Re: `gpart show` and secondary GPT header

On Jul 28, 2009, at 5:14 PM, Harald Schmalzbauer wrote:

> Marcel Moolenaar schrieb am 16.06.2009 18:28 (localtime):
>> On Jun 16, 2009, at 9:00 AM, Harald Schmalzbauer wrote:
>>> Marcel Moolenaar schrieb am 2009-06-14 07:23 (localtime):
>>> ...
>>>>>>> Jun 13 06:31:42 bigback kernel: GEOM: ad12: the secondary GPT  
>>>>>>> table is
>>>>>>> corrupt or invalid.
>>>>>>> Jun 13 06:31:42 bigback kernel: GEOM: ad12: using the primary  
>>>>>>> only --
>>>>>>> recovery suggested.
>>>
>>> I'm curious what the correct way to recover is.
>>> I also tried dd, but without success.
>>> `gpt` had a recover functionality if I remember corretcly.
>> Recovery is not yet implemented in gpart. That would be
>> the correct way to recover. And yes, gpt(8) has recovery.
>> Thus, gpt(8) is the right way to recover on -STABLE.
>
> Hello, I'd need some help.
> I tried to recover the secondary GPT after overwriting on the same  
> disk.
> I used 'dd if=/dev/ada0 iseek=1 count=1 oseek=1953525167 of=/dev/ada0'
> I thought this must work since the location of the secondary, stored  
> in the GPT header, is correct. But it doesn't.

The table is also duplicated. Thus, a few LBAs are different.
This means the checksum is different. Consequently, you can't
simply dd(8) to recover.
FYI,

-- 
Marcel Moolenaar
xcllnt_at_mac.com