Recovering a virtual disk on xenserver

Question

ZFS

Good afternoon!

Background - there is xenserver, on it 2 physical disks of 2 TB each are forwarded to the virtual machine on which FreeNas is installed. On a virtual machine, these two disks are combined into one array (zfs) and all this is given to the local network as CIFS.

There was an email failure at home, the computer rebooted and the network drive became inaccessible. FreeNas swears that it cannot get information about the size of the disk, in the log when loading the virtual machine (which lasts for many hours) there are messages

GEOM: ada2: the secondary GPT table is corrupt or invalid. GEOM: ada2: using the primary only -- recovery suggested.

“gpart recover ada2” does not give anything - the error “Input / Output error” crashes

I also tried using gdisk under Linux to check the table sections-errors did not show in gpt tables, while messages “Buffer I / O error on device ....” go to the console without interruption

At the same time, everything is OK with the physical disk (you can make 1 more logical disk and it will be normally available).

Who knows, please tell me how else you can try (it is possible to throw a damaged logical disk into any OS)

PS: there are no backups :(

Answer 1

Was everything built with mirroring in zfs?
Look at the errors in the host system, there is a high probability that io errors are coming.

Answer 2

zfs was built with more disk space.
There are no I/O errors on the host (xenserver)

Answer 3

Try to run mfsbsd, forward these disks there, and let's check the GPT labels.
It is possible that the xfs pool will bud in it.

Answer 4

gpart labels and sizes match?

Answer 5

Here is the error when trying to import a pool illumos.org/msg/ZFS-8000-6x

Answer 6

zpool import -fF share gives the output: cannot import 'share': one or more devices is currently unavailable
zpool import -fFn share does not print anything to the console.

Answer 7

Another interesting fact:

dd if=/dev/ada2 of=/dev/null count=1000
1000+0 records in
1000+0 records out
...

dd if=/dev/ada2p2 of=/dev/null count=1000
<Ошибки ввода/вывода>

Those. with the logical drive itself everything seems to be ok. The partition with zfs was damaged. Do you think it might be worth trying something like this www.lissyara.su/articles/freebsd/file_system/zfs_recovery/ ?