LiveKit Community

Yup, check out our docs for usage examples across platforms: <http://docs.livekit.io|docs.livekit.io> 

what is `livekitHost` set to? assuming your node app is running on the same machine, I can't see a reason for this to fail

yes, i use
```livekit-server --dev```
in same machine , server-sdk-go : perfectly work
But server-sdk-js : some function error

I'm curious about the NodeJs version of the people who use the server-sdk-js
And I wonder if they are using typescript

Hello, Can you tell me which version of NodeJS, server-sdk-js you are using?

I want to use server-sdk-js but mycase some function not work and cause error TT

Yes i created my server with typescript. It was an easy task tbh :D

why there is not tutorial guide from community for js server with unity ?

looks like this is because a publisher connection couldn't be established: <https://github.com/livekit/client-sdk-swift/blob/5cc3c001779ab147199ce3ea0dce465b846368b4/Sources/LiveKit/Core/Engine.swift#L238>

Still not clear what is happening. We restart the server and everything works fine again

<@U02FLGTTH8Q> if you have server logs before &amp; after this starts happening, could you please share them with me?

you can check them here is the same use case:
<https://livekit-users.slack.com/archives/C01KVTJH6BX/p1665059609461849?thread_ts=1664877114.008289&amp;cid=C01KVTJH6BX>

Hey Valentin, could you file an issue in GH with some details?  including versions, API called, and any other repro steps

your logs indicate something has disconnected your websocket connection, so please indicate if you are seeing anything in LiveKit server logs too

I have to add that the result of the `stopEgress` call doesn't indicate any error, but the `file.location` field is empty. This is probably not expected ?

RoomService is a part of livekit-server that you would run. There is nothing else to deploy :slightly_smiling_face:  Some docs <https://docs.livekit.io/guides/server-api/|here>.

room joining.png

Okay than this code should be working fine ?

livekit1x-samesource.png

livekit1.x.png

livekit0.x.png

Can you describe what your application is trying to do? The warning indicates you've published multiple tracks with the same `Source` field. are you sharing multiple screen shares?

Our application is very alike to a common video conferencing tool. It offers multiple users to share screens. We get an error when user stops then starts the screen share the second time. I’m currently using livekit 1.2x,

When I trace the issue I see that, the screen share publication drops from local participant. But it remains on remote participants. And somehow, there is no track underneath this publication.
When user shares screen second time, livekit throws a warning “received a second track publication for xxxx with the same source: screen_share” and generates a second publication with proper track in it. When I call participant.getTrack(Track.Source.ScreenShare) function, livekit gives me the first screen share publication that doesn’t have the track in it. And this causes issue on our application.

I compared the process with 0.15x. On 0.15x, when a remote participant stops the screen share, screen share publication is being dropped from both local and remote participants. So not matter how many times a user shares the screen, it causes no issues.

Hello <@U01KZ61GV5Z>,

We tested with several server versions, so 1.0.x works perfectly but after upgrade to 1.1x or higher we're facing with this issue.

are you unpublishing the previous track prior to publishing again? the error would show if unpublish was not performed.

we're found the problem when we're set the `TrackPublishDefaults.videoCodec` to h264 this is happen when we're back to vp8 everythings fine.

ah, yes! I believe this issue was just fixed. we'll have a release out with it later today

LiveKit does not support p2p, for reasons see discussion <https://github.com/livekit/client-sdk-js/issues/270|here>.

<https://docs.livekit.io/guides/room/connect/#connectivity|our docs> has a discussion around why TURN is used

Hi!
I don’t think there’s currently a built-in way of checking for expiration time in the server-sdk.
One way to do it would be
```const tokenPayload = token.split(".")[1];
      if (tokenPayload) {
        const tokenExpirationTime = JSON.parse(atob(tokenPayload)).exp;
...
}```


<@U02EU2ZPTQU> Thank you for your answer,
the problem was on the front side time, I just set it and it worked :slightly_smiling_face:

Welcome to Node.js v14.20.0.
or 18, there is no change

when I remove livekit-server-sdk from my project the problem disappears.

"dependencies": {
    "core-js": "^3.8.3",
    "crypto-browserify": "^3.12.0",
    "livekit-client": "^1.4.1",
    "livekit-server-sdk": "^1.0.2",
    "stream-browserify": "^3.0.0",
    "vue": "^2.6.14"
  },

const { defineConfig } = require('@vue/cli-service')
module.exports = defineConfig({
  transpileDependencies: true,
  configureWebpack: {
    resolve: {
      fallback: {
        "stream": require.resolve("stream-browserify"),
        "crypto": require.resolve("crypto-browserify")
      },
    },
  }
})

hi!
the `livekit-server-sdk` is not supported to run in browsers, but only in nodejs environments. you could for example use a server side route with nuxt or a nodejs cloud function in one of the popular serverless environments.
But of course you could also just let a node-script run on a VM of your choice.

Hmm.

I thought of it as a client that uses api's. this situation did not come to my mind. thank you.

happy to help! just for context, there’s also a security concern: you should never provide your livekit api key and api secret to a client facing environment as those credentials might get stolen by a malicious actor.
that’s why all APIs requiring those credentials are supposed to only run in a secure server environment where users don’t have direct access to

We do have a Go based SDK that could serve as a foundation for that. However, device management and media encoding/decoding will need to be implemented if you are looking to build client-side applications with it.

