I can’t explain it but I’ve noticed it seems to happen with the remote laptop is connecting to a Teams call. The laptop will become unresponsive and then within seconds it will disconnect and not reconnect. I am also connecting to the remote laptop via the same LAN network via the Internet. How can I determine what the issue might be. VPN is turned off.
If the issue comes up again, could you first check your router to see if the device has lost its network connection?
If it shows as disconnected, try unplugging and replugging the Ethernet cable to see if that restores it.
It’s not a network issue as all other laptops are working fine.
There is no ethernet, it is running on WIFI.
It was a very strange incident that happened 3 times for as long as 30 minutes or more.
Just to update the situation, I've just upgraded to 1.8, and now it's restarting every two minutes. I could not maintain a long session for more than a few minutes at a time.
I have the same problem. Since early morning, I have been trying to connect to KVM, and it works locally even when the indicator shows that there is no Internet connection, but I cannot connect for long from another device.
Could you export the logs to us? (flora.wan@gl-inet.com)
For now, it hasn't disconnected for some time, so I'm monitoring the situation. If I notice a pattern, I will send you the logs.
Hi, same thing is happening to me (Teams disconnects). KVM Pro, wired. Have the diag log if helpful.
Yes, please send over the log—that would be really helpful!
system_logs_20260324_160928.zip (118.4 KB)
Here you go!
Everything looks normal on our end — would you mind sending over a short video showing the issue? Thanks so much for your help!
Thanks Flora72. Well, I tried to get it on video, and of course it wouldn’t cooperate lol. Let me take that as a homework assignment to try to capture if/when it happens next time.
No rush at all.
Since the logs suggest this issue could be caused by a few different things, would you happen to remember roughly what time the problem occurred before you exported the logs? That would really help us narrow it down.
Hi, ok so it happened again where it disconnected on a Teams call, and this time I can provide more, including a video and time stamps. Do you have a location I can send an unlisted Youtube link? I don’t want to post on a public forum due to confidentiality concerns.
Here’s the breakdown:
Date: April 1, 2026 (no not a joke lol)
Video clip start: ~14:25:49 US Eastern
KVM window suddenly cuts video and audio at ~14:25:54
Video/Audio resumes around ~14:26:05
KVM logs attached, though I forgot to save them until 15:09 so I hope it includes the time period mentioned.
system_logs_20260401_150932.zip (110.7 KB)
It may be worth mentioning that I’m using a different laptop than before and the behavior was the same on both laptops. The network connection is hard wired and both the KVM and laptop are on the same Unifi switch with new network cables at 1Gbps. Continuous pings to the Internet (1.1.1.1) do not drop.
OK, I now fully understand what your issue is. The problem lies in your broadband having multiple exits and multiple public IP addresses. Our code is written such that if the public IP address is detected to have changed five times in a row, the device will restart. If each request to detect the public IP is randomly sent to either of these two public exits, theoretically, there is a 1 in 2^5 chance, or a 1/32 probability, that the video server will restart due to detecting a different public IP.
I'll give you a command to increase the number of detections to 10 times, so there will only be a 1‰ probability of this issue occurring. That basically means it's highly unlikely for this problem to appear.
cat > /etc/kvmd/override.yaml << EOF
janus:
check:
retries: 10
EOF
run this command and restart kvm
Wow ok yes that’s correct - I do have two ISPs that I load balance between. I put in the code, rebooted it, and will try meetings over the next few days.
If you don’t mind me asking though.. why in the world would you reboot the device under such conditions? I’m not a developer, but I am an IT professional since the 90’s lol.. wouldn’t it be better to toss up a warning message? I’m thinking because many situations where this type of device could reside (e.g. a data center) would definitely have multiple possible IPs.
This is the logic from PIKVM. Originally, the idea was that if your router re-dials and obtains a new public IP address from the ISP, then the WebRTC client should restart to use the latest public IP address, ensuring that your STUN hole punching can succeed. However, it did not consider that if you have load balancing across multiple ISPs, the public IP address would constantly change, causing the WebRTC client to restart repeatedly.
I've also forwarded the box to you because this issue is now returned for me when trying to access my reemote computer.
Got the email you sent. I'll reply soon.