Dale & Yvonne Ogilvie wrote:
> OK, so I have three suggestions thus far:
> tar - simple but you can't boot a .gz
> mondo - no SLES8 rpms, I sense the duct tape and wire rising... apart from 
> this looks promising.
> g4u - ftp based, looks less promising than mondo
> 
> And all of these methods would have to cope with the fact that the raid 
> controller on the server is not supported in the standard SLES kernel. Hmmm. 
> Perhaps I should try mondo on a less tricky to install box...


Boot with knoppix or something that does recognise your raid (with
knoppix, boot 'kernel26', which often helps to support SATA and more
RAID cards)

I'll assume you have two interesting partitions on your hard drive, hda1
for the place you installed your OS, and hda2 for a spare data partition
 that's much bigger.

mkdir /data
mount /dev/hda2 /data
dd if=/dev/hda1 of=/data/dd-image-of-hda1

Later, if you want to restore ...
dd if=/data/dd-image-of-hda1 of=/dev/hda1

This will copy everything of interest except for the MBR boot code.

Note 1 :-

If you want dd to run faster, you need to find an efficient block size
for it to use (and add the parameter bs=65536 if you choose 64k, for
example). I don't know what would be appropriate for your hardware and
kernel, someone else might do this more often and therefore have found
out. Try a few variations, using the 'time' command ... remember that
the first run will always be slower ... and the count parameter, so you
don't have to wait until the whole thing has been copied!

time dd if=/dev/hda1 of=/data/dd-image-of-hda1 bs=65536 count=2048

Note 2 :-

The resulting image will be the same size as your hda partition,
regardless of how much of it is actually being used for data. If you
want to hang on to the image file for a while, but need to reclaim some
of your disk space, compress it ...

bzip2 -9 /data/dd-image-of-hda1

This will take a very long time, though.

-jim

Reply via email to