Enable to share pci devices between numa nodes
Scheduler behaviour allows to boot VM with PCI devices *only* on NUMA nodes associated with these PCI devices. Before that nova booted VMs with PCI devices without checking NUMA nodes. But in some cases such hardcoded behaviour makes problems. User should be able to choose scheduling behaviour to be able to boot VMs on NUMA nodes without associated PCI devices.
Blueprint information
- Status:
- Complete
- Approver:
- Sylvain Bauza
- Priority:
- Low
- Drafter:
- Sergey Nikitin
- Direction:
- Approved
- Assignee:
- Sergey Nikitin
- Definition:
- Approved
- Series goal:
- Accepted for queens
- Implementation:
-
Implemented
- Milestone target:
-
queens-3
- Started by
- Matt Riedemann
- Completed by
- Matt Riedemann
Related branches
Related bugs
Sprints
Whiteboard
Gerrit topic: https:/
Addressed by: https:/
Enable to share PCI devices between numa nodes
Addressed by: https:/
Changed NUMACell to InstanceNUMACell in test_stats.py
Addressed by: https:/
[WIP] PIC NUMA policy PoC
Approved for the Queens release. -- mriedem 20171018
Addressed by: https:/
objects: Add PCI NUMA policy fields
Addressed by: https:/
trivial: Pass InstanceNUMATop
Addressed by: https:/
Add PCI NUMA policies
Addressed by: https:/
trivial: Modify signature of _filter_
Addressed by: https:/
doc: Document PCI NUMA affinity policy
Addressed by: https:/
Address nits in I46d483f9de6776
The last functional patch for this was merged on Dec 30, 2017. -- mriedem 20180102
Addressed by: https:/
Fix typo in release note
Addressed by: https:/
Update spec to reflect reality
