LiveKit Community

Screenshot 2021-08-26 at 9.58.34 PM.png

<@U02CAANFV25> what are the repro steps to get the gray video seen above?

1. create two participant token for same room 
2. open 2 different tabs from chrome browser and launch the app 
3. login with your token and urls first time then you have seen the issues 

the audio feature also not working properly

that's a known issue.. as mentioned in the other thread, we are working on it

for those that are interested in WebRTC, I wrote a post about simulcast and its inner workings: <https://blog.livekit.io/an-introduction-to-webrtc-simulcast-6c5f1f6402eb>

I had asked about this as well. It can be done with the serversdk, but unfortunately not with the client sdk: <https://livekit-users.slack.com/archives/C01KVTJH6BX/p1629331766106300>

This answers my question perfectly thank you! :slightly_smiling_face:

:wave: Hi everyone! :slightly_smiling_face:

Happy to chat more about this over DM (as I have with <@U0291E7JEP3>), shoot me a message! :)

hey Lois! Excited to hear more about your project and let us know how we can help or things we can improve! :pray:

Thanks <@U01KZ0SCDKL> I was able to get through most of the get started section today and wire everything up with docker-compose to easily stand up and down all the apps.

I’m sure I’ll run into issues soon though :smile: 

I believe recording is in the GoLang server sdk via its WebRTC implementation

"The Participant SDK gives you access programmatic access as a client enabling you to publish and record audio/video/data to the room."

we are working on recording currently, and it's almost ready to release. we ended up going with the approach of a web recorder (that uses an instance of headless chrome). This allows you to record any customizations/compositions of your UI.

some early docs are here: <https://docs.livekit.io/guides/recording/>

we'll be releasing this officially in the coming weeks

<@U027N5NE7K9> what are you building? would love to check it out too!

thank you so much, that is fantastic news!

<@U01KZ61GV5Z> in short, that’s a knowledge sharing platform that helps content creators to make some money and grow their audience

it’s still in the development, but as soon as MVP will be launched, I will love to share it with you

that’s not a pet project, we are a startup with team working on it, so I hope we will deliver it by December;)

that's awesome! excited to see it when you launch!

it's a bit tricky to tunnel with ngrok for WebRTC because the server itself has to know the IP to use to advertise to clients ahead of time. AFAIK, ngrok tends to assign a hostname after the tunnel is established.

the best way to test is to open up ports on your router, and start livekit with <https://github.com/livekit/livekit-server/blob/master/config-sample.yaml#L30|rtc.use_external_ip> set to true

Ok I changed the approach and it works! Now I have the server running on my laptop and the port forwarding configured on my router. Thanks for your help :+1:

Hey Marsouin! Thanks so much for the kind words. :heart:

The server and SDKs will be open source, forever. There will be a paid solution, but it will not negatively impact (probably the opposite) the open source offering.

We don't have any plans for a Vue SDK yet. We'll get to it eventually, but you're also more than welcome to contribute one!

<@U01KZ0SCDKL> thanks for enlightening me!
I definitely will share our Vue code as soon as we have something up and running.

welcome J! we are very interested in how it stacks up as well. we are particularly focused around quality/connectivity and performance. hoping it'll work out for you guys

Raja mentioned your history together too (small world!). Looking forward to chatting further!

let me know if you’re up for a quick call — it would be good to learn more about your future plan and share ours as well. if you’re up for it, feel free to find a time via <https://calendly.com/j-pop/30min> (i’m also available over the next couple of hours even though the calendar may not show those as options).

I'm also interested in these findings. We used mediasoup early on and the perf was really great, but the documentation/support was not so great.

Same here! I read somewhere that Pion doesn’t have great congestion control functionality unlike Mediasoup.

<@U02EJ2AP755> Just to clarify here, we do handle congestion control within LiveKit. Pion is a WebRTC implementation that makes it possible to handle all forms of RTCP feedback, which we are.

Separately, Pion 3.1 introduced native handling of TWCC, it's pretty exciting stuff that we plan to check out and potentially migrate to instead of our current TWCC.

While our examples cover files, the same method would work with other sources. Pion has some great examples of this with <https://github.com/pion/webrtc/tree/master/examples/rtp-to-webrtc|rtp> and <https://github.com/pion/example-webrtc-applications/tree/master/rtmp-to-webrtc|rtmp>.

<https://github.com/livekit/livekit-server/releases/tag/v0.13.0|v0.13.0> has been released :tada:

:wave:

Hey Ilya, I've never really used Hopin. But, I just read about their networking feature. It reminds me a bit of what Honk has been trying in consumer. You certainly could build this using LiveKit.

I'd imagine a very rudimentary version might keep track of all users or attendees in a database somewhere and then clients would randomly shuffle them into LiveKit rooms.

An alternative implementation (sans separate database ops) might place all attendees in a single room, and then leverage selective subscriptions to pair attendees with one another.

we’re going to try to build interviewing into our platform too

That's exciting! Def let us know if you need any help/guidance!

