debugging kernel panics

[NoOp]

On 06/13/2012 03:39 AM, Adam Funk wrote:
> I had 3 kernel panics in May on a machine that runs all the time as a
> home server and part-time desktop machine.  Each time I found the
> computer unresponsive with two of the keyboard lights blinking, & the
> syslogs & other evidence suggested that it had panicked once during my
> back-up script (which has several combinations of tar piped to gpg) &
> twice early in cron.daily (once during aide).
> 
> I suspected memory because a few months ago I added some secondhand
> memory that a colleague gave me, but I ran Memtest86+ (v4.20) for two
> full passes and got these results:
> 
> Cache  on
> ECC    off
> Test   Std
> Pass   2
> Errors 0
> 
> I'd appreciate any further debugging suggestions.
> 
> Thanks,
> Adam
> 
> 

You might install linux-crashdump
(http://packages.ubuntu.com/lucid/linux-crashdump) for more detailed
"This package installs all the needed infrstracture to get crashdumps
from your running system. If you are experiencing kernel crashes that
cannot be debugged normally (because the problem is infrequent or
panic's the system in such a way as to make information gathering
impossible), then you can install this package to automatically enable
capture of the vmcore on crashes. "
analysis:

$ sudo apt-get install linux-crashdump
$ sudo reboot
<http://lkcd.sourceforge.net/> for info, documentation etc.

In the past, I've found the kernel wiki pages to be helpful -
particularly the bug reporting pages. Perhaps these will help:
https://wiki.ubuntu.com/Kernel
https://wiki.ubuntu.com/Kernel/Debugging/Symptom
https://wiki.ubuntu.com/Kernel/Debugging
https://wiki.ubuntu.com/Kernel/FAQ
https://wiki.ubuntu.com/Kernel/Bugs
and if you are getting Oops before the panic:
https://wiki.ubuntu.com/DebuggingKernelOops

- Gary