[plug] linux alt to ghost for ntfs disk, impact on mbr/partition info

[Daniel]

Hello Plug,
I am considering using linux dd or g4u to take an image of an ntfs disk
& make a working copy on a smaller disk (then putting ext3 on big one).
Partimage is 'experimental' with ntfs and I aim to avoid Ghost or IC3
purchases (any better ideas than dd / g4u?).

Given the data itself will fit onto the new smaller disk:
What happens when dd 'runs out of track'?
dd gives an error, but does it leave the new disk in a useable format?

If it's going to bomb anyway then do exact counts matter if partition is
nowhere near full ?(calc guestimate below).

What of the mbr and partition info on the new disk?
Two ideas:
- - -
*1st idea:
to do an image of the 10gig partition on larger disk in full-
dd if=/dev/hda1 of=/dev/hdb2/image-hda1.img     

having partitioned and formatted destination with ntfs
build new disk from image-
dd if=/dev/hdb2/image-hda1.img of=/dev/hda1
I'm guessing the mbr would be invalid so might need one of 'their
recovery cds' to make disk boot.
Does a fresh format with ntfs & set active accept the new system & just
work?! (pure optomism)

- - - 
*2nd idea:
>From memory reserved portion is 512b of which 416 is mbr and rest is
disk info ...
If I am right the partition info is after 416b and upto & including 512b
My idea is to partition and format the new disk then
save the partition info:
dd if=/dev/hda skip=416 bs=1 count=96 of=/floppy/partinfo.img

to do an image of the 10gig partition on larger disk in full
dd if=/dev/hda bs=1024 count=10,485,760 of=/dev/hdb2/image-hda.img

ie I think ten gig should be:
1,048,576 (ie 1 megabyte)*1024 to get 1 gig * 10 [so div by 1024 for
this count ....but maybe exact amounts don't matter.]

build new disk from image
dd if=/dev/hdb2/image-hda.img of=/dev/hda
(don't need any counts do we?)
then
dd if=/floppy/partinfo.img of=/dev/hda skip=416 bs=1 count=96
 to put on the partition info we made when formatting the smaller disk
- - -


Am I on track with any of this? Guess there's probably a far better way.
Thanks for your help
Daniel.