[smartmontools-support] Calcuting smartctl output to recovery bad-sector on FreeBSD ZFS

[Dan Lukes]

On 11.3.2021 13:44, Budi Janto wrote:
> Mar 11 18:45:42 BEC-STG-P1 kernel: (ada2:ahcich5:0:0:0): 
> READ_FPDMA_QUEUED. ACB: 60 20 28 1f 90 40 b8 01 00 00 00 00

FreeBSD uses LSB, thus it needs to be read as 0x01B840901F = 7386206239

> root at BEC-STG-P1:~ # dd if=/dev/zero of=/dev/ada0 bs=128k count=1 
> seek=268435455
> dd: /dev/ada0: Input/output error

You commanded dd to seek 268435455 block while you claimed block to be 
128k. It's beyond end of the disc.

You should use seek=923275779 bs=4k count=1 instead

> root at BEC-STG-P1:~ # zfs get all | grep recordsize
> pool  recordsize            128K                   default

It's sector size of the ZFS, but we are speaking of physical disc here. 
Your physical disc have 4k sector size. Thus use 4k for dd

With bs=128k you are overwriting more than necessary.

Use physical sector size (4k) not the logical (512).
bs=512 may not work as you expect.

According my experience, some disks do no sector relocation during 
sub-sector write. To be on safe side, always write full physical sector 
to trigger relocation.

>> Sector Sizes:     512 bytes logical, 4096 bytes physical
>> 197 Current_Pending_Sector  0x0012   100   100   000    Old_age   Always 
>>        -       8

It seems there are (at least) two unreadable physical sectors on the 
disc. Consider to run full disc scan to identify the second defect.

Remember - sector relocation doesn't recover the data. Zeroed sector may 
be part of any file on your ZFS volume.

Dan