On Wed, May 14, 2014 at 10:28:13AM -0600, Eric Blake wrote:> On 05/14/2014 09:47 AM, Francesco Romani wrote: > > Hello, > > > > AFAIK, aliases cannot be set for Graphics devices. I tried this very simple experiment > > with libvirt 1.2.4 (from virt-preview on F20) > > > > relevant part of the input xml: > > > > <graphics type='spice' port='-1' autoport='yes' listen='127.0.0.1'> > > <listen type='address' address='127.0.0.1'/> > > <alias name='display.0'/> > > </graphics> > > Aliases are output-only; they are ignored on input. Libvirt demands > total control over the alias field, because it interacts with qemu via > the alias that libvirt chooses.Furthermore with older QEMU versions, we have no choice in aliases - we had to use what QEMU hardcoded Regards, Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|
----- Original Message -----> From: "Daniel P. Berrange" <berrange@redhat.com> > To: "Eric Blake" <eblake@redhat.com> > Cc: "Francesco Romani" <fromani@redhat.com>, libvirt-users@redhat.com > Sent: Wednesday, May 14, 2014 7:24:13 PM > Subject: Re: [libvirt-users] aliases for graphic devices > > On Wed, May 14, 2014 at 10:28:13AM -0600, Eric Blake wrote: > > On 05/14/2014 09:47 AM, Francesco Romani wrote: > > > Hello, > > > > > > AFAIK, aliases cannot be set for Graphics devices. I tried this very > > > simple experiment > > > with libvirt 1.2.4 (from virt-preview on F20) > > > > > > relevant part of the input xml: > > > > > > <graphics type='spice' port='-1' autoport='yes' listen='127.0.0.1'> > > > <listen type='address' address='127.0.0.1'/> > > > <alias name='display.0'/> > > > </graphics> > > > > Aliases are output-only; they are ignored on input. Libvirt demands > > total control over the alias field, because it interacts with qemu via > > the alias that libvirt chooses. > > Furthermore with older QEMU versions, we have no choice in aliases - we > had to use what QEMU hardcodedThanks to you and to Eric for the answers. If I understood correctly there are no plans to have libvirt-generated aliases for the graphics devices, is that true? If so, I think the only way to distinguish the graphics device is per type (spice, vnc,...), and the final implication of this hypothesis is that no more of one device per type is supported, right? I'm asking because I understood this from the documentation (and from a few experiments) and I just want to be sure my interpretation is 100% correct. Thanks and best regards, -- Francesco Romani RedHat Engineering Virtualization R & D Phone: 8261328 IRC: fromani
On Wed, May 14, 2014 at 03:41:38PM -0400, Francesco Romani wrote:> ----- Original Message ----- > > From: "Daniel P. Berrange" <berrange@redhat.com> > > To: "Eric Blake" <eblake@redhat.com> > > Cc: "Francesco Romani" <fromani@redhat.com>, libvirt-users@redhat.com > > Sent: Wednesday, May 14, 2014 7:24:13 PM > > Subject: Re: [libvirt-users] aliases for graphic devices > > > > On Wed, May 14, 2014 at 10:28:13AM -0600, Eric Blake wrote: > > > On 05/14/2014 09:47 AM, Francesco Romani wrote: > > > > Hello, > > > > > > > > AFAIK, aliases cannot be set for Graphics devices. I tried this very > > > > simple experiment > > > > with libvirt 1.2.4 (from virt-preview on F20) > > > > > > > > relevant part of the input xml: > > > > > > > > <graphics type='spice' port='-1' autoport='yes' listen='127.0.0.1'> > > > > <listen type='address' address='127.0.0.1'/> > > > > <alias name='display.0'/> > > > > </graphics> > > > > > > Aliases are output-only; they are ignored on input. Libvirt demands > > > total control over the alias field, because it interacts with qemu via > > > the alias that libvirt chooses. > > > > Furthermore with older QEMU versions, we have no choice in aliases - we > > had to use what QEMU hardcoded > > Thanks to you and to Eric for the answers. > > If I understood correctly there are no plans to have libvirt-generated aliases > for the graphics devices, is that true? > > If so, I think the only way to distinguish the graphics device is per type (spice, vnc,...), > and the final implication of this hypothesis is that no more of one device per type > is supported, right?Correct, there should be at most one entry per type. Regards, Daniel -- |: http://berrange.com -o- http://www.flickr.com/photos/dberrange/ :| |: http://libvirt.org -o- http://virt-manager.org :| |: http://autobuild.org -o- http://search.cpan.org/~danberr/ :| |: http://entangle-photo.org -o- http://live.gnome.org/gtk-vnc :|