[Fedora-xen] rawhide guests on F-8 host

Mark McLoughlin markmc at redhat.com
Mon Feb 25 10:22:31 UTC 2008


Hi,
	This build should fix the problem for you:

  http://koji.fedoraproject.org/koji/buildinfo?buildID=39764

On Fri, 2008-02-22 at 12:07 -0700, Orion Poplawski wrote:
> Mark McLoughlin wrote:
> > Hi,
> > 
> > On Thu, 2008-02-21 at 15:08 -0700, Orion Poplawski wrote:
> >> I just build a F-8 x86_64 guest on a F-8 x86_64 host, then did a yum 
> >> upgrade to rawhide.  Now the domain will not start.  Is this expected?
> > 
> > 	Thanks for reporting, I can reproduce here. Try this older kernel for
> > now:
> > 
> >   http://koji.fedoraproject.org/koji/buildinfo?buildID=32046
> 
> Is there any way to install this into my currently dead domain?

	Well, hopefully you still have the F-8 kernel installed and you can
just run that e.g. by doing "virsh console foo" before the grub timeout
expires and choosing the older kernel.

	Alternatively, you could mount the /boot partition and change the
default in grub.conf e.g.

  $> losetup /dev/loop0 /xen/foo.img
  $> kpartx -av /dev/loop0
  $> mount /dev/mapper/loop0p1 /tmp/foo
  $> emacs /boot/grub/grub.conf
  $> umount /tmp/foo
  $> kpartx -d /dev/loop0
  $> losetup -d /dev/loop0

	Once you've booted with a working kernel, then you can upgrade to the
newer one.

	If you don't have any working kernel in the domain, you could try
something like this:

  $> losetup /dev/loop0 /xen/foo.img
  $> kpartx -av /dev/loop0
  $> vgscan
  $> vgchange -ay VolGroup00
  $> mount /dev/VolGroup00/LogVol00 /tmp/foo
  $> mount /dev/mapper/loop0p1 /tmp/foo/boot
  $> chroot /tmp/foo yum --nogpgcheck localinstall kernel-xen-...rpm
  $> chroot mkinitrd --with xenblk -f /boot/initrd-2.6.21.7....rpm 2.6.21.7...
  $> umount /tmp/foo/boot
  $> umount /tmp/foo
  $> vgchange -an VolGroup00
  $> kpartx -d /dev/loop0
  $> losetup -d /dev/loop0

	Or maybe easier, boot the dead domain with a rescue CD.

Cheers,
Mark.




More information about the Fedora-xen mailing list