A few notes on the rescue use of the Debian Sarge installation media

Unfortunately the Sarge installation media has no resuce mode as the Woody media had. You can use the Woody CD if you don't need a newer kernel than 2.4.18, but if you need newer ones for hardware like SATA drives, the BroadCom TG3 NIC, or newer Compaq/HP Smart Arrays (there are a lot of other things that need newer kernels), you seem to be out of luck.

But you can boot from the installation media using the expert (Kernel 2.4.27) or expert26 (Kernel 2.6.8) options, paying attention to not overwrite anything on your system. You can do this using the full CD, the DVD, the NetInst CD, the TFTP netboot way, or the USB media (if your machine supports this).

After the boot you should set your country, your keyboard, detect and mount your CD-ROM drive and load installer components from your CD (if booting from a TFTP server the steps are a little bit different because you need to detect the network card and setup the network first, then load installer components and detect your hardware).

After the hardware detection you should be able to access a command prompt from your console 2 (with Alt+F2) and work there. If you have a software RAID you should setup the RAID configuration first because the system does not seem to find it automatically. Please pay attention to not format any of your partitions!

If you need to install some packages or work with lilo or kernel packages you can do a
chroot /target
because the installation mounts your disks under the /target directory. For some operations you need the access to the proc filesystem even in your chrooted environment:
mount -tproc none /proc
makes it available. If you are working on a software raid, I would suggest to wait with any operations until all partitions are synchronized because all operations could be very, very slow!

Another note: since the installation system uses parted instead of fdisk, the device files are somewaht different. Your first IDE disk will be at a path like
/dev/ide/host0/bus0/lun0/disc instead of /dev/hda
Instead of having your partitions at the addresses /dev/hda1 , /dev/hda2 , you fill find them at /dev/ide/host0/bus0/lun0/part1 , ....

For SCSI or SATA drives (these are using a sort of SCSI emulation the device files are
/dev/scsi/host0/bus0/target0/lun0/disc instead of /dev/sda
Instead of having your partitions at the addresses /dev/sda1 , /dev/sda2 , you fill find them at /dev/scsi/host0/bus0/target0/lun0/part1 , ....

Software RAID devices are under /dev/md/0 and so forth instead of /dev/md0 .

You can call parted to see the exat address of your first drive.

These device files are important for operations like mounts or calls to mdadm (maybe to initialize the RAID superblock with the --zero-superblock option).