frequent internet death in FreeBSD 6 - no idea why.

[Spadge]

Hi all.

I must admit to being completely stumped by this prooblem I am having 
with my FreeBSD 6 internet gateway. I don't even know where to start 
trying to troubleshoot it.

The symptoms are: every now and then, usually (but not always) when I 
try to open a web page (using a client machine further down the NATed 
network) the browser sits there not opening anything and eventually 
errors out. While this happens, the gateway computer drops its 
connection to the internet completely, the cable modem's "cable" light 
goes out, starts flashing and goes back on again as if I had just 
unplugged the cable. Everything on the NAT side of the gateway 
disconnects from the internet, usually (multiple chat protocols, web 
browsing). Rarely, things on the gateway disconnect from the internet 
(ircd).

I am trying to remember when this started happening, or when I noticed 
it at least (I am guessing they were around the same time) and it seems 
to me that it was around the time I had a pretty serious OS hang-up over 
some HDD 'setfeatures set transfer mode semaphore timeout' problem that 
caused a complete system lockup which required a hard reboot (physical 
button pushing). The HDD that was causing the problem was not the system 
disk, so all the data I am aware of having lost was just that: non 
OS-essential data.

As I say, I am really stuck here. I have tried swapping out NICs, 
changing WAN for LAN, putting the cards in different PCI slots, 
rebuilding world/kernel with dummynet, rebuilding world/kernel without 
dummynet. I am at a loss as to what to try to do next. How do I go about 
finding out what's causing the problem so that I can try fixing it?

spadge at tobermory$ uname -a
FreeBSD tobermory.home 6.0-RELEASE FreeBSD 6.0-RELEASE #2: Wed Nov 30 
02:19:31 GMT 2005     root at tobermory.home:/usr/obj/usr/src/sys/TOBERMORY 
  i386

spadge at tobermory$ diff /usr/src/sys/i386/conf/GENERIC 
/usr/src/sys/i386/conf/TOBERMORY
66c66,80
< device                apic                    # I/O APIC
---
 > # To make an SMP kernel, the next two are needed
 > options         SMP                     # Symmetric MultiProcessor Kernel
 > device          apic                    # I/O APIC
 >
 > # For ipfw/natd
 >
 > options         IPFIREWALL
 > options         IPDIVERT
 > options         IPFIREWALL_DEFAULT_TO_ACCEPT
 > options         IPFIREWALL_VERBOSE
 >
 > # For DUMMYNET packet shaping
 > # options         DUMMYNET
 > # options         HZ=1000
 >

spadge at tobermory$ ifconfig xl0
xl0: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
         options=8<VLAN_MTU>
         inet6 fe80::260:8ff:fe79:bcdb%xl0 prefixlen 64 scopeid 0x1
         inet 192.168.124.199 netmask 0xffffff00 broadcast 192.168.124.255
         ether 00:60:08:79:bc:db
         media: Ethernet autoselect (100baseTX <full-duplex>)
         status: active
spadge at tobermory$ ifconfig xl1
xl1: flags=8843<UP,BROADCAST,RUNNING,SIMPLEX,MULTICAST> mtu 1500
         options=9<RXCSUM,VLAN_MTU>
         inet6 fe80::250:4ff:fee9:114c%xl1 prefixlen 64 scopeid 0x2
         inet 82.18.8.27 netmask 0xffffff00 broadcast 255.255.255.255
         ether 00:50:04:e9:11:4c
         media: Ethernet autoselect (10baseT/UTP)
         status: active

Running ipfw/nated/named/bind. DHCP client to the WAN side, fixed IPs on 
the LAN side.

Not sure what further information is required.

If there's anyone who could help shed some light on this problem, how to 
troubleshoot it etc, please please please do write back, I am at my 
wit's end here.

Thanks.

-- 
Spadge
"Intoccabile"
www.fromley.com