I have a 5 drive RAIDZ volume with data that I'd like to recover.
The long story runs roughly:
1) The volume was running fine under FreeBSD on motherboard SATA controllers.
2) Two drives were moved to a HP P411 SAS/SATA controller
3) I *think* the HP controllers wrote some volume information to the end of
each disk (hence no more ZFS labels 2,3)
4) In its "auto configuration" wisdom, the HP controller built a mirrored
volume using the two drives (and I think started the actual mirroring
process). (Hence on at least on of the drives - a copied labels 0,1).
5) From there everything went downhill.
This happened a while back, and so the exact order of things (including my
botched attemtps at recovery) are hazy.
I tried using Jeff Bonwick's labelfix binary to create new labels but it
carps because the txg is not zero.
The situation now is I have dd'd the drives onto a NAS. These images are
shared via NFS to a VM running Oracle Solaris 11 11/11 X86.
When I attempt to import the pool I get:
root@solaris-01:/mnt# zpool import -d /dev/lofi
status: One or more devices contains corrupted data.
action: The pool cannot be imported due to damaged devices or data.
ZP-8T-RZ1-01 FAULTED corrupted data
12339070507640025002 UNAVAIL corrupted data
I'm not sure why I can't import although 4 of the 5 drives are "ONLINE".
Can anyone please point me to a next step?
I can also make the solaris machine available via SSH if some wonderful
person wants to poke around. If I lose the data that's ok, but it'd be nice
to know all avenues were tried before I delete the 9TB of images (I need the
zfs-list at thismonkey dot com
zfs-discuss mailing list