gmirror, gpart and MBR vs GPT in the Handbook

[Mike.]

On 12/2/2013 at 6:26 PM Polytropon wrote:

|On Mon, 02 Dec 2013 12:13:32 -0500, Mike. wrote:
|> My understanding is that MBR can be used with drives only up
to
|> and including 2TB in size.  So if I use MBR, the maximum
drive
|> size I could use would be 2TB.  Is that correct?
|
|Don't confuse partitioning schemes and file systems. Both are
|different things, happening on different "layers". To make it
|simpler than it probably is:
|
|Partitioning =
|	a) MBR with classic tools (fdisk, bsdlabel)
|	b) MBR with modern tool (gpart)
|	c) GPT (gpart)
|	d) Dedicated (only the "bsdlabel part")
|
|RAID concepts =
|	a) mirror
|	b) stripe
|	c) combined and extended forms ...
|
|File system =
|	a) UFS
|		option +U: soft updates
|		option +J: journal
|	b) ZFS
|
|Of course ZFS can handle things like "RAID concepts" already
|internally, whereas UFS would use gstripe and gmirror as
|"little helpers" - it will run on top of them, i. e., you
|initialize the device that represents the whole mirror
|instead of individually dealing with the drives that the
|mirror is constructed of.
|
|Regarding UFS's 2 TB limitation:
|
|http://www.freebsd.org/projects/bigdisk/
|
|Additionally, MBR can be troublesome on bigger hard disks
|or stripes.
|
|http://www.freebsd.org/doc/handbook/bsdinstall-partitioning.html

|
|Today's consensus seems to be:
|
|If you use ZFS, let ZFS deal with everything.
|
|If you use UFS, use gpart for preparation work. Use GPT when
|possible, MBR only in exceptions, and dedicated if and only
|if you really _really_ know what you're doing. :-)
 =============


Following the bsdinstall-partitioning link you cited, and then
following a link on that page, I wind up here:

http://en.wikipedia.org/wiki/Master_boot_record

which states:

"...The organization of the partition table in the MBR limits
the maximum addressable storage space of a disk to 2 TB ...."


That (and other places) is where I got the 2TB limit of MBR into
my head.  Am I misunderstanding that statement?


Thanks.