Multiple users have reported that this issue does not happen while using a mobile hotspot. Multiple users have also reported the issue occurs with ethernet or high network speeds.
In parallel to me rolling out some changes this week, I am curious about an experiment:
There is a tool called clumsy, which is available on Windows that lets you emulate characteristics of different network environments. Would anyone be willing to try using it to artificially/temporarily limit your bandwidth?
I would be curious if you still encounter the issue if you limit your bandwidth, to, say, 1000 or 2000 KBps. Please also set the filter to just “udp”, and be sure to click “start”. This should not impact your gameplay unless you go extremely low-- but of course you can always click “stop” if you experience any issues.
This might seem crazy but I’ve had about 3 of those “silent disconnect” in like 20 mins
and the last one was just now with clumsy on (with the settings you provided)
2025-02-11T18:04:20.433Z,116.433586,4648,6,Verbose [DFLog::RakNetMissingPingPong] Time since last (ms): ping recv 25001, ping sent 1475, pong recv 25001, pong sent 25001, pkt recv 20, pkt sent 10, reliable recv 24249, ack recv 24265, oldest unacked send 24784, oldest unacked recv 116434 | MTU size: 1396 bytes
2025-02-11T18:04:20.433Z,116.433586,4648,6,Verbose [DFLog::RakNetMissingPingPong] Other stats: dead false, outgoingWaiting true, acksWaiting false, netCapacity 65536, resendBuf msg113, resendBuf bytes22332, loss last sec0.82173324, limitedByBw false, outputQueueSize 0
2025-02-11T18:04:24.385Z,120.385551,3ba8,6 [FLog::Output] Settings Date header was Tue, 11 Feb 2025 18:04:03 GMT
2025-02-11T18:04:24.385Z,120.385551,3ba8,6 [FLog::Output] Settings Date timestamp is 1739297043
2025-02-11T18:04:26.065Z,122.065620,45c0,6,Verbose [DFLog::RakNetStoppedProduction] RakNet has not produced packets for 3601 frames / 30023 ms
Ah - thank you for offering. Actually I have made changes so that I no longer require you guys to share your client logs with me (they are automatically uploaded now)
Thank you to everyone who has been sending me logs, they were crucial in the investigation!
I’ve also had this issue for quite some time and decided to plug out my ethernet to try some with my normal wifi. I was inside of a roblox game and started up one of those internet speed tests. Specifically speedtest.net when i tabbed into roblox and noticed my ping froze. As soon as i cancelled the wifi test my roblox game went back to normal. This might just be an unrelated issue with the speed test taking up all my wifi or something or this issue is caused by other processes taking too much wifi
I really can’t bring myself to playing any Roblox game, it’s really driving my nerves. Now it’s not just silent disconnect but also includes “Player has been removed from the DataModel” as a follow up.
Testing this theory, capping my download speed has little to no effect, but capping my upload speed results in the game taking significantly longer to disconnect me.
Upon further investigation, it seems like games with less than 30 kb/s upload rarely disconnect me, but after that they take anywhere from 10~ minutes (50 kb/s) to less than a minute (120 kb/s). My Summer Car seems to be the fastest that I’ve found.
Also, I would like to volunteer for the beta test if possible.
I dont know why but i recently find large periods of lag spiking that can last minutes before i disconnect and it has just been getting on my nerves, for the past few days ive switched over to mobile data which is more expensive but i do it anyway because my average game lasts about 10 minutes.
Wanted to post an update, even w reduced mss I still see the silent drops, its intermittent and hard to consistently reproduce so idk whats exactly causing it
I tested by joining the same server in the game I linked earlier, it provides server ID and there didnt seem like any repeatable pattern between attempts
We have been observing players with connection issues in our game and even tried uploading the default Roblox baseplate to rule out any issues on our end.
We made a thread with a video of what we experienced. It seems these issues could possibly related to what users have been experiencing here.
