Link unexpectedly goes down and won't come back up without reboot

gibmat

Hello,

Back in April I put together a new machine to run pfSense on, and it's been working perfectly up until a couple of weeks ago. Now, somewhat randomly one of the four network links (always the same one – re1) goes down and won't come back up. I have to physically reboot the machine, since remote SSH access is only enabled on that link which is going down.

All NICs in my pfSense box have static IPs assigned and are directly connected to switches. No physical changes are occurring and on the switch re1 is connected to the LED indicator is still illuminated for that port. The box is running the latest release of pfSense: 2.2.4-RELEASE.

Here's what I've seen in the system logs:

Aug 26 07:14:42 	kernel: re1: link state changed to DOWN

Aug 29 04:45:24 	kernel: re1: link state changed to DOWN
Aug 29 04:45:29 	kernel: re1: watchdog timeout
Aug 29 04:45:29 	kernel: re1: reset never completed!
Aug 29 04:45:29 	kernel: re1: PHY write failed
Aug 29 04:45:29 	kernel: re1: PHY write failed
Aug 29 04:45:29 	kernel: re1: PHY write failed
Aug 29 04:45:29 	kernel: re1: PHY write failed
Aug 29 04:45:29 	kernel: re1: PHY write failed
Aug 29 04:45:29 	kernel: re1: PHY write failed
Aug 29 04:45:29 	kernel: re1: PHY write failed

Aug 29 07:24:54 	kernel: re1: link state changed to DOWN

Aug 31 03:12:47 	kernel: re1: link state changed to DOWN

In the first block of logs from Aug 29 above, I tried physically unplugging the Ethernet cable from the pfSense box, then plugging it back in. The link appeared to be back up by looking at the switch, however it didn't come back up on the pfSense end.

In all instances, there's no other messages in the system log for hours before the link state change line.

Any ideas as to what is causing this? This box has been running perfectly for several months, and no changes have been made to its configuration or the network it is connected to. This problem only started about two weeks ago.

Thanks,
Mathias

Gertjan

Hi,

Re-arrange your interfaces.
Exclude this realtek nic during testing.
See what happens.

gibmat

I can try rearranging the interfaces, but I can't exclude re1 since all four NICs are carrying active traffic.

I'll report back any changes.