1)Had trouble with poor error message which assumes that a bios originated "write" failure is due to the target partition not having enough disk. Actually partition had enough space but the root partition was beyond cylinder 1024 (motherboard=Intel OR840). Had to guess this might be the cause. 2)Using the nice GUI "back" button each time after a "no space" error to fdisk reformat after 3 times and finally placing the root partition below cyl 1024 the anaconda GUI broke messily, mumbling about device error. It seems there is a "leak" of some kind that is not reset after a failure message and hitting the "back" button all the way to fdisk screen. Anyway a hard reboot and filling in all the numbers again caused all to work (once the partition table was OK). Incidently, the disk scenario is 2*30GB drives, one with NT & win98 the other with Linux 7.0. Repeat this by purposefully putting /root beyond cyl 1024 and telling LILO to write itself there (and NOT the MBR). Then watch the install fail (after an optional format), click "back" until fdisk screen, fiddle and then return to install. Repeat another couple of times, with last time making a /root partition below cyl 1024, observe the crash. This install was about as difficult as installing NT and win98 on same disk. You are close to being easier than Microsoft !
Passing to QA to reproduce.
this problem is verified ... replacing the "linear" line by "lba32" in the lilo.conf allows you to have your /boot above 1024 cylinders, BUT the anaconda installer does not do this ... thanks for your report!
Essentially the same as bug 14351. *** This bug has been marked as a duplicate of 14351 ***