Multiple sequential reconnections impeding users to stay in call

[EvertonMJunior]

Hello,

We are having a recurring issue at least once daily where the users are reporting they are being sequentially reconnected, without being able to stay in the call and talk to the other user in the call.
All our calls happen with two users, one student and one teacher.

This is an example of a session where this happened: https://beta.dashboard.getstream.io/organization/1252570/1354017/video/call-viewer/default:ab28c39c-c5e1-4527-8b3f-6dfa7d42517f/b9d7f8c4-98d2-421e-ad9f-b55d75a0e84f/
Here's a screenshot of the issue:

Here are the event logs of one of the users sessions:

15:11:06
info
sfu_error

{
  "payload": {
    "error": "subscriber PC: negotiation failed"
  },
  "timestamp_ms": 1773684666160,
  "peer": "0-sfu-oci-brazil-vp2-02d57c0f3aae.stream-io-video.com"
}

15:11:06
info
callReconnect

{
  "payload": {
    "reason": "sfu error: subscriber PC: negotiation failed",
    "strategy": "rejoin"
  },
  "timestamp_ms": 1773684666162,
  "peer": "0-sfu-oci-brazil-vp2-02d57c0f3aae.stream-io-video.com"
}

15:11:06
info
Error

{
  "payload": {
    "error": {
      "reconnectStrategy": 3,
      "message": "subscriber PC: negotiation failed",
      "code": "participantMediaTransportFailure",
      "shouldRetry": false
    }
  },
  "timestamp_ms": 1773684666163,
  "peer": "0-sfu-oci-brazil-vp2-02d57c0f3aae.stream-io-video.com"
}

Here's also a video of the issue happening, this was in another session and was recorded by the teacher, where the student was suffering from the issue: https://imgur.com/pCRNKsz

(I blurred faces and names for privacy reasons)

Until last week, we were using v1.2.4 of the SDK, but since yesterday, all users are using it through the latest release, v1.3.0, we disabled access to the calls for app versions with the outdated SDK.

We hoped the update to v1.3.0 would fix this issue, but it still persists and was just reported. I also reported it through a ticket on your support, #78739 and we had reported it previously on Feb 3rd on ticket #77688, where we were told that "it looks like the client tried to connect to the call through a server which was full, and kept retrying that connection. It's an issue we're currently working on addressing and making sure the SDKs don't retry the same problematic server on loop.". It seems that either this issue was not yet fixed, or that this was not the actual only/root cause of the problem.

I'm reporting it here as well to try to increase visibility on this issue, as it's happening recurringly for more than one month and has not yet been fixed/identified.

Happy to add more details if needed.
Thanks again!

[Brazol]

Hi @EvertonMJunior, thanks for your patience. We’ve included the fix for the SDK joining full SFUs in the upcoming release, which we’ll publish as soon as possible. Please let us know if the issue persists after upgrading to the new version by reopening this issue or creating a new one.

[EvertonMJunior]

Hey @Brazol, thank you very much for the agility in attacking this issue. We're hopeful this will solve this problem, and will keep you posted. As soon as we get the new version, we'll update our app with it. Also, I saw the fix for #1189, and I think it could solve an issue we had reports too, but hadn't pinpointed to this iOS-specific problem.

Thanks a lot again!