SG-2440 randomly locks up WIFI and serial until reboot



  • I have a pfSense SG-2440 that will, for no apparent reason, halt its DHCP service on my WIFI bridge and stop outputting to the serial console until I power cycle it.

    This happened last Friday for the first time, and I found that the box kept getting into a boot loop where it would reboot immediately after having finished booting. At the suggestion of a support person at pfSense, I reinstalled the firmware using the vendor distribution. That fixed the boot loop. The next day, I rebuilt my configuration and validated all WAN- and LAN-facing services/features.

    Today, my WIFI client lost its DHCP lease shortly after connecting and fell back to a Link Local configuration. Once again, the SG-2440 did not output to the serial console. I could still connect to the WebUI via WAN, though, and the system logs indicated that all services appeared to be running fine.

    The WIFI interface is bridged to an OpenVPN interface and DHCP is hosted from the bridge interface, so I connected to OpenVPN: my client received a DHCP lease and held it for over 20 minutes until I disconnected. Only the WIFI interface and the serial console seemed to be failing. The status LED on the back of the SG-2440 was solid green, too.

    So I power cycled the box. Serial output started up as it booted. Watching the boot output, I saw these error messages:

    Synchronizing user settings…bridge1: error setting interface capabilities on igb1_vlan192
    bridge2: error setting interface capabilities on igb1_vlan1001
    bridge0: error setting interface capabilities on igb1_vlan172
    done.

    That's probably a red herring, but I want to include the information.

    After the box had booted, I could connect to WIFI and get a DHCP lease again. I could also ping the default gateway and navigate web sites. It's been holding steady for a couple of hours since.

    Has anyone seen anything similar to this? Is it a possible hardware failure, or a kernel bug? The correlation of DHCP failing and serial console output halting leads me to think it's deeper than a simple configuration problem.