LiveKit Community

Nice, let me know when you have created the repository :smile:

Neil, I could also create a repo in the livekit org, if it makes it easier for folks to contribute to.

yeah that's probably a better place for it

<@U01KZ61GV5Z> can you make this repo and give me write permission or ownership for it?

hey, I haven't ported my scene into a sample scene yet. I iterated in my own app to go faster, what issues are you seeing?

I'm using this code:
`RoomController roomController = GetComponentInParent&lt;RoomController&gt;();`
`await roomController.Join(token);`
`RenderTexture rt = await roomController.PublishRenderTexture("Test", "idk what is clientId", 1280, 720);`
`List&lt;MediaStreamTrack&gt; tracks =  roomController.GetSubscribedTracks();`
`foreach(var track in tracks){`
       `print(track);`
`}`

Can you explains me what is the clientId ?

client id is just a user-defined id used for bookkeeping so that livekit server responses can be correlated to the local publisher

Also, I'm getting this error on the console of the livekit-server
`2021-05-19T22:21:18.101+0200    ERROR   service/roommanager.go:366      could not decode trickle        {"participant": "test_participant", "err": "invalid character 'c' looking for beginning of value"}`
`<http://github.com/livekit/livekit-server/pkg/service.(*RoomManager).rtcSessionWorker|github.com/livekit/livekit-server/pkg/service.(*RoomManager).rtcSessionWorker>`
        `/home/feeps/livekit-server/pkg/service/roommanager.go:366`

can you try capturing from a camera instead of a render texture? I did my testing on a camera

<@U022KB72NFJ> the server error indicates you are not sending correctly encoded messages

image.png

I just changed the code to publish a camera instead of a rt

<@U01KZ61GV5Z> The error printed in the console is followed by a success message, not sure what to think

the encoding error is on the client.. the success messages from the server (i.e. server sent ice candidates to the client, but what client is sending to server could not be decoded)

hrmm I'm getting this too now :thinking_face:

ok fixed, the candidate in the trickle is supposed to be a json object and not a string, pushing up now

one of those "how did this ever work?" situations

<https://livekit-users.slack.com/archives/C01LPRVTQMU/p1621751561000400>

Any idea how one could do this on ContainerOS?

There's no `sysctl` command :thinking_face: I do see `/etc/sysctl.d` and I can edit that, but wonder how I can make it take effect immediately

Without needing to restart  the server container that is

good to know it works :slightly_smiling_face:

it doesn't hurt to modify it also.. we haven't tested the throughput differences with that.

Done :+1: will let you know if my users run into issues

for reference here the default on containerOS seems to be `212992`

maybe sufficient for ~5mbps of total inbound data

The good news is on containerOS it seems a docker image will inherit this from the host os. I execed into the running docker image and the rmem and wmem values mirror the ones I just set in the host

with LiveKit 0.9.8, it will warn users upon startup if receive buffer is too small

ooh very nice. I am on 0.9.0 right now, need to wait for a slower time to upgrade, seems like some users are on right now

:pray:  our multi-talented <@U01KZ0SCDKL> came up with it

<https://webrtchacks.com/facetime-finally-faces-webrtc-implementation-deep-dive/|https://webrtchacks.com/facetime-finally-faces-webrtc-implementation-deep-dive/>

Pretty interesting to see some of the features Apple is using with webrtc for FaceTime 

<https://sample-js.livekit.io/> is down ?

hey, yeah we've deprecated it in favor of <https://example.livekit.io>

you can still run the sample client locally, `yarn sample` in the JS repo

Image d’iOS.jpg

Image d’iOS.png

Yikes, do you mean the second screenshot? What’s off in the first one? Or do you just mean, in general?

second one is menu on mobile. guess we gotta fix that part

just pushed up some fixes for the styling issues you're seeing on mobile <@U022KB72NFJ>

I mean in general, the red line is a fail when I made the screenshot 

The images are aggressive because of the contrast but that’s my personal opinion :slightly_smiling_face:

Yeah, if we stick to that dark theme, we should (and will) make those images more legible.

it looks good now, good work :slightly_smiling_face: !

are you concerned someone getting the token from the web page that is rendered to your user? I think this depends on how you are delivering the token from your backend to the client

For me the website will be a place where you can join or create a room. As soon as one of these 2 actions is executed, the server sends an access token to the client who will then join a livekit node

The user must be logged in to make these actions

so it seems that it's extremely unlikely for someone else to get ahold of that token, assuming you are using HTTPS

on the livekit side, if you are deploying the server behind TLS (as we recommend), LiveKit SDK will only send the token to the server once, and that's over HTTPS/WebSocket

Yes, that seems very unlikely that someone steal a token. Is it viable to put the accountid on the participant metadata ?

you def can, though I'm not sure if it'll help. if someone was able to get ahold of the token, they will be able to connect as that account as far as livekit is concerned

it's sort of like a session cookie a server sets on a browser session

Yes I know :confused: I will assume that the access token will not be stolen. It's like stealing cookies from a browser, I don't think there's a lot of solutions to prevent that

@here we’ve updated a couple of our repo names for consistency - `server-api-js` is now `server-sdk-js` and `livekit-sdk-go` is now `server-sdk-go`

The website is awesome ! You did a great job guys :grinning:

Thanks <@U022KB72NFJ>! just in time for the launch later this week

LiveKit shouldn’t redirect a user to another node of another region so I think the answer is yes but I’m not sure 

I don't think it's strictly necessary.. I think what we would want is region awareness in the routing itself

the only thing going over redis are signaling messages, where we have a much higher degree of tolerance for latency

what matter is for the NodeSelectors to pick out a node for the room in the right region

Hey Folks! We are LIVE on ProductHunt: <https://www.producthunt.com/posts/livekit>. Give it a :thumbsup: if you have an account. If you’d like to share your experiences with it, that’d be amazing! :pray:

Woah blast from the pace, I think that 2011?

lol, did slack just pick up on a gravatar?

Hey <@U01KZ0UETNW>! It feels like a life time ago when you mentioned I talked you into playing with Pion

Yeah it does, last year I’ve spent a lot of time with the ion / ion sfu projects until livekit

awesome, let us know if you need any help. what are you building with it?

That'd be amazing! We haven't put HPA through a full cycle of tests yet.