On Tuesday 23 May 2006 09:17, Boyd Lynn Gerber wrote:
win 2000 5 GB 1.2 GB swap 118 GB Main Linux, everything all in the one large partition.
Hi Boyd, If I were you, since you can't backup your data, I'd pull hda out of the system and temporarily replace it with a cheaply purchased or borrowed 'spare.' All you need is a known working drive that is sitting around because it's now too small, say 8GB or 10GB or something along those lines. **Partition the spare to match your original hda, only obviously the slices must be much smaller. Remember, the first small slice is a primary and the remaining space is an extended carrying two logical partitions: hdn5 (swap) and hdn6 ("/") Be sure to keep hdn6 large enough to hold the installed 10.1. This step is important because you want the subsequent installation to match your original paths.** Install 10.1 on the spare drive. Treat this installation like it is going to be permanent, too, with respect to your desktop and software selections and configuration of hardware, mount points, etc. This system is hopefully going to be transferred to your original drive. Add the original drive back into the system, jumpered and cabled accordingly so you are booting from the spare and you are able to mount your original "/" partition manually. Boot 10.1 and test the original "/" filesystem for corruption. If, say, that partition is mapping to '/dev/hdd6' you'd run 'reiserfsck /dev/hdd6' as root. *DO NOT* attempt any repairs of any kind if corruption is found. If that happens, you MUST find a way to make a backup while the drive can still be read and, hopefully, all or most of your data can be copied wholesale to a safe place. If the filesystem passes the initial check, then mount it *read only* so you can visually sanity-check it for obvious problems. If you don't find any show-stoppers, you then: Boot to rescue (10.0 media is fine) via CD/DVD and manually copy the newly installed 10.1 *into* your original directory structure, the contents of one directory/subdirectory at a time. This distinction is important: don't overwrite directories unless you are absolutely certain the target directory has nothing unique in it. In other words, you want to overwrite (hence replace) the mangled/midmatched system files but leave everything else (any software you've added plus all your data) intact. Note: When you're done with this process, you'll still have to reinstall any 'extra' or 'third party' software to keep your rpm database consistent with the physical contents of the filesystem, but custom settings and original configuration files for that software should not have been impacted. Exceptions: * move your original /home/boyd to /home/.boyd and copy over the new /home/boyd * backup, don't overwrite, the original /etc/fstab. This is to preserve your non-Linux mount points and options. * don't copy the contents of /proc, /sys or /tmp... just confirm these directories still exist in the original filesystem and that they're empty. When this lengthy process is done, your original drive should now contain a system that will boot and run identically to the system you installed on the spare. Shut down, pull the spare drive out, return your drive cabling and jumpers *precisely* back to their original positions and boot from the original drive. *If* the drive does not boot, you should be able to boot into the system using an installation CD/DVD. Once booted, use YaST's bootloader configuration utility to reinstall Grub. Note: Pay particular attention to the device and slice assignments... I and others have had problems with it becoming 'confused' by 'too many' partitions and not using the correct partition numbers. Everything else, i.e. the paths, are still good but the actual partition numbers are wrong in some cases. If you have a problem accomplishing this recovery, drop back into SLE to discuss it before attempting 'ad hoc' deviations, OK? Good luck & regards, Carl