On Tue, Sep 03, 2013 at 05:26:30PM +0100, Dan Kenigsberg wrote:
On Tue, Sep 03, 2013 at 05:13:55PM +0200, Joop wrote:
> Joop wrote:
> >Dan Kenigsberg wrote:
> >>I'm only guessing here, but has there been a change in selinux-policy?
> >>Is selinux enforcing?
> >>
> >>Could you provide the output of
> >>
> >>ls -lRZ /etc/pki/vdsm ?
> >>Does these paths show up in /var/log/audit.log?
> >Attached both from working and non-working host and what the
> >update did on the now non-working host.
> >
> No /etc/pki/vdsm in /var/log/audit/audit.log. Only relation is a
> failure notice for starting test06 by libvirt, but no why that I can
> find.
those unconfined_* are fishy, but they exist on the working setup as
well. If Dan Walsh were not listening I'd suggest to try it with
setenforce 0, nonetheless.
Excuse my continued guesswork, but how about
$ chown vdsm.qemu /etc/pki/vdsm/keys/vdsmkey.pem
we once had a libvirt bug were auxiliary group kvm was not set on the
qemu process. Maybe history repeats itself.
Joop, do you have any news for us? More and more people are crashing
into this :-(
Bug 1006394 - after update machines fail to up with spice client
interface