LiveKit Community

ah I think I know what it is.. it's a timing issue.. good catch

I'll file an issue to fix.. it's on the client side

added logs to this issue: <https://github.com/livekit/livekit-server/issues/21>

what it looks like (to me) is that session 1 connects -&gt; session 2 connects -&gt; session 1 disconnects. We saw this with websockets and chrome before too, b/c the connectionId == userId it'll kill the old one. i think connectionId probably should just be a uuid per connection

are you trying to have the user in the call twice?

the behavior in the latest build is that the latest session to connect will retain the connection

b/c of chrome + keepalive it's totally reasonable to get the 2nd connect before the first fully disconnects

with 0.4.3, the behavior is that it'll allow the 2nd connection to work.. while dropping the first one

btw.. ur logs are using single node routing..

which doesn't work with this.. the issue should be resolve in multi-node

yah, didn't spend much time to fix the behavior for single node.. as it's supposed to be for dev validation

LiveKit provides audio level and a list of active speakers, but it updates every 500ms. for real time, you'd need to hook into a mediastream: <https://gist.github.com/yying/754313510c62ca07230c|https://gist.github.com/yying/754313510c62ca07230c>

how can i access the audio level that livekit provides?

want to try it out and see how well it works for our use case

give it a shot.. you can hook into it like this: <https://github.com/livekit/client-sdk-js/blob/main/test/sample.ts#L191>

`Participant.audioLevel` gives you the audio level as a float between 0 and 1

hmm interesting.. which version are you trying with? see if you can get webrtc-internals stats graph for the track that gets stuck

Hmm, I haven't seen this in a while. Seeing it on `0.5.0`

```Uncaught (in promise) DOMException: Failed to execute 'setRemoteDescription' on 'RTCPeerConnection': Failed to set remote answer sdp: Called in wrong state: stable```

wait wait disregard, I think I know whats happening

I have 2 nodes running right now that are sharing a redis instance

the app retired the connection multiple times and eventually succeeded

are the two nodes.. on the same physical machine ?

so currently it uses MAC address of the machine as node id.. so if two instances of the server register with the same node id, it'll wreck havoc in the routing logic

server 1:

```eth0      Link encap:Ethernet  HWaddr DE:65:73:BF:2E:20  
          inet addr:100.96.56.32  Bcast:100.96.56.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:9001  Metric:1
          RX packets:5477 errors:0 dropped:0 overruns:0 frame:0
          TX packets:10598 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:880746 (860.1 KiB)  TX bytes:1184265 (1.1 MiB)```
server 2:

```eth0      Link encap:Ethernet  HWaddr 0A:4D:8F:26:C5:08  
          inet addr:100.97.88.157  Bcast:100.97.88.255  Mask:255.255.255.0
          UP BROADCAST RUNNING MULTICAST  MTU:9001  Metric:1
          RX packets:3864 errors:0 dropped:0 overruns:0 frame:0
          TX packets:4961 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0 
          RX bytes:629425 (614.6 KiB)  TX bytes:661189 (645.6 KiB)```

can you print the node id when it starts up?

oddly enough I see this:

```{"level":"info","ts":1613594655.2477353,"caller":"service/server.go:110","msg":"starting LiveKit server","address":":7880","nodeId":"ND_0000000000000000","version":"0.5.0"}```

uh oh.. so there's a bug. in getting that Mac address

right, I mean more like OS/version/etc..  maybe it's an issue with docker runs

actually thats defined in your dockerfile