RTCSession Error

[sw7837]

Description

Element Web is fine with video and voice
When using ElementX on a mobile device, an error occurs
Sometimes you can talk for a few minutes, sometimes you can't get in and report an error,
Web and mobile are in a network,5Gwifi network speed stability is very good,
synapse and elementX are the latest versions
synapse built it,elementX didn't build it
The server is 16C32GB

RTCSession Error
The MembershipManager shut down because of the end condition: Error: Could not restart delayed event, even though delayed events are supported. tw: error sending request for url (https://chat.bitdu.love/_matrix/client/unstable/org.matrix.msc4140/delayed_events/syd_YkZjZSYxmnyWEACerjAg)

2025-04-15T20:26:57.029Z WARN livekit.transport rtc/transport.go:830 error reading data channel {"room": "!pLJrBmFowOYaQYdeyC:chat.test123.com", "roomID": "RM_nsG6Rpoota4y", "participant": "@nick:chat.test123.com:YKIXTBGXNH", "pID": "PA_bhtJgZptc4L4", "remote": false, "transport": "PUBLISHER", "label": "_lossy", "error": "[0xc00325a780] abort chunk, with following errors: (User Initiated Abort: Close called)"}
2025-04-15T20:26:57.029Z WARN livekit.transport rtc/transport.go:830 error reading data channel {"room": "!pLJrBmFowOYaQYdeyC:chat.test123.com", "roomID": "RM_nsG6Rpoota4y", "participant": "@nick:chat.test123.com:YKIXTBGXNH", "pID": "PA_bhtJgZptc4L4", "remote": false, "transport": "PUBLISHER", "label": "_reliable", "error": "[0xc00325a780] abort chunk, with following errors: (User Initiated Abort: Close called)"}
2025-04-15T20:26:57.032Z INFO livekit rtc/participant.go:1036 participant closing {"room": "!pLJrBmFowOYaQYdeyC:chat.test123.com", "roomID": "RM_nsG6Rpoota4y", "participant": "@nick:chat.test123.com:YKIXTBGXNH", "pID": "PA_bhtJgZptc4L4", "remote": false, "sendLeave": true, "reason": "CLIENT_REQUEST_LEAVE", "isExpectedToResume": false}

Steps to reproduce

When using ElementX on a mobile device, an error occurs
Sometimes you can talk for a few minutes, sometimes you can't get in and report an error,
Web and mobile are in a network,5Gwifi network speed stability is very good,
synapse and elementX are the latest versions
synapse built it,elementX didn't build it

Homeserver

chat.***

Synapse Version

latest

Installation Method

Docker (matrixdotorg/synapse)

Database

PostgreSQL

Workers

Single process

Platform

ubuntu
16c32g

Configuration

No response

Relevant log output

2025-04-15T20:26:57.029Z        WARN    livekit.transport       rtc/transport.go:830    error reading data channel      {"room": "!pLJrBmFowOYaQYdeyC:chat.test123.com", "roomID": "RM_nsG6Rpoota4y", "participant": "@nick:chat.test123.com:YKIXTBGXNH", "pID": "PA_bhtJgZptc4L4", "remote": false, "transport": "PUBLISHER", "label": "_lossy", "error": "[0xc00325a780] abort chunk, with following errors: (User Initiated Abort: Close called)"}
2025-04-15T20:26:57.029Z        WARN    livekit.transport       rtc/transport.go:830    error reading data channel      {"room": "!pLJrBmFowOYaQYdeyC:chat.test123.com", "roomID": "RM_nsG6Rpoota4y", "participant": "@nick:chat.test123.com:YKIXTBGXNH", "pID": "PA_bhtJgZptc4L4", "remote": false, "transport": "PUBLISHER", "label": "_reliable", "error": "[0xc00325a780] abort chunk, with following errors: (User Initiated Abort: Close called)"}
2025-04-15T20:26:57.032Z        INFO    livekit rtc/participant.go:1036 participant closing     {"room": "!pLJrBmFowOYaQYdeyC:chat.test123.com", "roomID": "RM_nsG6Rpoota4y", "participant": "@nick:chat.test123.com:YKIXTBGXNH", "pID": "PA_bhtJgZptc4L4", "remote": false, "sendLeave": true, "reason": "CLIENT_REQUEST_LEAVE", "isExpectedToResume": false}

Anything else that would be useful to know?

No response

[sw7837]

Using Element is fine
Using ElementX results in an error

[velx4]

Sounds more like an issue with Element X rather than Synapse.

[ciyam]

I have been experiencing the exact same problem (and submitted a couple of rage shake reports) - it happens using either the Apple or Android versions of Element X and am likewise having no issues at all with audio/video Element Calls between desktop computers.

