LiveKit Community

Not that random, and we’ve talked about doing something like this! Would love to hear if there are other folks interested. :blush: 

What sorts of admin tasks did you have in mind, or that would be helpful <@U046CJ4SRGU>?

<@U01KZ0SCDKL> The things I would find helpful might become more apparent when I'm up and running with livekit after e2ee is supported (currently waiting). My main thoughts are around no-code Observability, management and maintenance.

On initial thoughts, basically a high level dashboard that lets me monitor the health of the deployment, wraps server api's you are already providing (view and manage rooms and participants). Be able to change configuration easily (webhook urls etc) and most importantly be able to basically have a UI that can start and help manage upgrade deployments (might require some decoupling of the admin dashboard service from the main deployment). It doesn't have to look fancy, just enough to be able to control the deployment without the need for custom code.

I guess the idea would be to lean into the Livekit's goal to abstract as much of the pain points of running your own webrtc stack so that a single person to manage the deployment without any custom code.

For inspiration, along the lines about how dagster dashboard does it where they provide a UI admin panel that users are expected to use to control and manage their deployment. (see <https://youtu.be/UrjHVJ2GZN0?t=507>) I have linked to the exact time you can see their dev experience.

Basically, have it so once livekit has been deployed. The admin panel can be used to easily access all the server apis, configuration settings and upgrading so you never really have to think about writing custom code or handling deployment upgrades (almost like running a self hosted webrtc saas).

9b486984-a996-4f39-81a3-9664a0bbd84d.png

something like this that we can log into as a backend self hosted admin dashboard :joy:

but with additional functionally that includes the management and maintenance as mentioned above.  <@U01KZ0SCDKL>

aka basically a ui wrapping helm deployment upgrade commands i guess

On second thoughts, If you are using this dashboard as a propriety feature, I wouldn't be against self-hosting the deployment (for compliance), but piping the telemetry to Livekit Cloud and pay to get the cloud dashboard experience. I would do this in a heartbeat.

There would also be some kind of auth between the self-hosted deployment and my livekit cloud account so I can issue management and maintenance commands as well (as mentioned above).

<@U046CJ4SRGU> thanks for sharing all the detail here!! I’ll check out the Dagster video once things have calmed down towards the end of the week. Pretty much since we started working on Cloud, we’ve been thinking about how we can offer/package the analytics for self-hosted projects. As one option, we’ve talked about exactly what you mention too — being able to link self-hosted deployments to a Cloud account. All this will require some more thinking, but I’ll be sure to keep you updated and will likely have more questions for you in the coming months. :slightly_smiling_face:

<@U01KZ0SCDKL> Yeah there's no rush! I can imagine the rush the team is going through right now haha.

Yeah  that would work, I need to be in control of the data through the my own infrastructure (for compliance and with e2ee), but I'd be happy to have the Observability and maintenance connected to Likekit Cloud.

Hello <@U046B0LNLD7>, screen sharing is already available.

Have a glance at this example <https://github.com/livekit/client-sdk-js/tree/main/example>

You can also have a look at the doc here about screensharing: <https://docs.livekit.io/guides/room/publish/#screen-sharing>

if you’re already running a livekit server, you can also use <http://example.livekit.io|example.livekit.io>, which has screen sharing available

Thanks <@U032TUAL4KT>, I think what <@U046B0LNLD7> is talking about is the default playground code doesn’t integrate the screen sharing APIs/functionality. He’d like to use playground for some quick tests with a customer. <@U046B0LNLD7> updating playground is on the list of TODOs, but probably won’t happen for at least a month.

Oh my bad. <@U01KZ0SCDKL>, you seems to be right on the dot about it. ahaha oups, after rereading, that makes more sense :slightly_smiling_face:

Thanks for your input. Yes what <@U01KZ0SCDKL> is saying is exactly what we're looking for. Thanks for the update on the timeline.

