epic UFS crash

[Šimun Mikecin]

12. 12. 2011., u 20:04, siur <siur.vvm at gmail.com> napisao:

> Some days ago I had situation with server running under FreeBSD 7.3.
> Power supply unit suddenly broke down so server switched off
> incorrectly. After that hardware raid controller marked one disk in
> RAID1 as degrated. We've boot system from other disk and found out,
> that almost all data was gone. To be exactly, it's looks like
> everything created during last uptime had just dissappeared. I mean,
> literally, server looked like from the past -- there was no data newer
> than summer 2010. All databases, websites, user's files -- everything
> was created/modified a year ago.
> Fsck created much staff in lost+found directories, but, for example,
> directories looks like empty for 'ls -la', but hex dump shows
> information (that's, actually more like file, not a directory).
>
> So, could anybody help me find out how it possibly could happen? (my
> creepy story contains very few details, I understand). Does anybody
> saw something like that?
> I appreciate any ideas, all that shit just blowing my mind.

That is what could happen if you are using RAID controller in
write-back mode without battery-backed RAM for write-back.
What kind of UFS is it? UFS1 or UFS2?
UFS+SU? GEOM Journaling?