn build`
6. Repro and post logs.

I upgraded the livekit version and followed the instructions you sent.
But I got error in "yarn build"

```$ rollup --config --bundleConfigAsCjs &amp;&amp; yarn downlevel-dts
[!] RollupError: Could not resolve entry module "rollup.config.js".
    at error (/Volumes/Work/upwork/toptal/NERD/exy-react-native/node_modules/livekit-client/node_modules/rollup/dist/shared/rollup.js:274:30)
    at ModuleLoader.loadEntryModule (/Volumes/Work/upwork/toptal/NERD/exy-react-native/node_modules/livekit-client/node_modules/rollup/dist/shared/rollup.js:24301:20)
    at async Promise.all (index 0)```

when I checked room.sid, it was different object.
Also, when rejoin, it didn't show other participants.

Very weird that the sid is different though

try <https://github.com/davidliu/client-sdk-js#dl/v1.8.0_logging>

zsh: no matches found: <https://github.com/davidliu/client-sdk-js#dl/v1.8.0_logging>

when I run following command
```yarn add <https://github.com/davidliu/client-sdk-js#dl/v1.8.0_logging>```

is this visible to you? <https://github.com/davidliu/client-sdk-js/tree/dl/v1.8.0_logging>

try just replacing the livekit-client package in your package.json directly with that url

error <https://github.com/davidliu/client-sdk-js/tree/dl/v1.8.0_logging>: Extracting tar content of undefined failed, the file appears to be corrupt: "Invalid tar header. Maybe the tar is corrupted or it needs to be gunzipped?"

I think `yarn add` requires a specific format to take in a repo, easier to just replace the line in the package.json directly

"livekit-client": "<https://github.com/davidliu/client-sdk-js#dl/v1.8.0_logging>",

so, this issue happens on ios real device, too

looks like the logs from metro are getting shortened and missing info, can you grab the logs from xcode or console?

hmm it looks like they’re different rooms completely, with different room names as well. If you’re able, try verifying that the tokens are generated with the same room name

yeah `9e1ca7fee9d2` on first connection, then `b3011fdc65ee` on second connection

oh, thanks. probably, it might be the problem.
Thanks for your support.

telegram-cloud-photo-size-2-5310221282316502890-y.jpg

is the camera mirrored if you use <http://example.livekit.io|example.livekit.io> or the example app contained in the sdk?

you can post process the video by adding a `videoProcessor` to createVideoTrack to flip the image, but I’m not sure if there’s anything to prevent the rear camera mirroring from happening in the first place. this is actually the first I’ve heard of it happening for the rear camera and I can’t find any resources on it

<@U01NZKTNE1H> We are observing the same issue with livekit-client js sdk for all of the mobile cameras.

<@U045CJPACG1> is this the back camera? or front facing camera as well?

back camera. Never had a issue with front camera. Issue always persists with back camera.

<@U045CJPACG1> is it mirrored if you try something like this? <https://webcamtests.com/viewer>

is the mirror happening when you view the camera locally, or only mirrored for remote viewers?

On both. when local.partcipant ( view camera locally) stream we get mirror effect as well. As the local stream is mirrored , we assume the same stream is forward for remote participants as well

tested on <http://example.livekit.io|example.livekit.io> as well?

&gt; is the camera mirrored if you use <http://example.livekit.io|example.livekit.io>
I’ve tried both devices and all of 4 cameras operating normally without mirroring.

Yes as <@U052TD1AABE> mentioned, it worked in latest <http://example.livekit.io|http://example.livekit.io>.

<@U052TD1AABE> I’ll create an app for diagnosing, found some flags for mirror mode deep in the camera stuff

&gt; is the camera mirrored in the example app contained in the sdk?
I’ve tried to connect to our stream from the SDK sample and it works correctly. Also, I tried the same in our app - everything works correctly. I already noticed this behaviour before - the issue is not permanent

<@U045CJPACG1> your issue may be in your own code potentially then; we don’t do any camera mirroring on the js client AFAIK. can continue to help you out, but will need you to post a separate message to <#C01KVTJH6BX|helpdesk> since it’s a separate client sdk

I noticed a new issue with the front camera on Pixel 7 Pro. 
It reflects the device rotation

This is the video from android sdk sample and the comparison to default camera app

The same issue I have on Samsung device. It’s a new one, I haven’t seen it before

ah, I can repro on my device. can’t believe I never noticed this

That’s fine :slightly_smiling_face:

What about the main issue with Pixel 7 Pro back camera - it is not reproducible for now, unfortunately. Although, I didn’t update the SDK and didn’t change anything

ick. maybe device restart fixed it? :sweat_smile:

<https://github.com/davidliu/Camera2VideoMirrorMode|here’s the test app I made for testing out mirror modes>

by default it (and our sdk) uses MIRROR_MODE_AUTO, <https://github.com/davidliu/Camera2VideoMirrorMode/blob/main/Camera2Video/app/src/main/java/com/example/android/camera2/video/fragments/CameraFragment.kt#L367|configured here>. Would be good to check if it’s disrespecting the mirror mode in a separate app, as well as if it disrespects MIRROR_MODE_NONE

I made some tests.

Everything is fine with back camera on Pixel 7 Pro with both AUTO and NONE mods.

But if I switch from AUTO to NONE and start front camera the picture rotates as in my video <https://livekit-users.slack.com/archives/C01KVTJH6BX/p1682506881662249?thread_ts=1681233896.151589&amp;cid=C01KVTJH6BX|above> (wrong direction).

ahh, for front cameras, the front mirroring is removed. this will need to be reapplied for local participants using front cameras to fix

<@U01NZKTNE1H> thanks, will check from our codebase further  and confirm

the renderer views each have a `setMirror` method that can be used as needed

Unfortunately that would require a lot of extra plumbing on our end. What’s your goal, though? Ideally you can learn whatever you are curious about from the cloud dashboard

(So if there’s something you’d like to see there that isn’t there yet, let us know!)

for debugging purpose, debugging just got easier by reading logs

Hello <@U03NBU0MGA1>, May I ask, how are you generating your Token?

Hi <@U032TUAL4KT>

``` .\livekit-cli.exe create-token --admin --room "room1" --identity rubi  --api-key xxx --api-secret xxxx```

As a side note, try to threading your issues, if you don't mind.

the issue is likely your token has expired. if you are trying to generate a token for testing, try using `--valid-for 72h` to create a token that's valid for 3 days