Can a WAN interface down block the whole system?

Started by elreyquerabio, January 15, 2026, 11:08:27 AM

Previous topic - Next topic
My OPNsense has crashed a couple of times for no apparent reason. The system simply wasn't responding: no connected device was getting an IP address, although it was responding to pings. So I had to reset the system.

I've attached the log, but it's quite perplexing. The only thing I see is that the WAN DHCP6C interface stops responding (confirmed at 2026-01-15T01:41:07 by the Home Assistant log). I really don't see anything else that would indicate a problem.

Can a WAN interface failure cause the entire system to crash? I'm asking the experts to clarify this point because I don't know how to interpret it.

Thank you very much.

January 21, 2026, 03:19:16 PM #1 Last Edit: January 21, 2026, 03:39:14 PM by fastboot
Based on the log you attached, there is no indication of a full system crash or kernel panic.

What the log clearly shows is:
- Repeated PPPoE link failures (LCP: no reply to echo requests, reconnect loops)
- Frequent WAN down / up cycles
- Continuous dhcp6c "Network is down" errors while the PPPoE interface is offline

Eventually a reboot (---<<BOOT>>---), which strongly suggests a manual reset or watchdog, not a spontaneous crash

This behavior is consistent with an unstable WAN / PPPoE connection or NIC driver issue, not with a WAN problem taking down the whole system. Normally even PPPoE reconnect loops do not stop LAN DHCP or the GUI.

If the box becomes unreachable (no GUI, no SSH, no LAN DHCP) while still replying to ping, likely causes are:
- NIC driver lockup
- Hardware issues (RAM, storage, PSU)
- High load or kernel deadlock triggered by repeated interface resets

In short:
The WAN failure itself is almost certainly not the root cause. The log points to WAN instability acting as a trigger, exposing an underlying system or hardware problem.

Suggestion: As you did not mention the hardware of the appliance. I would do a stress test with the machine. For instance I take this quite seriously,  therefore I test my machines full 24-72h via stress-ng and others like memtest+ etc...
I tested my Protectli 6600 the same way. Result was that it's extremely stable.  In other words "Rock Solid".  I know there are other views in this forum about Protectli, but I can tell only good things. Especially about the support.