Prev: Open XML - sent to cservices@bsi-global.com
Next: I don't understand
From: Tim on 5 Apr 2007 07:03
Jim Howes wrote:

> The Netgear Storage Central is an annoying, nonstandard lump of hardware
> that pretends to be a network disk server.
>
> The quality of software support and the driver software itself on a
> windows box has rendered the valuable contents of the RAID-1 mirrored
> discs in the thing
> unreadable. The box now admits to having two drives, but is unable to do
> anything useful with them.
>
> Digging around the disk, I find what MIGHT be a UFS-like filesystem some
> 96k into the drive,

Plausible - that 96k may be a proprietry RAID superblock.

> and I am busy dumping all 160Gb of it to an image file
> on a
> larger disk where I can play with it (might be a while). I did some work
> a long time ago to read SunOS4 formatted UFS optical disk volumes on a
> linux box..

Good move.

> Does anyone know what sort of filesystem is on these disks?

No. But "file" will probably tell you, if it is an FS that linux can
understand, ie "file <diskimage>" - worth a shot.

> All I've worked out so far is that it appears to be in NUXI format, i.e.
> two
> big-endian words in a little-endian longword. Eek. I could be wrong, it
> could be 'XINU', but "My Documents" in what looks like it may be a root
> directory apepars as "yMD comunest"

Wibble. If "file" doesn't tell you anything, do you see anything interesting
looking at the first few kB of the filesystem - perhaps a literal string
giving an FS type name (allowing for weird endianess).

> Any information welcome
>
> Regards,
> Jim

Sorry - cannot help WRT this device - but can it be "persuaded" to run with
one disk and present the contents? I guess you've tried a certain amount of
this with two drives, just wondering if connecting one drive alone would
let it think it's got a viable degraded array...

Tim
From: Jim Howes on 5 Apr 2007 09:52
Tim wrote:
>>
>> Digging around the disk, I find what MIGHT be a UFS-like filesystem some
>> 96k into the drive,
>
> Plausible - that 96k may be a proprietry RAID superblock.

There's not much in it; a few log messages around the 6k mark, but nothing much
else. This is, I think, the problem, the control blocks have got hosed

>> Does anyone know what sort of filesystem is on these disks?
>
> No. But "file" will probably tell you, if it is an FS that linux can
> understand, ie "file <diskimage>" - worth a shot.

Unfortunately not. I tried 'file' on every individual sector of the first few
megs of what looks like a filesystem, nothing found apart from snippets of text
files and other files that are on the disk, all byte-swapped.

>> All I've worked out so far is that it appears to be in NUXI format

> Wibble. If "file" doesn't tell you anything, do you see anything interesting
> looking at the first few kB of the filesystem - perhaps a literal string
> giving an FS type name (allowing for weird endianess).

I have found a root directory that contains something like

<longword> <longword> <longword> NAME
where the last longword is a big-endian byte count which appears to relate to
the length of the entry, I am guessing that one of the other longwords is
probably an inode number.

The fun part is working out where the inode table is, how big it is, how many
inodes there are in each cylinder group, and where the block numbers are for
each file (and whether there are dind, tind, etc. blocks about)

My current plan involves blanking a drive, creating a volume on it with the
right name, and then copying the 96k superblock to the non-blanked drive,
inserting that in the thing and trying to persuade it that there is a valid
filesystem there. Probably won't work, in which case I'm back to the inode
chasing...

Jim
 | 
Pages: 1
Prev: Open XML - sent to cservices@bsi-global.com
Next: I don't understand