Closed adduarte closed 3 years ago
/lgtm
We might refine this if the need arises, but I'm happy to wait to see if that happens.
/hold Turns out we can proably do all of this by just including the correct fields into standalone_parameters.yaml like so:
#0b:00.0 VGA compatible controller [0300]: Matrox Electronics Systems Ltd. G200eR2 [102b:0534] (rev 01)
NovaSchedulerDefaultFilters: ['RetryFilter','AvailabilityZoneFilter','ComputeFilter','ComputeCapabilitiesFilter','ImagePropertiesFilter','ServerGroupAntiAffinityFilter','ServerGroupAffinityFilter','PciPassthroughFilter','NUMATopologyFilter']
NovaSchedulerAvailableFilters: ["nova.scheduler.filters.all_filters","nova.scheduler.filters.pci_passthrough_filter.PciPassthroughFilter"]
nova::pci::aliases: [{"name": "VGAalias","vendor_id": "102b","product_id": "0534","devic_type": "type-PCI"}]
NovaPCIPassthrough: [{"vendor_id": "102b", "product_id": "0534"}]
KernelArgs: "iommu=pt intel_iommu=on"
here is the documentation on how to do it. Is pretty simple, the only question is where we can add all of those, because the manual expects a separation of controller and compute node, but this all in one single place. https://access.redhat.com/documentation/en-us/red_hat_openstack_platform/16.1/html/configuring_the_compute_service_for_instance_creation/configuring-pci-passthrough
All the necessary parameters to configured pci passthrough can be specified in the local_override.yaml file