LiveKit Community

the idea is you would use a load balancer that can handle TLS termination. if you are not using one, then 7880 would need to be exposed.

How about if I don't have load balancer before 7880，then 7880 is exposed, right?

like If I want to take h264 over vp8, how to tell client to do that

I don't think this is currently possible in Flutter or mobile SDKs.

Screen Shot 2022-06-20 at 10.58.59 PM.png

nice info! planing to try electron soon :slightly_smiling_face:

Its a voice chat app for remote teams. Kind of like Clubhouse for business. <https://getvocal.app>

Nice. Are you using LiveKit for the screensharing also?

<@U03DUKK5K0T> why did you have a hard time with it? Was there something not clear enough in the documentation? Any words of wisdom to share with someone that’s going to go through the same flow weeks from now? :smile:

Electron.js uses a different library for screensharing. So i had to extend some of the livekit JS classes to get it working

<@U03DUKK5K0T> is this something that can make it to their documentation <@U01KZ0SCDKL>? I would imagine screensharing is right there in the top use-cases with video and audio chats.

<@U03DUKK5K0T> would be helpful if you post about it somewhere, even in their Github with a link here for the rest of us. Thanks!

Yeah, <@U03DUKK5K0T>, I'd love to better understand the workarounds necessary to get screen sharing working in this environment. Happy to consider how we can make this easier for those building Desktop apps with Electron.

cc <@U02EU2ZPTQU> 

<@U01KZ0SCDKL> I think Electron is the go to for desktop apps built with js: Slack, Whatsapp, Skype, Discord, and others too are wrapped using it.

so we’re also considering it for out upcoming product *that might also have screensharing*

so would be great for this part to be in the docs when we get to that point

V1.1 is here , I love that we can track the total participants in a room :tada:
<https://github.com/livekit/livekit/releases/tag/v1.1.0|https://github.com/livekit/livekit/releases/tag/v1.1.0>

:wave: Hi everyone! I found about livekit while I was browsing for a webRTC implementations on top of pion. Really excited about the work that’s being going on. This was exactly what I wanted.

Hey <@U03M8ELQTRS>, welcome! Two things here:

1. Reading this may shed some light on our ambition/commitment to this project: <https://blog.livekit.io/the-metaverse-needs-open-infrastructure-39bc3f2805a7/|https://blog.livekit.io/the-metaverse-needs-open-infrastructure-39bc3f2805a7/>
2. We're working on a (paid) cloud-managed offering of LiveKit, which will allow us to continue working on this for many years to come.

thanks <@U01KZ0SCDKL>, will get familiar with the project moving forward. Have just found it 15 min or so ago. Sounds like a good plan!

Def let us know if you have any questions!

sure thing! Really early in product discovery mode, so it might take weeks until we dive in, but this just made it to the top of my list as a tech stack for our audio needs.

:wave: Hi everyone! Thanks for inviting me. Using LiveKit in one project. And one more planned.

live text.gif

<@U01KZ0SCDKL> another one I found for text is <https://matrix.org>.

Ideally we’d use LiveKit for all 3 - text, audio and video.

<@U03M8ELQTRS> yes, you can use LiveKit for text. See this: <https://docs.livekit.io/guides/room/data/#data-messages>

I noticed that the same tech - webRTC is used for chat apps, and these guys seem to work on that, but it’s not open source

<https://pusher.com/tutorials/webrtc-chat/>

what I’m going for is, instead of showing three dots that indicate someone is writing, to actually start displaying the text people are writing. I’m aware of the implications of that UX wise.

even if someone isn’t doing it this way, and showing … instead, it still seems like the Live Kit should contain the text part too.

note: I didn’t yet go through the docs, not sure if it’s already there.

<https://docs.livekit.io/guides/room/data/#data-messages>

yep thanks, wondering if this can be expanded into something that puts live text on equal footing with audio and video, as per my example before

Sorry if I'm missing something, but just have to bind event listener to the user input, and publish input data to room/participant

we’ll look into it, need to see what Pusher offers on that front as well. It just seemed to me as they’re adding more stuff.

yeah it definitely can be expanded from there. we've built the primitives for data first (after all, media is just a specialized form of data). next steps are higher level constructs to work with data

I hope it’s something you’re looking into.

