Begining of xfs partition destroyed

[smuszlak]

Hi all,
Sorry for this long email, but I do hope you've got a solution to  
recover my lab's data (mostly unanalysed research data, a big part  
being mostly videos).

Hardware is 3ware 9096SA-4I RAID controler with 16 disks of 931.5M  
totalizing 13.6 To on an lvm xfs formated volume in RAID 5. It is in  
VERIFYING state in the 3ware bios.

I don't now original state of the server as I wasn't it's  
administrator and there isn't much written procedures left. (no idea  
if lvm patritions got expanded and so on)

I guess it was the /data partition of the FreeNAS volume on
lvm with VG called rosalbarray, and one LV

As I saw it with df -h there was a small partion prior to the VG, I  
guess it was a /boot partition on the raid but outside of the lvm.

But I did a mistake. Here it is :
I added a small IDE disk to receve a new FreeNAS OS.
At the formating step to install the os, booting form iso on CD from  
usb-CD reader I was on the other server waiting for the install disk  
to load when the keyboard and mouse switch (with a remotte button)
changed machine inextedly and I hit return on the format step but on  
the wrong device :
FreeNAS installed itself at the begining ot the 13.5T of data thus  
destroying partition information and about 2Go of /data as well as the  
supposedly /boot partition...

Now I ve booted on an crunchbang live cd where I've installed lvm2 package.
And of course what I see are 4 partitions created for FreeNAS and  
"unallocated" free space of 13ish To

Since yesterday I am running
xfs_restore

Regarding data, I beleave disk it is about 50% full with about 5 to 7  
To filled up.
Regarding lvm stucture all I have is speculations that needs to be checked.
Regarding original FreeNAS version, I have found an FreeNAS 8.3 CD and  
it is the one I subsequently used.
Is there a way to recover xfs partition size and data (minus the about  
2Go) from the secondary superblocks and then restore it to a mountable  
state ?

when xfs_repair is over (it takes ages) I will post it's result

for now the output is
Phase 1 - find and verify superblock...
bad primary superblock - bad magic number !!!
attempting to find secondary superblock...
with some
found candidate secondary superblock
-- 
Sacha Muszlak