Re: qemu-kvm (1.0) sqeeeze-backports request to fix libvirt (0.9.8-2) and debian linux-ha clusters
-----BEGIN PGP SIGNED MESSAGE-----
On 18/01/12 20:05, Michael Tokarev wrote:
> It is not easy to backport qemu-kvm since it requires backporting
> seabios and vgabios (new qemu-kvm does not work with old *bios) and
> a few other packages like spice, rbd/rados, libiscsi and others.
> While the "other" packages are just extra features and optional,
> *bios (and ipxe) are mandatory, but new versions of *bios will
> break old qemu and old qemu-kvm, so whole qemu needs to be
> backported at the same time too.
> I considered uploading a backport of qemu-kvm to squeeze several
> times (I maintain main qemu-kvm package) but it is still not done
> due to the above reasons.
> The better route will be to fix libvirt to not use features which
> are known broken. Just IMHO anyway.
> And the highest priority is to mark new libvirt as Break'ing old
> qemu[-kvm] properly.
> As for the particular bug you mentioned (#637990), I think I can
> just fix it for squeeze in 0.12. Not sure if I can catch the next
> point release however. And there's one security fix pending for
> squeeze release of qemu-kvm, too but that one is difficult to
> backport (I tried and even succeeded but am still not sure I got it
What would be a feasible solution for the current regressions due to
the changes in libvirt?
Is it possible to get the previous version of libvirt back in
squeeze-backports? I would like to be able to downgrade if an upgrade
of qemu-kvm is to difficult in the short term.
I currently have a broken cluster (and postponed upgrades on several
others to prevent them from breaking) and that ain't good.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)
-----END PGP SIGNATURE-----