Hey Karthick, you can actually just use data packets to broadcast to the room! <https://docs.livekit.io/guides/room/data/#data-messages|docs here>

I'll whip up an example to publish by next week

Here it is. a full conferencing example including chat: <https://github.com/livekit/meet>

congrats guys! you are doing an amazing job with Livekit :heart:

Congrats to the whole team! Wishing the team amazing success with this new feature! :partying_face:

Congrats! I'm sure this is going to blow up.

Thank you so much for your support everyone! :pray:

We now have a <#C048FRL1N2C|> channel for any Cloud specific questions or issues.

Nice. We're trying to come up with a migration plan as soon as our product get traction.

Let me know if we can be helpful when you get there <@U032JMDR8TD>!

&gt; 100ms of latency.
This a jab at <https://www.100ms.live/> or just a coincidence

Definitely not a jab or reference to them. We don’t do that sort of thing here.

100 milliseconds is kind of the bar/benchmark for true realtime latency/performance. I’m willing to bet that inspired Kshitij and team’s choice of company/product name. :wink:

thanks Max! feels like we've been talking about this forever.

I appreciate your advise on "nothing shared". we ended up taking a very similar approach to that. :pray:

load testing instructions here: <https://docs.livekit.io/oss/benchmark/>

<@U01KZ61GV5Z> For Dashboard setup any Docs is available?

<@U01KZ61GV5Z> How to enable the metric part? .To use the prometheus

please see docs here: <https://docs.livekit.io/oss/deployment/#configuration>

<@U01KZ61GV5Z> In livekit prometheus metrics, i want to connect the metrics to grafana dashboard any posible ways to connect this??

Hey, you can use the metrics we export just like any standard metrics.  We provide a template grafana dash here: <https://github.com/livekit/livekit/tree/master/deploy/grafana>

<https://docs.livekit.io/oss/deployment/vm/>

this should have the steps to get it set up on a VM

yea, but when running the docker command, its asking a turn domain name. so we need to setup a turn server on our own, is it that way?

livekit has an embedded turn, but for TURN/TLS it needs to have a domain name

it helps you set up a TURN server, and is asking you which domain you plan to use with it

So during docker run, first asks a primary domain name, then a turn domain name. So what i understood is, we have to create a domain name which corresponds to the primary domain name, then create a sub domain for the primary domain, point it to our livekit server(vm) and assign it to the turn domain name field. is it like that?

you'd point both domains to the same address. they could be subdomains. i.e. `<http://livekit.yourdomain.com|livekit.yourdomain.com>`, and `<http://livekit-turn.yourdomain.com|livekit-turn.yourdomain.com>`

i have a subdomain as <http://abc.xyx.com|abc.xyx.com> which have its ssl certificate. so what will be the primary domain and turn domain in this case, i have to give.

you cannot use the same domain for both turn and primary

Awesome write up!! Always nice to read how the underlying structure works!

PSA: JS SDK v1.5.0 has been released. We recommend upgrading as it has a compatibility fix for the newly released Safari 16.
Safari 16 broke getDisplayMedia requests that specify an `ideal` dimensions constraint. Full release notes: <https://github.com/livekit/client-sdk-js/releases/tag/v1.5.0>

That was really fast! Nicely done!

Let us know if you have any feedback about the migration process! We'd love to make it easier.

Apologies I have no feedback other than it was easier than Agora :stuck_out_tongue:

