[fedora-virt] Windows XP KVM annoyance

[Daniel P. Berrange]

On Thu, Apr 01, 2010 at 07:32:52AM -0400, Tom Horsley wrote:
> > > even if it's the same layout
> > > something almost certainly moved things or flipped a flag of some sort.
> > 
> > Well, what?  it's a bug.
> 
> It certainly seemed a bit like a bug to me, but I also can't imagine
> how it could keep things absolutely identical unless it did something
> like save the details of the pci slots, interrupt assignments, etc.
> in the .xml files it dumps, so when you reload the xml in a new
> distro, it can replicate the details.

As of Fedora 13 we *will* be saving the PCI slots - you'll see in the XML for
any new guest you define, it will have loads of PCI info there. We also now
have ability to specify an exact CPU model, so we no longer susceptible to
changes in host CPU flags.

In Fedora 12 we merely save the machine type eg you'll see the machine type
in the XML be changed from 'pc' to a specific version like 'pc-0.11'. This
is enough to keep some OS happy some of the time, but PCI slots is what we
really need for the full job.

> 
> I've seen the dumped xml - it ain't got that stuff in there :-).
> 
> It isn't so bad when I'm just trying to boot the machine on a new
> distro, but I thought live migration was part of the feature set.
> This seems like an attempt to migrate from f12 to f13 would move
> the hardware out from under me, something I doubt even a linux
> KVM could survive :-).

When we have PCI slot persistence you should be able to live migrate between
different versions of Fedora and have everything just work.


Daniel
-- 
|: Red Hat, Engineering, London    -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org -o- http://virt-manager.org -o- http://deltacloud.org :|
|: http://autobuild.org        -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-   F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|