[arch-releng] grub install not listing partitions

[C Anthony Risinger]

On Sat, Dec 24, 2011 at 3:22 AM, Dieter Plaetinck <dieter at plaetinck.be> wrote:
>
> was this supposed to be an answer to my question?
> it's really not clear what your point is.

you asked if <insert list of three examples> could be done, i stated i
used one of them *right now* ... i was indirectly implying it could be
done and offering it forth as a partial answer.  of course, i meant
grub could be independently/safely installed to each raw device
_backing_ a softraid ... but maybe i misinterpreted your 18 words, and
you were reeeeally asking if grub could be installed on a blockdevice
_resulting_ from the softraid assembly (or, for example, the creation
of an LV), in which case the answer is also yes, because virtual
machines do exactly this when passed an LV or any other block device
(or raw file/etc) for use as a rootfs.

does it make sense, generally, at install time, via AIF or any other?
not in any useful way i'm aware; BIOS certainly won't find it -- maybe
thru some crazy chainload from hell, idk -- but it *could* be safely
done, in my understanding.

quite frankly, i'm fairly certain you can go right ahead and install
grub to whatever you want ... be a normal device, or a partition
derived from kpartx scanning a loopback mounted file.  i was simply
non-authoritatively stating that every tool/scheme/standard/etc i've
used from your examples and beyond appear to avoid clobbering the MBR,
and i assumed you were speaking of the MBR as that is where the grub
would reside ... unless we are talking about UEFI booting here, in
which case the (required?) GPT layout sorta guarantees a place for
bootloaders?

soo ... maybe it's not really clear what your question is?  to me,
anyway.  an entire block device formatted with an FS might be
problematic -- IIRC btrfs has an unconditional gap at the start of the
FS -- i'm not 100% on that, and i have no idea about other FS -- even
`mkswap` will preserve the MBR unless your forcibly tell it otherwise.
 seems like the "correct" answer could vary, with a heavy lean toward
"yes'm, it be ok".

alas, if the preceding messages were completely useless in addressing
your question, my apologies, for our time is lost forever :-( ... ever
... ever .... [ever] ... [ever] ... [...]

> also what is an 1ehm-eye-bee?

my overly-exaggerated-and-ultimately-ill-fated attempt at
expressionless, text-powered humor ... along with a general desire to
avoid being pounced.

http://mailman.archlinux.org/pipermail/arch-releng/2011-November/002255.html

... maybe we should start over.  Howdy partna! i'm fairly new round
these parts, but y'all seem like a nice crew to hang with.

-- 

C Anthony