On Mon, Aug 18, 2014 at 10:03:06AM -0400, Matthew Miller wrote:
On Sun, Aug 17, 2014 at 11:15:36PM -0400, Dusty Mabe wrote:
> Just wondering if we should work around  in our cloud/atomic
> kickstart scripts for building images by copying the bits in
> /usr/share/syslinux/mbr.bin to the mbr of the disk.
> I've been doing some 'manual' virt-install builds of the cloud/atomic
> images using the same kickstart/tree from koji and I notice that
> my guests don't boot from disk unless I cat mbr.bin to /dev/vda.
> Thoughts? I can work on drafting and testing a work around if desired.
Yeah -- see comment #13 in the linked bug.
I think the comments from the bug show that there isn't a clear *generic*
solution for this problem. However in the case of building new images for
cloud/atomic I think we can possibly make some assumptions that allow us to
work around the problem.
I am not super familiar with the build process for these images but I assume
that they are building up disk images (qcow2 or raw) that are probably backed by
files. If that is the case then we shouldn't be worried about overwriting any
important data as we know exactly what is in the disk image and what the purpose
of the image is. We should be able to do this in the kickstart files and reference
the bug so that if/when the bug is fixed we can remove the workaround.
Thoughts? Am I oversimplifying this?