haha, that's fantastic to hear! Dev exp is a primary design goal for us

Assuming javascript:
```const room = new LiveKit.Room()
await room.connect(serverUrl, token)```

There's a full working example here: <https://github.com/livekit/client-sdk-js>

what do you mean by `before I will enter in room`  in terms of LiveKit connection state?

iOS does not allow you to use ear piece speaker unless the AudioSession is set to `playAndRecord`

Yes - You are right - I use mode `playAndRecord, but I need that when user connected to room - voice will be hearing from ear piece speaker`

are you managing audio session by yourself?

but it is No results - please give me some breadcrumbs - where to go&amp;

You can run the core server locally in principle. But probably your web app will become slow. So you do need a separate server.

Hello. Please I need help with screen sharing. My screen is showing in inverted mode when I share screen in conference room

image.png

Hmm appears it was a codec issue. h264 not supported my chrome clients? :confused:

Hey Mark, I think you meant LiveKit instead of ion?

We'll look into this. I think the Chromium version that we are bundling doesn't support H.264.

Filed issue <https://github.com/livekit/egress/issues/165>. we'll look into this one today.

<@U01KZ61GV5Z>, does that mean Livekit does some form of translation? SFUs as far as I've known, everyone had to use the same codec

<@U044UD10VPE> Some clients send h264 and some send vp8. Livekit doesn't transcode, and I don't believe there's any requirement that they're all the same (unless you use enabled_codecs on the room)

<@U044UD10VPE> just to confirm, this only happens when Egress is deployed on arm64 instances. On amd64, we are using Chrome, which has no problem decoding H.264.

<@U01KZ61GV5Z> im on an Intel(R) Xeon(R) CPU E5-2678 v3 @ 2.50GHz for the server, and intel 9900k on client

<@U01UB0LHTTQ> So that was the issue! We were explicitly enabling vp8 on the startup config! after disabling that it all worked as expected. Thank you!

I ran some 50 person load tests on calls today. It was interesting toi see the loads not handled well on older laptops/hardware

are you talking about Egress specifically? or are you talking about the core server?

a lot of it depends on the client.. and how much data each person is subscribing to

I might not be the best person to answer, but our SFU started as a pion fork, so it might be worth trying. <@U02CM28PB4Y> or <@U01KZ61GV5Z>  might know more

We are probably not compatible.  Need to take a look, We used very specific portions on ion-sfu which did not involve protocol/wire messages exchange.

Looking at <https://github.com/ionorg/ion-sip|ion-sip>, the repo looks empty. So, unable to ascertain what it will take.

<@U02GAC58A79> I see you are the committer on that repo. Do you know the status of that project?

<@U02CM28PB4Y> Sean has an example app too : <https://github.com/pion/example-webrtc-applications/tree/master/sip-to-webrtc>

for sure, you can use this example with LiveKit using the Go SDK.

but as examples go, they will take you 50% of the way there. there'll be a lot of cases to handle for a production-grade solution

<@U02CM28PB4Y> This <https://github.com/ionorg/ion-sip|ion-sip> repo is currently suspended due to no manual maintenance, but I plan to try to develop sip-gateway for livekit at the end of next month (January 2023),  also based on <https://github.com/cloudwebrtc/go-sip-ua>,  <@U044UD10VPE> if you have any idea about the function in sip gateway, welcome to join the discussion.

Boom! Thanks <@U02CM28PB4Y> and <@U02GAC58A79> !

We have not seen any issues with WebSocket in corporate environments. I'm wondering what you are seeing?

<@U01KZ61GV5Z> we have a lot of experience installing our product into bank's on premises networks..
Pretty much all of them don't allow Websockets within their corporate networks because they deem them insecure. (mostly because their firewalls can't handle them).
While we use Websockets for our cloud product 95% of our on premises setups have to use comet :disappointed:

interesting.. it's deemed insecure even when deployed as WSS? If it's deployed over HTTPS, there shouldn't be any differences between a websocket connection vs HTTP long polling to the firewall.

While it's possible to switch to another transport for the signaling messages, it'd be a major lift.

I'm not saying this makes sense but here's an outtake of an official statement by one of the worlds biggest banks:

&gt; *Why Websockets have been disabled?*
&gt; *Risk*
&gt; Websockets have been disabled as a part of &lt;XXX&gt; compliance as they are vulnerable to different types of security attacks as below.
