This message was deleted.
# cloud
s
This message was deleted.
m
hm…do you have a link to the session and participants in the Cloud dash @important-boots-3535?
feel free to DM me if you don’t want to put it here (fwiw access control will prevent any non-LK core folks from accessing it)
i
Sure, DM'ing now
d
Hey Alex, looking at the code snippet you sent over, I think the issue is the track had been muted too quickly immediately after publishing. There's a limitation to our SFU stack where you cannot immediately mute a published track (it's better to not
setMicrophoneEnabled(true)
initially if the user didn't desire to publish audio
ok, looking at the logs, our engineers have identified an issue. we'll take a look from our side.
f
@important-boots-3535 In the session you shared, would you happen to remember which way the audio unmute did not work? I see two users 9924 and 9925 and both have muted audio for over 10 minutes in that session. Wanted to understand if the failure was one way or both ways suffered the same issue. If only one side, do you remember who could not hear whom?
i
Failure was both ways in that session. However, we have had reports where the failure affects only some participants (some can hear the unmuted participant and some can't).
f
Thank you Alex. Will investigate.
i
Thanks!
An update - We were setting dtx to false. Since changing it to true, we have been unable to trigger the bug after a few tests in staging. This seems an odd fix, but we will run dtx=true in prod. and see if we still get reports of the issue.
f
Thank you Alex. Yeah, that is just a happenstance. We will dig into this.
🙏 1
Alex, sorry about more questions. I am not able to reproduce the problem. But, I am trying to look at the logs and check for patterns that I can try reproducing. Would you happen to remember when the not able to hear happened? For track published by participant 9924 (and subscribed by 8095), I see mutes at (all times IST - easier for me as that is my timezone 🙂 , but the diffs are not more important) 000328 - subscribe 000336 - muted 001652 - unmuted 003258 - muted 003305 - unmuted 003310 - muted 003311 - unmuted 003407 - muted 003409 - unmuted 003412 - muted 00:42.52 - unsubscribe For the track published by 8095 (and subscribed by 9924) 000329 - subscribe 000332 - muted 003321 - unmuted 003321 - muted 003404 - unmuted 003411 - muted 004252 - unsubscribe
i
Hey Raja, sure - For 9924, the "not able to hear "happened at
00:16:52 - unmuted
For 8095, the "not able to hear" happened at
00:33:21 - unmuted
I'm not sure exactly what the threshold is, but it's at least 10 minutes and possibly closer to 15 minutes. So when we tested, if we unmuted after > 15 minutes muted, we could reproduce the bug every time.
If it helps, I can probably find and send you links to more sessions that his was happening in.