Sorry, sorry, I was 'tuning' the IDE chips in a wrong way. Since ever
and on every machine (~ 50) I'm using
hdparm -c1 /dev/ha[a-d]
to enable fast 32-bit I/O.
But with i820 chipsets, this will corrupt data!!
After 100 tests and reboots I changed to
hdparm -c3 /dev/ha[a-d]
and the world is okay. Thank you Wintel once again, I have never seen
this kind of problem with any ServerWorks or VIA chipset.
Russell Cattelan schrieb:
>
> Simon Matter wrote:
>
> > Hi everybody,
> >
> > I'm using XFS since 1.0 and it has worked perfectly everywhere. I used
> > it on top of MD Raid5 and on LVM on top of MD Raid5.
> >
> > But now I came across a problem and I don't know how to solve it.
> >
> > I installed RH7.1-XFS on a new DELL Precision 220 Workstation, one PIII
> > 933, 256M RDRAM, Intel Pro/100 ethernet. I removed the CDROM and
> > installed 4 IBM IC35L060AVER07 60G HD's. The onboard Sound and USB is
> > disabled. The chipset is i820 (Camino).
> >
> > Initial install went fine, using 180GB MD Raid5. Then I tired to update
> > some RPM's and rpm segfaulted. Later after a reboot, it went fine. The
> > next reboot I saw some messages that /sbin/mkkerneldoth segfaulted,
> > another reboot went fine. I started to play with the noapic and pci=xxx
> > options but no result. 1 of 5 times I boot the system, it doesn't work
> > properly and many applications segfault. beside that everything seems
> > okay.
> >
> > Any idea what I can do?
>
> Sounds like bad hardware.
>
> >
> >
> > Thanks
> > Simon
> >
> > --
> > Simon Matter Tel: +41 61 695 57 35
> > Fr.Sauter AG / CIT Fax: +41 61 695 53 30
> > Im Surinam 55
> > CH-4016 Basel [mailto:simon.matter@xxxxxxxxxxxxxxxx]
> >
> > ------------------------------------------------------------------------
> > Name: dmesg
> > dmesg Type: unspecified type (application/octet-stream)
> > Encoding: base64
> >
> > Name: lilo.conf
> > lilo.conf Type: unspecified type (application/octet-stream)
> > Encoding: base64
>
> --
> Russell Cattelan
> cattelan@xxxxxxxxxxx
|