WatchGuard SSL VPN/OpenVPN: Inactivity timeout (--ping-restart), restarting

You know the drill.

Have a couple of WFH users out of ~50 who use our WatchGuard SSL VPN and constantly complain that it disconnects. Other VPN users are fine when these disconnects happen. Authentication is RADIUS-based w/ Azure MFA.

Basic troubleshooting done:

  • Reinstalled the VPN client
  • Updated to the latest version of the VPN client
  • Updated the firmware on the firebox hosting the VPN
  • Tried using the vanilla OpenVPN client
  • Supplied a CAT6 cable for the user to use at home for a wired connection
  • Completely replaced the end user devices
  • Checked that the WFH user’s IP range does not overlap with the internal VPN subnet (it doesn’t)
  • Changed the connection port to UDP 443 instead of TCP
  • Adjusted the encryption algorithm to AES-128-GCM for better performance

None of these worked.
For each user, logs show the exact same message when their VPN disconnects:

Inactivity timeout (–ping-restart), restarting

My understanding is that the VPN server pings the client but does not receive a response, leading to connection termination and restart. I am at my wit’s end and don’t know what else to investigate to prove it’s not on our end.

One user admitted that this problem started after switching ISPs. I even tested one of the replaced devices at my home and could connect continuously for over 8 hours without seeing that message in the logs.

Any suggestions on how to proceed?