Albatron KM18G PRO/RedHat 9.0 - disk errors and system seizures...

Albatron KM18G PRO/RedHat 9.0 - disk errors and system seizures...

[Brian Dushaw]

Dear Linuxers,
    I have a new KM18G PRO motherboard from Albatron, an nforce2 system.
I've put together a generic system with 2X512MB Kingston HyperX memory
sticks, 80 GB Maxtor drive, LG GMA-4020B DVD drive, athlon 2500XP+ (barton).  
I also have USB devices: Epson C80 printer, Linksys wireless ethernet v2.5, 
and a compactflash reader.  I am trying to use the nvidia graphics drivers, 
which seem to work o.k.  A plain-jane system (apologies to Jane) - no 
overclocking, etc.
    I've installed RedHat 9.0 three times now - going on the fourth.  Most
recently I upgraded the kernel with the RedHat update, to similar effect.  The
problem seems to be two fold:  system lockups and disk errors.  The
system lockups happen irregularly.  The disk errors seem to do things like
corrupt the shared libraries (e.g., "bad magic" something, so the libraries
are not recognized as ELF anymore).  In one case, I was upgrading the RedHat 
glibc package when the upgrade crashed; then I was in big trouble.  Things are
very unstable, although everything seems to install o.k..
    Nothing much appears in the logs - dmesg or /var/log/messages.
    Possible theories:  the nforce IDE driver is unstable, the two memory
sticks in "twin bank" configuration don't cut the mustard, flakey motherboard,
motherboard bios settings that linux has trouble with (PnP, ACPI/APIC, MPS
Version Control (set to 1.4)), Memory timings (set to a conservative Optimal,
By SPD), anything else?  I don't think I have an overheating problem.

    Anybody have ideas on what to try?  I got the nforce2 for performance; I
had not realized that the linux support for it was so tenuous...

    One other note:  I do have Windows2K on the system as well.  I did have
a certain amount of trouble installing it, but I figured that was normal.  
Windows 2K seems to be by-and-large stable.

Thx!

Brian Dushaw

(I am not subscribed; please cc me directly, thx.)

Re: Albatron KM18G PRO/RedHat 9.0 - disk errors and system seizures...

[Brad Laue]

Brian Dushaw wrote:
>     I've installed RedHat 9.0 three times now - going on the fourth.  Most
> recently I upgraded the kernel with the RedHat update, to similar effect.  The
> problem seems to be two fold:  system lockups and disk errors. 

The RedHat kernel is massively modified, especially in version 9.0 of 
the distribution. Try a vanilla kernel and see if the system continues 
to misbehave.

Brad

-- 
// -- http://www.BRAD-X.com/ -- //

Re: Albatron KM18G PRO/RedHat 9.0 - disk errors and system seizures...

[Brian Dushaw]

On Sun, 4 May 2003, Brad Laue wrote:

> Brian Dushaw wrote:
> >     I've installed RedHat 9.0 three times now - going on the fourth.  Most
> > recently I upgraded the kernel with the RedHat update, to similar effect.  The
> > problem seems to be two fold:  system lockups and disk errors. 
> 
> The RedHat kernel is massively modified, especially in version 9.0 of 
> the distribution. Try a vanilla kernel and see if the system continues 
> to misbehave.

A fourth install...

I've tried the 2.4.21-rc1-acX kernel (after trials and tribulations with
RedHat - the compilation kept segmentation faulting).  This seems to be
somewhat more stable, but still unstable.  Sorry to be vague, but if I could
be more specific I could probably solve the problem...  I still get X-windows/
system seizures, although I can't say I have any disk errors now  (except 
perhaps for those caused by the lockups.)  I am going to try an offboard
video card to see if that helps (the onboard video apparently uses system
ram for its memory - I don't know if this makes a difference to linux or not).

I tried using only one memory card, which did not solve the problem (i.e.,
twin bank memory is not the problem).

Disk access is about twice as fast (~60 MB/s by hdparm -tT) now with the new 
kernel, if only the system were stable!  I watched Lawrence of Arabia on 
DVD (3.5 hr movie) with this same system in Win2K without a problem, so the 
problem would appear to be linux OS specific.

Still trolling for advice, thanks, 

B.D.