I understand that this feature automatically schedules the guest to be
on a NUMA node to which the PCI device is associated.

For a guest that spans multiple NUMA nodes, is the NUMA node
information associated to PCI devices passed to the guest?
Say a guest is configured to use 2 PCI devices(associated to numa
nodes 0 and 1 on the host). The guest is also configured to have 2
numa nodes.

I tried this with the latest version of openstack (master) and the
numa_node information associated with the device shows up as -1 in the
vm.

> cat /sys/class/net/eth1/device/numa_node
 -1

For such cases, knowing the numa node information in the VM would help
applications choose the device based on the associated numa node.Is
there any blueprint that addresses this issue?

__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: [email protected]?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev

Reply via email to