I wonder if you're seeing: LOG.warning('This host appears to have multiple sockets per ' 'NUMA node. The `socket` PCI NUMA affinity ' 'will not be supported.') In the logs at libvirt drive init time, given that this is an AMD EPYC system. It would make sense for `socket` to not work on these, but we should definitely handle the failure better. If you do see it, I think we can verify the feature, but file a new bug to track the more graceful failure.
Could you provide a DB dump of the deployment? Specifically, I'm looking for the the numa_topology JSON blob column in the compute_node row.
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 (Release of components for Red Hat OpenStack Platform 17.1 (Wallaby)), 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/RHEA-2023:4577