Server Connection Timed Out

Summary

I created a mechanic for my minigame and after a random time the server froze and every client got kicked with the reason network connection lost.
After days to research the issue the problem disappeared when I stopped using the material parameters in verse. I used it on a spawned prop and was updating a material parameter every 0.1 seconds, I also used moveto on the prop.

The weird part is that the server crash could even happen a couple minutes later after the loop was broken and I was not updating the parameters anymore.
It happened during live version, playtest and live session. Later I also tried to replicate the crash by just spamming a lot of parameter updates on a spawned prop but never succeeded so not sure what the exact cause is. I just know after removing the material parameter updates in verse it does not happen anymore.

As this is my Fortnite Mares submission I have the version without the material parameters usage published so I also attached the private island code with the issue.

Please select what you are reporting on:

Verse

What Type of Bug are you experiencing?

Verse

Steps to Reproduce

Use materials parameters in verse.

Expected Result

No sever crash

Observed Result

The server crashes and everyone gets kicked for network connection lost

Platform(s)

All platforms

Island Code

5268-8951-6128 (private version: 5662-6887-0219)

Additional Notes

LogNet: Warning: UNetConnection::Tick: Connection TIMED OUT. Closing connection… Elapsed: 30.00, Real: 30.00, Good: 30.00, DriverTime: 1056.85, Threshold: 30.00, [UNetConnection] RemoteAddr: 18.170.205.200:15005, Name: IpConnection_4, Driver: Name:IpNetDriver_4 Def:BeaconNetDriver IpNetDriver_4, IsServer: NO, PC: NULL, Owner: ValkyrieBeaconClient_4, UniqueId: MCP:fc48bafda23d4db098b955ae2824684c
LogNet: Error: UEngine::BroadcastNetworkFailure: FailureType = ConnectionTimeout, ErrorString = UNetConnection::Tick: Connection TIMED OUT. Closing connection… Elapsed: 30.00, Real: 30.00, Good: 30.00, DriverTime: 1056.85, Threshold: 30.00, [UNetConnection] RemoteAddr: 18.170.205.200:15005, Name: IpConnection_4, Driver: Name:IpNetDriver_4 Def:BeaconNetDriver IpNetDriver_4, IsServer: NO, PC: NULL, Owner: ValkyrieBeaconClient_4, UniqueId: MCP:fc48bafda23d4db098b955ae2824684c, Driver = Name:IpNetDriver_4 Def:BeaconNetDriver IpNetDriver_4
LogValkyrieRequestManagerEditor: Warning: Encountered Network Error!
LogValkyrieRequestManagerEditor: Error: Failed to handle connection attempt (Network error occurred when connecting to a server after matchmaking completed. Error: UNetConnection::Tick: Connection TIMED OUT. Closing connection… Elapsed: 30.00, Real: 30.00, Good: 30.00, DriverTime: 1056.85, Threshold: 30.00, [UNetConnection] RemoteAddr: 18.170.205.200:15005, Name: IpConnection_4, Driver: Name:IpNetDriver_4 Def:BeaconNetDriver IpNetDriver_4, IsServer: NO, PC: NULL, Owner: ValkyrieBeaconClient_4, UniqueId: MCP:fc48bafda23d4db098b955ae2824684c)
LogValkyrieSummary: Destroying Valkyrie Beacon

I can confirm this issue. Been debugging a server freeze issue during the day, and I’m seeing the same problem here where the game server looks to freeze randomly when using material parameters.

In the short-term I imagine I can work around this using material instances instead of parameters.

To assist any debugging Epic side, here’s a snapshot reference ID where this issue occurs:

[2024.10.14-16.53.18:178][750]LogValkyrieSummary: Sending Content Update
[2024.10.14-16.53.18:178][750]LogValkyrieSummary: Calling UpdateContent on Valkyrie beacon with: ProjectId(c4f890e0-4dcf-5844-1f51-26acf36004ee) SnapshotId(2df9656b-f51f-4d11-aa6d-540eb884bc2f) ModuleId(f989622c-4d30-6831-9f49-078af5e03d60) ModuleVersion(479)
[2024.10.14-16.53.18:179][750]LogValkyrieSummary: RPCing content cook request to server: ProjectId(c4f890e0-4dcf-5844-1f51-26acf36004ee) SnapshotId(2df9656b-f51f-4d11-aa6d-540eb884bc2f)

1 Like

We are suspecting this behavior too. These freeze issues are nightmarish to debug.

EDIT: Circumstantial evidence confirms this. Not a single freeze since we commented out verse material parameter stuff

The status of FORT-809334 incident has been moved from ‘Needs Triage’ to ‘To Do’.

1 Like

We’ll have someone take a look.

1 Like

any update on what exactly was happening here, so we can avoid the issue until it is fixed?