Date: Thu, 13 Dec 2001 18:41:40 +0100 From: "Anthony Atkielski" <anthony@atkielski.com> To: "FreeBSD Questions" <freebsd-questions@freebsd.org> Subject: Uptime not so good after all -- why does my net connection go dead? Message-ID: <002201c183fd$6d028210$0a00000a@atkielski.com>
next in thread | raw e-mail | index | archive | help
I thought my FreeBSD system was going to stay up forever, based on what I
had heard, but I had to boot it today. For the umpteenth time, the OS
abruptly and silently decided to stop communicating with my router. It had
no trouble talking to the other PC on my LAN, but it absolutely would not
talk to the router. As far as I could tell, it would not respond to traffic
from the router, nor would it send traffic to the router.
The FreeBSD system contains a 3C905 3-Com 100 Mbps Ethernet NIC connected to
one port of a 3Com five-port switch. The other PC, running Windows NT, is
connected to another port on this same switch. The uplink of the switch is
connected to one of the ports on a NetGear ADSL router.
I've ruled out most potential causes:
- It's not the router; the router continued to talk with the NT machine, but
could not talk with or even ping the FreeBSD machine.
- It's not the WAN; traffic from the router was ignored even for local
purposes, such as syslog logging.
- It's not the switch; I have the same problem when connecting both local
machines directly to the router.
- It's not the FreeBSD machine's NIC; the NIC continues to talk to the NT
machine, and I can also make it work with the router by adding a new IP
address to the interface ("ifconfig xl0 xxx.xxx.xxx.xxx alias").
- Rebooting and/or power-cycling the router have no effect.
- Power-cycling the switch (there is no reboot function) has no effect.
- Taking the interface down with ifconfig up and down has no effect.
Nothing seemed to make the problem go away, so after two weeks of continuous
uptime, I finally bit the bullet and rebooted the machine. The problem was
gone when the machine came back up. I did not power-cycle the hardware.
This means that the NT machine still holds the record for uptime by a very
handsome margin (several weeks).
I'd like to know exactly what is happening inside FreeBSD when it decides to
consign this particular IP address to the Twilight Zone for one particular
destination/source (the router). Obviously, this is a mission-critical
issue, as no production system can afford to be completely deprived of
external network connectivity.
I used to have this problem a lot more until I discovered that the router
was sending out DHCP and RIP traffic to the LAN. I turned that off and the
problem _seemed_ to go away. Unfortunately, it looks like it simply became
less frequent instead. Once in two weeks is still completely unacceptable,
however.
The same thing happened again as I was writing this message. Obviously
there is a problem.
To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?002201c183fd$6d028210$0a00000a>