I am the co-founder of Iamears which is an anonymous mental health community

Wanted to know more about livekit and it's experience

Can I connect with some founders or people from livekit who can helps us with this?

Hey all, just a friendly reminder to thread your messages. :pray: There’s almost 1K folks in this channel and threading makes messages easier to skim through and follow along with (for everyone). cc <@U03M8ELQTRS> <@U03M6MREVNZ>

Hey Nelson, can you share some context about what you are doing? It sounds like you are doing egress of some sort.. which kind? and how is it started?

I am using egress. I want to upload live broadcast to cloud via s3 protocol, the following is my code using golang

image.png

This is the log of the livekit service container and the log of the egress service container respectively

are you publishing to the same room? it looks like there are no publishing participants in the room.

since you have a video produced, it means that the pipeline is working well. it's just that the template is rendering a black background, since it hasn't seen any video tracks published

I use *<https://example.livekit.io/#/>* demo to create a room and live stream, is it not a publisher?

yeah that is a valid publisher. are you sure they are joining the same room though

Yes, I'm sure, and, before I created another publisher, and also joined the room was also a black screen

right, I'm checking because this would be the behavior if you are recording an empty room.

The token generated by the following command
```docker run --rm -e LIVEKIT_KEYS="***: ***" \
    livekit/livekit-server create-join-token \
    --room "test-room" \
    --identity "aaa"```


can you attach the full server logs as a file.. it's very difficult to parse details from screenshots

egress.rtf

livekit-server.rtf

```func RoomCompositeEgressRequestFile() {
	fileRequest := &amp;livekit.RoomCompositeEgressRequest{
		RoomName: roomName,
		Layout:   "grid-dark",
		Output: &amp;livekit.RoomCompositeEgressRequest_File{
			File: &amp;livekit.EncodedFileOutput{
				FileType: livekit.EncodedFileType_MP4,
				Filepath: "/livekit-demo/test2.mp4",
				Output: &amp;livekit.EncodedFileOutput_S3{
					S3: &amp;livekit.S3Upload{
						AccessKey: AccessKey,
						Secret:    Secret,
						Region:    Region,
						Bucket:    Bucket,
						Endpoint:  Endpoint,
					},
				},
			},
		},
	}
	info, err := egressClient.StartRoomCompositeEgress(ctx, fileRequest)
	fmt.Println(err)
	streamEgressID := info.EgressId
	fmt.Println(streamEgressID)
}```

This is the egress configuration file
```log_level: debug
api_key: ***
api_secret: ***
ws_url: livekit:7880
insecure: true
redis:
    address: redis:6379
    username: ""
    password: ""
    db: 0
s3:
  access_key: "***"
  secret: "***"
  region: "***"
  bucket: "***"
  endpoint: "***"```

looking at the logs, it shows you had a single user join the room, and the egress participant never joined. were you recording during this time?

your server logs do not show you've attempted to egress the room.

What is the reason for this? How to deal with it?

I heard from you just now that I may be recording, I retry the demo to live broadcast, and upload through s3 except for a black screen and no sound

I'm afraid you'd have to dig into this.. As I said, your server logs do not show any egress activity, it's just a single publisher in the room. Your egress instance is probably interacting with another instance of livekit server.

I just create a container, and the room is also created randomly, it is impossible to say that it is associated with another livekit service, and if there are two services, the interface will conflict, so this possibility is very low, right The mirror I used are the latest

Could it be a container permission issue?

Through the *srs* streaming service, normally there is streaming output, but there is no audio and video output. At present, I have no clue of investigation. Can you give me some direction and help?

this is what I said before:
```your server logs do not show any egress activity, it's just a single publisher in the room. Your egress instance is probably interacting with another instance of livekit server.```