&gt; • *No authentication during the handshake process* - WebSocket protocol doesn’t let a server authenticate the client during the handshake process. Only the normal mechanisms for HTTP connections are available. That includes HTTP and TLS authentication and cookies. The upgraded handshake still occurs from HTTP to WebSocket. But, the HTTP sends the authentication information directly to WS. This can be exploited through Cross-Site WebSocket Hijacking.
&gt; • *Unencrypted TCP channels* - WebSockets can be used over an unencrypted TCP channel. This leads to issues that are listed in the OWASP Top 10 A6-Sensitive Data Exposure.
&gt; • *Vulnerability to input data attacks* - Input data attacks could lead to common yet dangerous attacks such as Cross-Site Scripting.
&gt; • *Data masking* - Data Masking prevents security tools from actions like identifying a pattern in the traffic. Software like DLP (Data Loss Prevention) isn’t aware of the existence of WebSockets. This makes them unable to perform data analysis on WebSocket traffic. This also prevents these software programs from being able to identify things like malicious JavaScript and data leakage.
&gt; • *Authorization/authentication* - WebSocket protocols don’t handle authorization/authentication. Any application-level protocols need to handle this separately. Especially in cases when sensitive data gets transferred.
&gt; • *Tunneling* - WebSockets let anyone tunnel an arbitrary TCP service. An example is tunneling a database connection directly through and reaching the browser. In the case of a Cross-Site Scripting attack it evolves and ends up becoming a complete security breach.
&gt; • *DoS attacks* - WebSockets let an unlimited number of connections reach the server. This lets an attacker flood the server with a DOS attack. This greatly strains the server and exhausts the resources on that server. Then the website slows down greatly.
We have addressed all of these issues in our product but still they don't want to enable WebSockets. A lot of this isn't logical, technical reasoning, it's mostly politics.

Thanks for sharing this. Agreed this list seem rather arbitrary and questionable technically.

Yes it is but we face pretty much the same reasoning with all financial institutions

right, I'm open to a different mode of signal transport. If you guys are interested in working on it, I'm happy to help with getting that merged.

Thanks for the feedback. We'll definitely think about it! I guess this would mean an update to not only the server and Web SDK but basically all clients right?

I think it's fine to only add it to the web SDK for now. The server can support either transport.

:+1:  I'll run in through internally and get back to you as soon as I know more

Hey you guys seem to have solved the "Bluetooth Headset Microphone as Audio Source Using Flutter" problem that I've been stuck on for weeks.    Is there any way to use flutter_webrtc to just pull audio data from a bluetooth headset mic?

Saw your recent repo on <http://meet.livekit.io|meet.livekit.io>. NIce! Any reason you chose Next.JS this time as opposed to React?

Hey William, some answers:
1. Yes
2. LiveKit 
rooms horizontally scale, more details here: <https://blog.livekit.io/scaling-webrtc-with-distributed-mesh/|https://blog.livekit.io/scaling-webrtc-with-distributed-mesh/>

Thanks a lot <@U01KZ0SCDKL> :raised_hands:

Could you please share with me the timeline about the C++ client SDK?

Uh a C++ SDK would be something we're also interested in

<@U01KZ0SCDKL> have you already started with the C++ framework?
We are currently doing this with a combination of FFMPEG and libwebrtc on C++. If you have a preview we'd love to have a look / maybe contribute if you'd like to.

Hey, we’re currently building a Native Unity SDK on top of a common core written in Rust. It should be available at the end of this year. We’ve already planned to build a C++ SDK for early 2023 using the same core 

<@U022KB72NFJ> let me know if you need help / somebody to review with the C++ SDK

Hi! Just joined here to ask this question :wink: ! Awesome your building a Unity Native SDK.

Hi everyone ! I'm learning to use the go sdk provided by livekit, butI'm currently not sure how to use it as a client to manually subscribe.

<@U03GFRSKKST> the stuff you’re building looks cool, but this post feels a little bit like an ad/marketing and not appropriate for this community. If Atmoky has a LiveKit integration for spatial audio and someone is asking about that feature, I think it’s fair to refer them to your site. Thanks for your understanding.

Hi Russ, it was not my intention to do ads here. Just think this is a great fit of complementary tech. However I fully understand and respect the rules on this server. Happy to stay in touch!

it worth taking a look: <https://twitter.com/littleredspam/status/1588640168122146818>

are you using the swift SDK? is there audio being published?

and how are you "selecting earpiece as output"?

I resolved this issue by upgrading flutter_webrtc sdk to 0.9.12