disklabel differences FreeBSD, DragonFly
Matthew Dillon
dillon at apollo.backplane.com
Sun Jul 30 17:01:06 UTC 2006
:
:>>>>> "Dmitry" == Dmitry Marakasov <amdmi3 at mail.ru> writes:
:
:Dmitry> * Matthew Dillon (dillon at apollo.backplane.com) wrote:
:>> felt that 8 partitions is restrictive. My main home server has 10
:>> and the main DragonFly box has 11.
:>>
:>> There is another solution for FreeBSD folks, however. You *DO*
:>> have four slices to play with. You can put a disklabel with 8
:>> partitions in it on each one (for 32 total). It isn't as
:>> convenient, but it does work.
:
:Dmitry> About `lack' of partitions - don't forget that labels can be
:Dmitry> nested. Just do `bsdlabel -w /dev/ad0s1e` - you'll get
:Dmitry> /dev/ad0s1ea.
:
:Don't also forget that gpt(8) exists and seems to provide for large
:numbers of partitions. It even seems to be compiled into GENERIC by
:default.
:
:Dave.
Yah, well... I'd be a bit leery of using anything more complex then
a basic disklabel. The more complex the setup, the more likely that
a disk crash will become unrecoverable.
I had an issue a little while back with a disk crash where the OS's
insistence on reading numerous sectors at both the beginning and
end of the disk before 'recognizing' it as a disk prevented me from
being able to access the disk at all, even when all I wanted to do
was to make a disk image of e.g. '/dev/ad6' and skip over the bad
sectors. It pissed me off so much I rewrote the code in DragonFly to
not actually try to interpret a slice or partition table unless something
needing the slice or partition table was accessed (like '/dev/ad6s1').
A linked or recursive partition table makes things all that much more
fragile. Recursion isn't as big a deal as linking. Linked partition
tables are a disaster waiting to happen. No thanks!
-Matt
Matthew Dillon
<dillon at backplane.com>
More information about the freebsd-hackers
mailing list