How to troubleshoot the problem?
docker-compose.yaml
```version: '3'
services:
  livekit:
    image: livekit/livekit-server
    container_name: livekit-server
    command: --config /etc/livekit.yaml --node-ip 127.0.0.1
    restart: unless-stopped
    networks:
      - webrtc-network
    depends_on:
      - redis
    ports:
      - "7880:7880"
      - "7881:7881"
      - "7882:7882/udp"
    volumes:
      - ./livekit.yaml:/etc/livekit.yaml
    labels:
      - "traefik.enable=true"
      - "traefik.http.routers.livekit.rule=Host(`livekit.****.****`)"
      - "traefik.http.routers.livekit.entrypoints=websecure"
      - "traefik.http.routers.livekit.service=livekit"
      - "traefik.http.services.livekit.loadbalancer.server.port=7880"

  egress:
    image: livekit/egress:v1.0.4
    command: mkdir out
    container_name: livekit-egress
    environment:
      EGRESS_CONFIG_FILE: /out/config.yaml
    networks:
      - webrtc-network
    depends_on:
      - redis
      - livekit
    volumes:
      - ./livekit-egress:/out

  redis:
    image: <http://ydjharbor.jingzhuan.cn/public/redis:4.0.14-alpine|ydjharbor.jingzhuan.cn/public/redis:4.0.14-alpine>
    container_name: livekit-server-redis
    ports:
      - "6380:6379"
    networks:
      - webrtc-network
    restart: on-failure

networks:
  webrtc-network:```

I'm not sure you can deploy livekit this way..

node ip should not be 127.0.0.1.. that would not allow connections to it

Should it be the host's ip or the docker ip?

you shouldn't need to set it.. see `use_external_ip` parameter for it to autodiscover IP

I deployed 127.0.0.1 before, the demo can be connected successfully, I am also very confused

with docker.. you should be using host networking.. ports are going to be a mess without it

please refer to docs for more details: <https://docs.livekit.io/deploy/>

<@U03DUKK5K0T> could you share some more context (and code) to give us an idea of what you're trying to accomplish?

The events are successfully being received by the endpoint, but even with the express.raw() middleware, req.body is a JSON object and therefore errors when input into webhookReceiver.receive

I'm able to convert it back to a string with `JSON.stringify(req.body)` to resolve that issue

But then the Auth header has problems. The checksum fails and I also get `JWT NotBeforeError`

Screen Shot 2022-06-28 at 7.18.50 PM.png

These are my current workarounds. `tokenVerifier.verify` still fails _almost_ every time (weird that it doesn't fail _every_ time).

Chris, when you say `the new livekit server`. has the behavior been different before? I don't believe there are any recent changes to the payload type

<@U01KZ61GV5Z> by `new` I mean the most recent version of the livekit server with webhooks enabled. I didn’t have webhooks enabled in previous versions, so I don’t know if the latest version has a regression.

I see, I believe it's always posted content-type as `application/json` . I think you can still configure your webserver to parse it as plain-text. Since you are seeing an object there, that route configuration probably didn't apply correctly. It will not work to re-serialize it back, since the signature is generated on the original string, and re-serialized copy might not match.

It's a fair point that perhaps we should just send it as `application/octet-stream`, so JSON parsers don't get too aggressive there.

<@U01KZ61GV5Z> that might help. You could also just continue passing it as JSON and rewrite the webhookreceiver to accept the JSON object?

Frankly, I only see a need for verifying the auth token using a Livekit SDK call. I can deal with the req.body on my own.

Also, shouldn’t req.get(‘Authorization’) still return a valid token?

once it's parsed into JSON, the original string is lost, so it'd be impossible to compute a checksum to ensure validity.

<@U01KZ61GV5Z> ah okay. I dug deeper into the express.js docs and found another way to get the raw body even when content type is application/json. See this if you’re curious: <https://stackoverflow.com/questions/68680900/express-4-17-get-raw-body-for-one-endpoint|https://stackoverflow.com/questions/68680900/express-4-17-get-raw-body-for-one-endpoint>

For now, would help to clarify in your docs

<@U01KZ61GV5Z> Just confirmed that it works perfectly like this:

```app.use('/livekit', express.raw({type: '*/*'}), liveKitRouter);```

```const webhookEvent = webhookReceiver.receive(req.body, req.get('Authorization'));```


Screen Shot 2022-06-29 at 6.13.16 PM.png

<@U01KZ61GV5Z> yikes, still experiencing this intermittent NotBeforeError: jwt not active

Its being caused by this line _some_ of the time:
```const webhookEvent = webhookReceiver.receive(req.body, req.get('Authorization'));```


are the clocks synchronized on your server and client?

<@U01KZ61GV5Z> yes, they are both synchronized. 

they are both UTC.. so it shouldn't be.. I've never seen this before. If you get this again, can you paste the authorization token? I'll look into it tomorrow.

New guy: I’m trying to get the <https://github.com/livekit/client-example-swift> up and running. Do I need to get my own token?