Server Fubar'ed - Ideas welcomed

[Alma J Wetzker]

Michael Hipp wrote:
> Well the problem I thought I had with a full /home wasn't even close.
> 
> Msg on console:
> Filesystem "hda2": Corruption of in-memory data detected. Shutting down 
> filesystem: hda2. Please unmount the filesystem, and rectify the 
> problem(s).
> 
> Hda2 is the root fs. Thought it might be a bad disk, but 2 different 
> tools say it is ok.
> 
> Found the CPU fan to be turning at about 3 RPM. Replaced it. I'm now 
> assuming the thing crashed originally due to overheating.
> 
> Anyway, it won't run for long at a time before the message above 
> reappears and only a hard reboot will bring it back to life. (Has the P4 
> processor been damaged?)
> 
> The kernel update linux-image-2.6.10-5-386 is in a partially installed 
> (severely inconsistent) state. It won't remove and it won't install 
> without crashing the box. Evidently it crashed originally when 
> installing that update. Coincidence?
> 
> dpkg says to reinstall it. But there's no option for "reinstall", only 
> "install". And that crashes.
> 
> Any ideas?

Can you use knoppix to tweak the kernel install issue?  I understand 
that this is a production box, but can you get it to fail under knoppix?

If you suspect overheating, it could be any component(s) of the main 
board or connected to the main board that have failed.  My inclination 
would be to fix the kernel update with a bootable distro and replace the 
mainboard and processor.  If curiosity (or economics) drives you to do a 
full diagnostic and post mortem, start with something that you can get 
working reliably.  Good luck!

     -- Alma