8.2-STABLE ZFS degraded and "freezing"

[mailinglist at modernbiztonsag.org]

Dear List,

I was running a ZFS pool, consisting of 6 disk (2 raidz1 striped) under
8.2-STABLE. The OS was running from a dedicated geom mirror. One of the
ZFS disks died and the machine just froze. I've disconnected the driver
and now it shows symptoms of very high I/O wait with no response.

The first case was normal boot process and it froze at the "mounting local
filesystems" part. For nearly 60 minutes I saw from the HDD leds that the
ZFS array was working but after that no response. Even ctrl+alt+delete
wasn't working.
In single user mode I've disabled the ZFS rc script and after that the
system booted fine. then I've started ZFS manually and it was stuck at the
mounting state. I saw with zpool iostat that it was reading around
200-500kbytes constantly from the pool for about 60 minutes. During that
time i could use the machine, even check ZFS with zpool status of zfs
status. The only issue was that df only showed one of the 6 zfs
filesystems mounted from the pool. After about 60 minutes of constats HDD
working everything stopped, like when there's a very high I/O wait because
of a faulty HDD. Even commands running from the dedicated gmirror drive
stopped working. I've checked every HDD with badblocks and only one (the
original faulty drive) showed errors.
I've booted up a 9-BETA install CD and from the rescue shell I've run a
scrub process and it was finished in 21 hours and if fixed about 3MB worth
of data from 800GB. Even that didn't help.
Could it be that some metadata got damaged and the zfs filesystems cannot
be mounted? I'm not an expert on ZFS, i'm just starting to learn it and
this was just an experimental system but i just don't want to start it
over. If there is a slight possibility for recovery i would like to do
that, but i don't know how to narrow down the problem.
Please, if you have any idea let me know of it.

Best regards,
Mate