gmirror, gpart and MBR vs GPT in the Handbook
mueller6724 at bellsouth.net
Tue Dec 3 07:09:20 UTC 2013
from Warren Block:
> As far as I recall, MBR expects 512-byte blocks and uses 32-bit
> values. The total number works out to 2TB (2^32*512).
> Just recently, someone in the forums had an external drive, MBR
> format, but 4K blocks. There were problems, although I don't know if
> FreeBSD was at fault.
> I have not tried MBR on a drive larger than 2TB. I suspect it would
> be safe to use the first 2TB and ignore the rest, but that really
> ought to be verified.
There are ways (kludges) to get MBR on a drive > 2TB.
Western Digital My Book 3TB USB 3.0 hard drive was formatted with a single huge NTFS partition with some software.
In order to copy the stuff, which I've never used, I had to use the System Rescue CD.
Subsequently I migrated to GPT, deleted the huge partition, and made my own GPT partitions.
OpenBSD has a kludge with fdisk + disklabel whereby everything over 2TB goes into one (or more) disklabel partitions, but boot partition must be entirely within first 2TB. I never tried that and don't plan to.
fdisk can't see the slice that goes past 2TB, but OpenBSD can, according to what I read on OpenBSD online documentation.
I never tried that. OpenBSD can't access my hard drive at all for lack of support for USB 3.0 and GPT.
DragonFlyBSD latest release, 3.6.0, downloaded, bzip2 -d and dd'ed to USB sticks recognizes my hard drive partitions but can't mount any.
So all I can do with OpenBSD is download the Live USB from liveusb-openbsd.sourceforge.net, run 7z e on the download, which is 7-zipped.
7-zip (archivers/p7zip) is in FreeBSD ports, which is how I built it.
NetBSD supports GPT, but the dk wedges are not as elegant as what FreeBSD and Linux do.
NetBSD seems to be just starting with USB 3.0 support in -current.
NetBSD and OpenBSD still use preassigned device nodes with MAKEDEV such as Linux and FreeBSD had some years back.
This is actually sort of a combined response to two threads, other one being Re: kernel "mismatch" on r256420.
More information about the freebsd-questions