help creating new gmirror > 2TB

Frank Leonhardt frank2 at fjl.co.uk
Thu Aug 31 14:11:09 UTC 2017


On 31/08/2017 01:35, William Dudley wrote:
> I've never heard of AFD, and is having 4K sectors something I have control
> of?  Or is tthat something the manufacturer controls?
>
> My drives are: 2 of Seagate 4TB IronWolf NAS SATA 6Gb/s NCQ 64MB Cache
> 3.5-Inch Internal Hard Drive (ST4000VN008)
>
> Which doesn't have the magic word "AFD" anywhere in the description.
>
> Also, I happen to like the idea of "fixing" the gmirror code to put it's
> metadata in the penultimate sector if it discovers GPT
> in the last sector, but I suppose I shouldn't get my hopes up.
>
> Bill Dudley
>
> On Wed, Aug 30, 2017 at 5:57 PM, Frank Leonhardt <frank2 at fjl.co.uk> wrote:
>
> "I'm sure you'll be okay with MBR assuming it's an AFD (4K sector drive)."
>
> On 29/08/2017 21:12, William Dudley wrote:
>>> Hi,
>>>
>>> I want to create a simple mirror > 2TB on a FreeBSD 10.3 system.
>>>
>>> I have 2 identical 4TB disks.
>>>
>>> The examples in freebsd handbook "geom-mirror" pages show creation of a
>>> 2TB
>>> mirror using
>>> MBR partitioning, and that has an upper limit of 2TB.
>>>
>> I still get a bit worried about this, but I'm 99.9% sure you'll be okay
>> with MBR assuming it's an AFD (4K sector drive). The bodge/workaround
>> works. If it's SAS instead of SATA, all bets are off.
>>
>> Although I use ZFS a lot, I still prefer geom mirror for twin-disk
>> systems. I feel a lot more comfortable booting from it in the event of a
>> failure. ZFS has its good points, but so does UFS.
>>
>> Trying to get geom mirror to work with GPT as it stands just leads to
>> pain. I've taken a look at the code with a view to fixing this is no one
>> else does, but UFS is so un-cool in most circles and I don't fancy doing it
>> alone in case I zap someone's data. it doesn't look that tricky to move the
>> metadata somewhere else, and by checking for a GPT you can select between
>> the old/new block. It's unexpected interactions I'm worried about.
>>
>> I think we're safe with MBR until we pass 16Tb.
>>
>> Regards, Frank.
>>
>>
>>
Sorry about the AFD - I think it's just called AF Disk now (was also 
known as Advanced Disk Format at one time).

Basically, to get around the 2Tb crunch on MBR the size of the 
"hardware" sectors was increased from 512 to 4096 bytes (or octets if 
you want to be modern!) It does reduce the ECC space needed on the disk 
too, but could cause alignment problems if the OS wasn't aware and was 
translating 512-byte blocks. IIRC Microsoft sorted this with Windoze 8.

But it's only putting off the evil hour when we have to move from MBR to 
GPT, and break geom mirror.

Regards, Frank.



More information about the freebsd-questions mailing list