ZFS pool data recovery

[CyberLeo Kitsana]

On 02/18/2013 08:06 AM, Jonni Nakari wrote:
> It seems, that while testing suspend to RAM on my machine by running
> "acpiconf -s 3" I managed to break a RaidZ zpool. The machine went to
> sleep fine, but after waking up commands (e.g. reboot) reported I/O
> error. When booting after a hard reset the machine fails to mount the
> root filesystem:
> 
> Trying to mount root from zfs:/vault/root []...
> Mounting from zfs:/vault/root failed with error 5.
> 
> I booted the system from FreeBSD-9.1-RELEASE-amd64-memstick.img and
> tried to import the zpool in the shell:
>  # zpool import -f vault
> cannot import 'vault': I/O error
>     Destroy and re-create the pool from
>     a backup source.
> 
> Command "zpool import -nfF vault" from the memstick system gives no
> output. (I thought it should always give some output?)
> 
> What should I try next to repair the zpool or recover the data? I
> started "zpool import -nfFX vault" but it seems to take quite long.
> 
> Some more information about my system:
> zpool vault consists of 5 block devices:
> whole disks: ada1, ada2, ada3
> cache: ada0s1e
> log: ada0s1d
> 
> The system boots from a UFS filesystem ada0s1a. The zpool and rest of
> the system was created with FreeBSD 9.1-RELEASE.

Does dmesg or syslog show anything with regards to the supposed I/O
errors? Do the disks pass a read-only scan or 'long' SMART test? While
unlikely, it is possible that one of the devices picked that precise
moment to fail.

It's also possible a disk did not appreciate being put to sleep; I've
had several SSDs that simply stopped responding under such conditions,
prompting warranty replacement.

-- 
Fuzzy love,
-CyberLeo
Furry Peace! - http://www.fur.com/peace/