+bfe0: watchdog timeout -- resetting

[Jan Catrysse]

Dear subscribers,

I am currently running a production gateway:
FreeBSD 6.2 STABLE
Onboard Broadcom BCM4401 Fast Ethernet (bfe0)
3 x PCI intel Gigabit Lan (em0, em1, em2)

This machine is my firewall/gateway running PF.

In my /var/log/messages I found:
Nov 20 10:22:42 techdesk kernel: bfe0: watchdog timeout -- resetting
Nov 20 10:22:42 techdesk kernel: bfe0: link state changed to DOWN
Nov 20 10:22:44 techdesk kernel: bfe0: link state changed to UP

It seems to happen every day or two and from time to time twice a day.

The bfe0 is my external interface connected to the internet (6Mb/s).

I don't have any free PCI slots to put in another card for testing;
furthermore I don't really like putting down the server for too long
either.

The system is mainly running:
pf (no queuing / no balancing / no red / ... )
isc-dhcpd (a small number of internal requests)
named (a small number of internal requests)
ftp-proxy
pftpx

I don't think the external connection has to suffer from heavy loads, it
only accepts 6Mb/s max. Behind the gateway the only heavy incoming
traffic is for the FTP server. Only a bunch of simultaneous connections
but at high speed...

To be frank the CPU and Memory are idle most of the time ;-) .

Someone has a clew on what this could be or how to do further
diagnostics?

Kind regards,
Jan Catrysse