The other day I mailed out about the new PC I'd happily installed RH7.1
on, in a 4-IDE controller motherboard with 2 ATA100 drives and 2 older
EIDE drives, and how I managed to get it booting off drives past the
1st 2.
Optus Cable is due tomorrow arvo to install a cable modem - this is the
reason that I bought a copy of Windows 98 for the new machine. :-(
Anyway, last night I tried to boot up '98, just to make sure all was
well, and it wouldn't boot. The HDD activity light flashes for about
0.1 secs every 8 seconds, and the boot proceeds a little further. It's
like the drive is timing out after an initial successful read.
The old drives are now slaves to the 40Gb drives - on Sunday they
stopped being recognised ,on the 4th IDE controller, and in desperation
I pulled everything apart and shuffled all the drives around so that
the cables reached everything, and fixed up all the drive device names.
The (suspect) 4th IDE controller is no longer in use.
After that it seemed good. One odd thing though, was that while
everything looked good and worked well under RH7.1, when I booted up
RH6.2 from the old 8Gb drive, I got lots of "/dev/hdf: lost interrupt"
errors while it was booting - in particular, while it was doing the
partition check on hdf (the older 4Gb drive); similarly on hde (one of
the new 40Gb drives). Ditto for hde, the new 40Gb drive that's master
of the pair (on the 3rd IDE controller).
The period between "lost interrupt" error messages, remembering, was
probably also about 8 seconds. It took maybe 3 minutes for Linux to
give up on the drives and let the boot proceed without those drives.
So, using RH7.1, I copied the home directories off the 4Gb drive and
onto the 8Gb one, make a symlink for /home on RH6.2, and booted up with
hda=noprobe and hdf=noprobe to get the old RH6.2 system working.
Win98 is on the 1st 3Gb partition on hda and RH7.1 on the next, on the
same drive. And note that it's the 1st primary HDD. Yet Windows 98
won't boot.
So, my question (in desperation) is, does anyone recognise these
symptoms? I'd think it was hardware except that everything works well
under RH7.1. It's just the older OSes - RH6.2 and Win98 that have the
problem.
The problems that made me think the 4th IDE controller had simply
failed seem suspicious/significant. (Looking at old /var/log/messages,
I noticed that every time the 4th IDE controller worked well, the RH7.1
boot had autosensed the BIOS properties of the 2 old drives on that
controller as hdX:dma, and every time it failed it reported them as
hdX:pio. Yet there is no BIOS option for fiddling that. I wondered
whether it was randomly sensing it as one way or the other.)
If no one has experienced this, then there is the option of installing
Windows again. But I'm scared that may wipe out my Linux partitions,
and since it seems almost like a hardware problem, I'm not at all
confident that a re-install wouldn't make things substantially worse.
The only other option I can think of is to get the motherboard
replaced, but again, since everything *was* working fine under Win98,
and *does* work fine under RH7.1, it may not be the solution.
I should finally note that I've installed no extra sw on Windows; I've
not really touched it. The only change was putting LILO on the MBR,
and it was still working fine after that.
All I've changed is the cabling and physical arrangement of the drives,
and some BIOS settings for the order of devices that the system will
boot from.
Last thing I tried was removing "linear" from lilo.conf and re-running
lilo, but that made zero difference.
Any consulting detectives out there? I'm certainly willing to pay for
help on this one. I'm expecting that I'll have to ring Optus and
cancel tomorrow's installation, to be honest. Then arrange another
some weeks down the track. :-(
FWIW, the motherboard is an Abit SA6.
luke
--
SLUG - Sydney Linux User Group Mailing List - http://slug.org.au/
More Info: http://lists.slug.org.au/listinfo/slug
