Civileme, Problem is now solved. I dropped my clock rate back to normal (350, <sniff>) and I got past the hanging point. Then I think I got as far as selecting the security level, and the fsck of my Linux partitions kept giving a "code 4 or signal 0" error (that may not be exact, sorry again).
So I booted rescue and fsck'ed my partitions manually, correcting errors like no "lost+found" and several other things. I was then able to complete the install successfully. FYI, the "ide0=noautotune" didn't seem to have a visible effect and I found the hard drive CHS values reported by the BIOS and fdisk were the same. Thanks for steering me in the right direction, and for the valuable hard drive info. (I assembled this machine long before I even considered Linux.) - Kathy > From: civileme <[EMAIL PROTECTED]> > To: [EMAIL PROTECTED] > Subject: Re: [newbie] Partition Check Hangs > Date: 17 May 2002 12:01:05 -0800 > > That is really a partition check hang. > > 1. Drop your clock rate. > > 2. make sure you have a 40-pin cable on the drive, not an 80 to limit > UDMA to 33MHz or less. > > 3. Next drive acquisition, please consider NOT buying a brand which > supports (officially) only Windows and Solaris (and the Solaris for SCSI > drives at that). There is more trouble with WDs than all the others > combined, in linux. > > 4. F1 at the splash screen and > > linux ide0=noautotune > > because WDs usually rate themselves higher than they run well at. > > (Example WD200AB on a PIII 933 with VIA KT133A running RH 7.1 was set to > UDMA66 and producing a rousing 1.6M/sec when tested with > > hdparm -t /dev/hda > > 5. By dropping the data rate to 33MHz, a slight performance > improbvement was noted, from a 1.6Mb/s at 66 to 9.4 Mb/s at 33. > Obviously there were many errors and rereads at the higher speed. > > Finally on some WDs, there is a problem with estimation--since they are > single-platter with two heads physically, their Cylinder, Head ans > Sector numbers for Logical Block Addressing must be calculated. A few > models show more capacity (closer to tru physical capacity) at a > solution that provides 240 heads rather than 255. The difference is > between the 32-bit math of the kernel and the 16-bit math of the BIOS. > > So see what the CHS numbers are in the BIOS and compare those to > > F1 at the splash screen and > > rescue > > booting a normal rescue and > > # fdisk -l /dev/hda > > anfd looking at the reported CHS numbers. > > By now, we think we have closed off those approximation errors but there > is always the chance that there is a cheap drive out there that we > missed in the exception table, and there is a method of overriding the > calculated CHS numbers with a kernel message, so check. If the numbers > are different, bingo. > > linux has much tighter timing that Windows in its drivers, so > overclocking is often enough to kill a few devices... It is better > unless you have top-notch hardware to run linux at designed clock speeds. > > Civileme
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com