[escix]

Just to confirm, I have been using elementX with selfhosted { Livekit and JWT } MatrixRTC without any problems in android.

Calls span from minutes to hours without any problems. Not element-call but widget mode. Previously element-call has also worked without any issues.

[anoadragon453]

@sw7837 I've moved this issue from the Synapse issue tracker to Element Call's. Please can you also "rageshake" from within the Element Call settings. That would help the team get the full logs (and the logs of any other clients present in the room at the time).

[fkwp]

are you using synapse workers? and is there a chance the workers were restarted or so?

[ciyam]

I am wondering if the livekit-jwt service has something to do with this as I see the following (redacted) lines in its log:

May 10 17:09:54 <server> lk-jwt-service[32152]: 2025/05/10 17:09:54 Request from [::1]:57378 at "https://appassets.androidplatform.net"
May 10 17:09:54 <server> lk-jwt-service[32152]: 2025/05/10 17:09:54 Request from [::1]:57378 at "https://appassets.androidplatform.net"
May 10 17:09:54 <server> lk-jwt-service[32152]: 2025/05/10 17:09:54 Got user info for @<user1>:<server>
May 10 17:10:04 <server> lk-jwt-service[32152]: 2025/05/10 17:10:04 Request from [::1]:57378 at "https://call.element.io"
May 10 17:10:04 <server> lk-jwt-service[32152]: 2025/05/10 17:10:04 Request from [::1]:57378 at "https://call.element.io"
May 10 17:10:04 <server> lk-jwt-service[32152]: 2025/05/10 17:10:04 Got user info for @<user2>:<server>

As Livekit is self-hosted I don't understand why "appassets.androidplatform.net" and "call.element.io" are actually doing there in the log (the latter being from my laptop and the former being from my Android phone).

The following is the redacted content of https://host/.well-known/matrix/client:

{
    "m.homeserver":
    {
        "base_url": "https://<server>"
    },

    "org.matrix.msc4143.rtc_foci":
    [
        {
            "type": "livekit",
            "livekit_service_url": "https://<server>"
        }
    ]
}

I must admit I found the documentation concerning the "rtc_foci" to be contradictory (some examples include having host/livekit/jwt and others with host/livekit/sfu and another stating you only should put host which is what I had thought was working.

[fkwp]

If is referring to the same in the following both lines that is indeed wrong.

"base_url": "https://<server>"
"livekit_service_url": "https://<server>"

Request from [::1]:57378 at "https://appassets.androidplatform.net" or
Request from [::1]:57378 at "https://call.element.io" are referring to the Element Call client which was requesting the JWT token:

• https://appassets.androidplatform.net hints at the in-app bundle
• https://call.element.io hints at Element Call configures via a Widget URL

for details refer to https://github.com/element-hq/element-call?tab=readme-ov-file#-element-call-packaging and https://github.com/element-hq/element-call/blob/livekit/docs/embedded-standalone.md

I must admit I found the documentation concerning the "rtc_foci" to be contradictory (some examples include having host/livekit/jwt and others with host/livekit/sfu and another stating you only should put host which is what I had thought was working.

While there are many ways to achieve a single goal we do recommend the pattern as outlined here: https://github.com/element-hq/element-call/blob/livekit/docs/self-hosting.md

However, the issue as outlined by @sw7837 seems to be more related with an unreliable connection to the homeserver such that the delayed_event is kicking in and the call is ended. Hence, the Q if synapse workers are part of the setup.

[ciyam]

If is referring to the same in the following both lines that is indeed wrong.

I'm not sure that wrong is accurate as it does work that way (i.e. the livekit-jwt log would show nothing at all and you'd get a FOCI error if it couldn't connect).

But in any case I have now changed it to include the /livekit/jwt endpoint and it still behaves exactly as before.

So I am able to start an Element-Call between my laptop and Android phone and it appears to be working perfectly for a certain amount of time (sometimes as much as ten minutes if lucky) but then I end up getting the RTC Session Error (which occurs in the Android Element-X app).

[escix]

If is referring to the same in the following both lines that is indeed wrong.

I'm not sure that wrong is accurate as it does work that way (i.e. the livekit-jwt log would show nothing at all and you'd get a FOCI error if it couldn't connect).

But in any case I have now changed it to include the /livekit/jwt endpoint and it still behaves exactly as before.

So I am able to start an Element-Call between my laptop and Android phone and it appears to be working perfectly for a certain amount of time (sometimes as much as ten minutes if lucky) but then I end up getting the RTC Session Error (which occurs in the Android Element-X app).

it should be lk-jwt-service url.

livekit log should be showing some info (lk-jwt-service log will be empty).

Also turnserver log will also be helpful.