Strange booting problem

[jd1008]

On 07/01/2015 03:14 PM, Matthew Woehlke wrote:
> On 2015-06-30 19:01, jd1008 wrote:
>> So, how can I proceed with a brand new drive,
>> dd /dev/zero into the first ... say 4K bytes, partition
>> it with fdisk, do not mark any partition bootable, so
>> that bios will skip over it ?
> Don't know why no one's mentioned this, but... you could always just
> install an actual bootloader on the drive that boots from the device
> from which you really want to boot. (I think you can do this with grub...)
>
> Of course, plugging that drive into any other computer might make for an
> interesting experience :-).
>
I am sorry - but ...
the design and implementation of the traditional
(msdos) scheme and ( from what I understand so far
from all the respondents), even gpt, effectively render
the disk to have a signature which BIOS interprets
as a valid partition table AND as bootable, and thus
hangs there looking for what does not exist.

Why the design mixed 2 different things into 1, I have
no idea. But AFAIAC, it sucks and blows atthe same time.

Theoretically, supose I want my PC to have 2 identical drives,
partitioned identically, both bootable.
Say the boot order is cd-rom, drive A, then drive B.
CD-rom is empty, bios moves on to drive A. Somehow
drive A's boot code is corrupt (say somehow all nulled).
PC will never move on to drive B.

So, instead of fixing the issue, we invent new, complex
schemes that require even more complex SW like VM's,
LVM's, .... etc etc.. to solve a problem created by a very
silly error: partitioned means bootable as far as BIOS
is concerned. At least, that is what I have come to understand
and experience (when I removed the boot signature bytes).