[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Problems with grub2/initramfs-tools in chroot



On Thu, Nov 20, 2014 at 9:11 PM, Scott Ferguson
<scott.ferguson.debian.user@gmail.com> wrote:
> On 21/11/14 15:13, Ross Boylan wrote:
>> Over the last week I've repeatedly found my machine unbootable, in the
>> sense I couldn't get to a working system without intervention.
>> Sometimes I couldn't even get the grub2 menu.
>
>
> Tick

I don't understand what you mean by tick.

[snip]

>> Could changing the boot order in the BIOS change the drive mappings
>> and screw up grub that way?
>
> Yes.
>
Whichever drive I booted off of sda and sdb always referred to the
same physical devices.  And that was even though I booted into
different operating system instances.

But I guess the fact that linux can keep the names safe (via udev
remembering serial numbers I suspect) doesn't really speak to the
drive mappings that grub sees in its early operation.

[snip]

>
> I've had a 'similar' problem, in my case it was solved with:-
> grub-install /dev/$Whatever

I either executed that explicitly or, I assume, the grub package
installation machinery did it for me.  But I still had trouble.

grub needs to know where it should jump to. I don't know how it can
figure that out from inside a chroot.  For example, say /boot inside
the chroot is mounted from sdb2. To the chroot, it's just part of the
filesystem.  How is grub-install to figure out that, when it loads
from the start of the disk, it should look for the grub directory (not
/boot/grub) of the appropriate partition?  The only possibility I can
think of would be that it looks in /etc/fstab.

>
> If the problem is GRUB, and you are actually booting from sdb (due to
> the BIOS settings making it the boot device.
>
> Note: double check device names (you can use the GRUB device name) with
> the mount command and use SMART to determine which device it the one set
> to boot from in the BIOS.
>
>
> Hope that helps.
>
>
> Kind regards

Thanks for your response.
Ross


Reply to: