Mark Kane wrote:
On Tue, Aug 22, 2006, at 20:08:57 +0200, Daniel A. wrote:
Hi,
I am currently having a big deal of issues with the ethernet cards on
my server, as well as a possible motherboard problem.
This causes my server to reboot occasionally, but never in a
predictable pattern. A while back, the reboots only occured once
every two weeks - But now they're almost every other hour. I cant
take it anymore.
So, do you know if the "watchdog timeout"s caused by bad network
cards can cause a complete system crash/lockup/reboot? If yes,
shouldn't a kernel core be dumped? And if no kernel is dumped, does
it mean that the issue is potentially with my IDE controller or
harddrive? I noticed that, at a point where I had no internet
connection (And hence, my server was not used - but still turned on),
the server stayed up for 20 days straight without any issues at all -
So my #1 suspicion right now is my network cards.
Hi. I'm not completely sure on this part. The only watchdog timeouts
I've had was indeed a bad used network card, and it happened right
away when I started to configure the box. I replaced it with a brand new
one and it started working perfectly. I don't recall any crashes or
reboots, but it did constantly flood the screen with watchdog timeouts
and I couldn't do anything with the machine.
Are there any other messages in /var/log/messages or dmesg that could
be related, or on the console itself when the machine reboots? Maybe it
is another problem causing the crashes that also causes problems with
the networking. It does seem weird that the onboard and the PCI would be
having the same problems though.
Hi Mark,
I really don't get any other messages, no. The only things remotely
close to errors I get are the watchdog timeouts. Also, I've enabled
all.log, and it does not seem as if there is any pattern to the crashes.
FYI, this is an old Intel Pentium 4 server (2.4GHz). I've removed one
faulty RAM block of 512MB, so I'm down at a measly 512MB instead of 1GB.
I've tested it for 16 hours with memtest (Not 86) on a live system
without any errors. I know that I should probably test it offline with
memtest86 - But I just want the server to run :(
Also, I'm getting a "Kernel Trap 12" or something - Which is the only
thing I could read before the server rebooted, one day when I was lucky
to have a CRT screen attached to the server AND while I was staring at
it. And there are *never* any log traces or anything else, except when I
execute "last", where it just says that my server rebooted at the given
time.
Maybe this will help in debugging it if it is a kernel panic:
http://www.freebsd.org/doc/en_US.ISO8859-1/books/developers-handbook/kerneldebug.html
Considering that I am quite poor - What PCI ethernet cards should I
buy for my server? And could you suggest a cheap motherboard which
isn't bad too?
I like NICs from 3com, and Linksys' LNE100TX has been pretty stable
over the years. Both are around $20 here in the US, but I'd go with the
3com for new purchases.
As far as the motherboard, I'm not sure what type of CPU you have, but
I've been buying MSI boards lately and they have been working out well.
They're pretty well priced also.
Pentium 4 Socket 478. But maybe I should upgrade my CPU/mobo combo if I
find something good and cheap.
By the way, what version of FreeBSD? Perhaps if it's an older version
and these problems have been happening since the beginning, it may be an
incompatibility and already fixed in a newer version. However that
wouldn't explain why it has gotten more frequent over time.
Version 6.1-RELEASE, upgraded from from a 6.0-REL.
-Mark
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hardware
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