<https://www.figma.com/proto/hEe4eNwUVboxgBVmSdMNr0/Live-Interview?page-id=2%3A2&amp;node-id=46%3A338&amp;viewport=251%2C48%2C0.4&amp;scaling=min-zoom&amp;starting-point-node-id=2%3A10|https://www.figma.com/proto/hEe4eNwUVboxgBVmSdMNr0/Live-Interview?page-id=2%3A2&amp;nod[…]=251%2C48%2C0.4&amp;scaling=min-zoom&amp;starting-point-node-id=2%3A10>

adding <@U02FH322XC5> here, the incredible CTO of VanHack

Looks pretty slick! I especially like the e2e integration with the full interview process. Having planned interview questions there and notes is great. Since you know about schedules, for multiple interviews, will the questions other colleagues ask or are planning to ask be there? Would also be rad to have a coderpad or something integrated for engineering interviews. But, really, don't take my feature requests too seriously, you know your product best!

good call about coderpad cc <@U02FH322XC5> on that

“will the questions other colleagues ask or are planning to ask be there?” good idea to add that too!

Just thinking about the types of things I want when I interview candidates. :blush: 

I'm totally fine if you want to do it! My only concern is getting it done for the next release :smile:
Is it something that you'd want to tackle in the next days already?

Yes I'll start working on it right now ;)

<@U01KZ61GV5Z> I failed and I directly pushed to the main repository ...

I did a reset to `d83b7b543d28e6a98d1737be8a7318346779c8fc`

yes! we are looking into that as we speak. The idea is to introduce a region field to `Node`, and introduce a new `NodeSelector` that's region aware

How far along is that and could you use some testing/help with it? I realized that I need to make some changes to my "prefer local node" patch now that the room allocation code moved in v13. If this is somewhat close I won't bother making those changes.

we are pretty close! will likely have a PR this week!

here's the PR: <https://github.com/livekit/livekit-server/pull/135>

Thanks! I'm trying it out and I see how it should work, but I seem to be having an issue. I posted about it here: <https://livekit-users.slack.com/archives/C01KVTJH6BX/p1633923684148800>

<@U02CEMTDG2E> isn't that a DNS level thing?

<@U0291E7JEP3> That’s part of it. The issue is described more here <https://github.com/livekit/livekit-server/pull/93>

Hi everyone, kudos for the great software you're all creating! :muscle:

Flutter users, <@U029G6PA4D7> has just released a major update that's a significant upgrade over the previous version. Stability, simulcast, background audio, etc. More <https://pub.dev/packages/livekit_client/changelog#050|details on pub.dev>

From what I read on the announcement, this is only a Turn server at the edge. When using an SFU, you really don't need a turn server 99% of the time, as it's more used for Peer-to-Peer and LiveKit is an SFU.

Peer-to-peer has many drawbacks, even with a good turn server, the main being that each guest has to encode/upload their video to each other guest, it means you can't scale unless everyone has a really awesome upload speed. Most providers, at least in the US, provide really really bad upload speeds.

It looks like they're working on new APIs focused on streaming, not sure how directly that will compete if they're positioning it as an SFU.

Thanks for enlightening me <@U02BD5MV5NX>!

<https://docs.livekit.io/guides/recording/>

<@U02CPFCCG8N> just a heads up that this currently beta and the APIs are subject to change. Would love to hear your feedback once you have a chance to try it!

Make sure the protocol repo is at the same level as the Android sdk folder

Yes the protocol project is there. I already restarted and invalidate everything. Still the same error

as in your directory layout looks like this?

root:
-protocol
-client-sdk-android

note: the repo layout has been just updated so that the protocol repo is in a submodule. after pulling, ensure that the repo is initialized with `git submodule update --init`

<@U01NZKTNE1H> is it possible to draw something on the screen with the finger and show it on the other side of the meeting?

at the moment we don't have screencast functionality, it's currently in the todo list

But I want to draw on the vídeo coming from the back camera vídeo

ahh, lemme think. we don't support that out of the box

I think you'd need to create a custom VideoCapturer to process the video with whatever you're drawing. we also don't expose the ability to create video tracks with LocalVideoTracks at the moment, so it'd require some tweaking on the sdk side

Ok I will go back on that later. I tried the Android app on a Xiaomi box but it says: Failed to open camera. The trueConf app works fine but I think it's the only one that is able to use correctly the camera on an Android box app

There isn't a better channel for this so...

relevant: <https://www.videogameschronicle.com/news/the-entirety-of-twitch-has-reportedly-been-leaked/>

awesome! the bandwidth needs depends on the bitrate you are using for HD, but assuming 720p, it'd be <https://github.com/livekit/client-sdk-js/blob/main/src/room/track/options.ts#L196|around 2.5mbits> per stream.
if each person needs fetch 8 streams, then the bandwidth requirements is roughly 7 * 8 * 2.5 =~ 140mbps.
FWIW, you would save a lot of bandwidth if using simulcast &amp; using lower res streams for those that are showing up in a small view in the UI