We currently default that instances should be pinned to the first CPU of
the nova-compute host. We want to remove this default, and have the
default as all CPUs. This patch removes it for fdio and ovs-dpdk
scenarios.
opnfv-tht-pr: 74
JIRA: APEX-187
Change-Id: Iacaa699eb3948b58beff1628dfd1bf008c4c9c5a
Signed-off-by: Tim Rozet <trozet@redhat.com>
hugepages: 1024
hugepagesz: 2M
Compute:
- nova:
- libvirtpin: 1
kernel:
hugepagesz: 2M
hugepages: 2048
hugepages: 1024
hugepagesz: 2M
Compute:
- nova:
- libvirtpin: 1
kernel:
hugepagesz: 2M
hugepages: 2048
hugepagesz: 2M
hugepages: 1024
Compute:
- nova:
- libvirtpin: 1
kernel:
hugepagesz: 2M
hugepages: 2048
hugepages: 1024
hugepagesz: 2M
Compute:
- nova:
- libvirtpin: 1
kernel:
hugepagesz: 2M
hugepages: 2048
intel_iommu: 'on'
iommu: pt
Compute:
- nova:
- libvirtpin: 1
kernel:
hugepagesz: 2M
hugepages: 2048