The SDK is compatible with <https://github.com/pion/mediadevices/|Pion mediadevices>, which implements some of these capabilities

Could you please help to point the repo about the "Go based SDK that could serve as a foundation"?

<https://github.com/livekit/server-sdk-go>

In the livekit-client SDK (<https://github.com/livekit/client-sdk-js>), it uses webrtc adapter instead of using go version based webrtc, why don't we use go version based webrtc here <https://github.com/pion/webrtc>? I am thinking with <https://github.com/pion/webrtc>, we have more control in the webrtc layer. Thanks.

Hello all, did any here tried implementing LiveKit with any of your unity3d project ?

<@U044SAWPQ21> i'm also looking for it. Let's what comes out

<@U043YJFANAK> have you implemented it with Unity3d ?

hey guys, please thread the convo and/or move it to DM :slightly_smiling_face:

this is a public channel with over a thousand people in it. thank you for understanding :pray:

Token generation is working but not able to join room.

Hi Nuno, could you share raw logs with me?

would be great to see logs before it stops working, and all the way to after.

6500318.log

Hi, pls take a look at the attach. Tks, Nuno

the log message you included is only informational and would not by itself cause any issues.

I think you might be seeing the same issue as reported <https://github.com/livekit/livekit/issues/1060|here>. which has to do with a compatibility issue with using a single port for UDP and AWS instances.

Are you running with single port mode?

I'll check on this. I haven't specified a udp_port so I assume I'm not on a single port setup.

I have tested on <https://livekit.io/connection-test> and I'm getting intermittent passes and fails

think i've figured it out, i was missing UDP ingress access on ports 50000-60000

Screen Shot 2022-10-04 at 5.07.27 PM.png

Hey David, can you confirm if you are connected via TURN/TLS or regular TURN?  To check, you can visit <chrome://webrtc-internals> and see the specific TURN url it ended up using (attached)

Even if it's using TURN/TLS properly, there could still be strict corporate firewalls that ends up blocking the domain/IP. Certain customers may need to whitelist your domain in their firewall rules

image.png

I see a few instances of RTCIceCandidate_… (local-candidate). They differ in networkType and priority but otherwise look the same. url is the same on all of them (see screenshot).

Does this mean it’s not using TURN/TLS?

I’m reading the docs for firewall settings <https://docs.livekit.io/deploy/ports-firewall/>. Could the problem be that I have defined turn.tls_port in my config yaml  while also using a load balancer?

correct, try removing the `udp_port: 443` from your config.. having it may be causing the browser to prefer TURN/UDP instead of TURN/TLS.

removing `udp_port: 443` from the config seemed to do the trick. I’m now seeing my browser use TURN/TLS. Thanks for the help!

yup, you can <https://docs.livekit.io/guides/server-api/#listrooms|ListRooms> with a server API

there's an issue filed for this: <https://github.com/livekit/livekit-helm/issues/7>

would welcome PRs to make this happen

I can try making a pr on the helm chart later today

from what I see, the entirety of Values.livekit is being templated into the configmap?

yeah. LiveKit can read keys from `LIVEKIT_KEYS` env var, so perhaps it's just mapping the secrets to that.

then I may be able to just add a way to add secrets to env vars

from what I see, it doesn't template in `LIVEKIT_KEYS`  to envs

yes, the current helm chart doesn't support secrets. but that'd be the piece to add: pulling in a secret into `LIVEKIT_KEYS` env var

As a possible workaround until we look into this, you can try EGRESS_CONFIG_BODY

```EGRESS_CONFIG_BODY=`cat ~/livekit-egress/config.yaml````
is worth a try

I am using the command:
"docker run --rm -e EGRESS_CONFIG_BODY="$(cat config.yaml)" -v /root/record:/out livekit/egress"

However I get the following error:
+ export TMPDIR=/tmp/lkegress
+ TMPDIR=/tmp/lkegress
+ pulseaudio -D --verbose --exit-idle-time=-1 --system --disallow-exit
W: [pulsaudio] main.c: Running in system mode, but --disallow-module-loading not set.
N: [pulsaudio] main.c: Running in system mode, forcibly disabling SHM mode.
I: [pulsaudio] main.c: Daemon startup successful.
+ '[' -z /tmp/lkegress ']'
+ mkdir -p /tmp/lkegress
+ exec egress
{"level":"info","ts":1665045471.3235445,"logger":"egress","caller":"redis/redis.go:53","msg":"connecting to redis","nodeID" :"NE_bjdCZqenEuLW","sentinel":false,"addr":"localhost:6379"}
unable to connect to redis: dial tcp 127.0.0.1:6379: connect: connection refused


It does not connect to redis-server, I have tested redis with:
redis-cli -h localhost -p 6379 ping

The result is:
 PONG

So redis is working perfectly.

figured this out, the key specified needs to be in the keys config.

I think this could be better documented?

i’ll take a look at where we can clarify this in the documentation Ryan, thanks for posting about it.

I came across this too.  It was only until I found this thread it got me to the next step.

This must have slipped through the cracks, going to take a look a updating the docs today.

Gentle ping here - joined this Slack today to find the answer to this issue.

Updated docs would be appreciated, happy to submit a PR