samba - Apr 2003 - Help with critical error

So...

I upgraded via RPMs (actually, I uninstalled 2.2.7, and then installed 2.2.8
because of the dependencies... Is there a better method to upgrading?) this
past weekend.

Monday, end of the day, the entire machine crashed. No unusual
circumstances, it sits there acting as a PDC on a Mandrake 7.2 Pentium 200
with 40MB of ram with not much else going on.

Suddenly the whole machine locks up, with a message like this printed on the
console:

Unable to handle kernel paging request at virtual address 0000fc3b
*pde=00000000
oops:0000
cpu:0
Process smbd (pid 18567, stackpage=c160f000)

...memory addresses, etc...

Kernel panic: Aieee, killing interrupt handler



Any ideas what the heck may have caused this? I sure don't want it to happen
again... My thought is that Linux ran out of memory (only 40MB physical, but
300MB of swap space). It did last 8 hours in full production use with 2.2.8,
so I am not sure that it is an issue with 2.2.8 over 2.2.7.

I rebooted, which took 3(???????) tries before LILO took charge. This is
especially worrisome. Why would it not boot straight away? Is the above
error possibly due to failing hardware?

Brent Torrenga
brent.torrenga@torrenga.com

On Tue, 2003-04-08 at 10:23, Brent Torrenga wrote:
> So...
> 
> I upgraded via RPMs (actually, I uninstalled 2.2.7, and then installed
2.2.8
> because of the dependencies... Is there a better method to upgrading?) this
> past weekend.
make sure you're using 2.2.8a - it includes an important security fix
(different from the one in 2.2.8)
> I rebooted, which took 3(???????) tries before LILO took charge. This is
> especially worrisome. Why would it not boot straight away? Is the above
> error possibly due to failing hardware?
this part certainly has nothing to do with samba...

like you - i suspect flaky hardware.

linux should never panic even it you've run out of memory - you should
just get an out of memory error.

brad
-- 
Bradley W. Langhorst <brad@langhorst.com>