need help with ext2 and jfs fundamentals for data recovery

hello,

by a small typo i just managed to created a second loop device onto my existing raid md0, instead of md1. so, while the JFS volume was still mounted, i formated it over with ext2. it is still mounted but unsurprisingly 80% of the directorrys disapeared. (what an irony! the purpose for all of this was backuping to dvd-raid5!)

frankly, i dont see a huge chance to recover more than the lucky inodes, but at least i have to try to make up for my mistake. please lend me your filesystem knowledge.

it would be verry useful to know exactly wich blocks got affected by a complete mkfs.ext2 over a 1709303660K blockdevice. so i could concentrate on the good ones. either its a simple layout or maybe its possible to change the mkfs code to print out those block numbers,

whats left?

there are all paths and filenames within symlinks on a seperate disc. there are also the md5 checksums for every file for verification. sadly i dont have the filesizes for them. but with the names, it might be possible to find the corresponding inodes lowlevel.

the raid5 (8 discs) was degraded during this exident, but the excluded disk is healthy, so there are redundant information about the original jfs state. even the exact state, because it was mounted read only. but i dont think this could be of any use.

i dont have a plan yet. from afar i thought of a block-map where step by step those get ruled out, wich are destroyed, buttigned to allready rescued files, known to be empty. starting blocks of files could be found by greping for the headers of known filetypes. but then what? i guess, the only solution, is to find healthy inodes, wich are cut off from the directory tree and putting them into lost&found. but im afraid, that without the superblocks, it will never be mountable, or fsckable again.

yep. heres the fsck:

jfsfsck -ndev-loop2 jfsfsck version 1.1.7, 22-Jul-2004 processing started: 9-1-2005 7.38.51 The current device is: dev-loop2 The superblock does not describe a correct jfs file system. If devicedev-loop2 is valid and contains a jfs file system, then both the primary and secondary superblocks are corrupt and cannot be repaired, and fsck cannot continue.

well thats what i call a challenge! please let me know any ideas, suggestion or reference

andreas