Re: Problem creating filesystem
lsorense@csclub.uwaterloo.ca (Lennart Sorensen) writes:
> On Fri, Sep 09, 2005 at 02:14:08PM +0100, Gaius Mulley wrote:
> > I'm interested - as I had a similar experience yesterday. I used the
> > SID netinst dated 6 Sept iirc. My experience was:
> >
> > 250 GB drive - bios occasionally didn't spot it, the netinst SID
> > couldn't detect the correct IDE kernel modules to load for the disk.
> > I gave up on this disk and moved to:
>
> SATA or IDE 250G?
IDE 250G on the same cable as below..
> > 20 GB drive - mkfs ok (but it took ages - maybe I was having dma
> > timeout during this, and ide reset messages - I didn't look).
> > Later on when installing packages onto the drive I had many ide
> > timeouts, dma problems (just like the ones you reported).
> > However I also tried installing an older Suse 9.1 (64 bit)
> > and it also took ages to mkfs and died during installing packages.
> > Eventually I gave up and used another disk:
> >
> > 10 GB drive - which worked fine.. mkfs, install, later I compiled gcc
> > etc etc
>
> Hmm, a 10G disk is likely old enough to be at most UDMA33. It would
> work fine with a 40wire ide cable.
yes true..
> > My next test will be to replace the IDE cables with brand new ones
> > (the cables came from an older machine which also had the 20 and 10 GB
> > drives).
>
> Make sure to use an 80wire ide cable, since the MB supports faster than
> UDMA33 and if any drive supports faster than that, it will try to use it
> and that only works reliably with an 80wire ide cable. A 20G might be
> new enough to do UDMA66.
ahh, this sounds logical and likely sums up what was happening on
this machine. I've just ordered brand new 80 wire IDE cables just to
be sure - will test next Wednesday..
> > The motherboard is the MSI-9131 SSI Mainboard (K8D Master -F)
> > using the AMD 8131 chipset.
>
> It could of course be a bug in that chipset or the kernel driver for
> that chipset. Or maybe that chipset doesn't agree with certain ide
> drives on some timing issue.
>
> > My drives were also jumpered as master on IDE primary and the CDROM
> > drive was master on IDE secondary. I only ever had one disk and one
> > CDROM drive attached to the motherboard at a time. The CDROM checked
> > its image correctly.
>
> No drive should ever operate as slave without a master. That violates
> the ide specifications and is not required to work reliably at all.
> Sometimes it does, often it does not.
thanks for the advice..
Gaius
Reply to: