Note this is a tracking bug for delivering support of virtio gpu in nova. upstream blueprints related to this have already been approved. https://blueprints.launchpad.net/nova/+spec/libvirt-video-device-models https://blueprints.launchpad.net/nova/+spec/image-metadata-prefiltering i am reopening this as i cannot fine the original BZ for this RFE and need to have a BZ for intenal tracking of this feature.
note the main functionality of this feature has already merged so im going to move this to post alther the second half of this feature which was enhancing the scheduler has been defered to U. i will clone a second bz for the remaining work whic will land in Ussuri/16.1 to track the remaining work and re target this to 16.z. the remaining code should be backportable so we can validate the completed feature after GA. these were tracked as two seperate feature upstream and should have been tracked as two seperate BZs downstreamm too.
the hw_video_model is listed in the useful image porperties upstream https://docs.openstack.org/glance/latest/admin/useful-image-properties.html#image-property-keys-and-values hw_video_model The graphic device model presented to the guest. hw_video_model=none disables the graphics device in the guest and should generally be used when using gpu passthrough. vga, cirrus, vmvga, xen, qxl, virtio, gop or none. and the glance metadefs have also been extended to enable auto polulation of the filed in horizon https://github.com/openstack/glance/blob/0a29d31c1322462ff50ec20aff3c3ffb3bdf37c5/etc/metadefs/compute-libvirt-image.json#L71-L85 an explantion of the feature and the relevent change can be found in the release note https://github.com/openstack/nova/blob/master/releasenotes/notes/extend-libvirt-video-model-support-d630b99ef5039f51.yaml --- features: - | In this release support was added for two additional libvirt video models: ``gop``, the UEFI graphic output protocol device model; and the ``none`` device model. Existing support for ``virtio`` has been extended to all architectures and may now be requested via the ``hw_video_model`` image metadata property. Prior to this release the ``virtio`` video model was unconditionally enabled for ``AARCH64``. This is unchanged but it can now be explicitly enabled on all supported architectures. The ``none`` video model can be used to disable emulated video devices when using pGPU or vGPU passthrough.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2020:3148