Back to bug 1446311

Who When What Removed Added
Stephen Finucane 2017-04-27 16:18:01 UTC Target Release 12.0 (Pike) 13.0 (Queens)
Red Hat Bugzilla Rules Engine 2017-04-27 16:18:15 UTC Target Release 13.0 (Queens) 12.0 (Pike)
Red Hat Bugzilla Rules Engine 2017-05-02 13:35:09 UTC Target Release 12.0 (Pike) 13.0 (Queens)
Joe Donohue 2017-05-04 14:02:27 UTC Blocks 1389435, 1389441 1419948, 1422243
Stephen Gordon 2017-05-30 21:30:02 UTC URL https://blueprints.launchpad.net/nova/+spec/share-pci-device-between-numa-nodes
RHOS Integration 2017-05-31 04:03:33 UTC Whiteboard upstream_milestone_none upstream_definition_pending-approval upstream_status_unknown
Severity high unspecified
Joe H. Rahme 2017-06-07 14:36:18 UTC CC jhakimra
QA Contact panbalag gszasz
Yoshiki Ohmura 2017-08-01 13:03:52 UTC CC awaugama
QA Contact gszasz awaugama
Blocks 1399897 1427361
Stephen Finucane 2017-08-30 16:02:26 UTC Status NEW ASSIGNED
Assignee eglynn sfinucan
Stephen Finucane 2017-08-30 16:04:24 UTC Status ASSIGNED NEW
Stephen Finucane 2017-08-30 16:05:01 UTC Status NEW ASSIGNED
Scott Lewis 2017-09-29 18:18:22 UTC Target Release 13.0 (Queens) ---
Stephen Finucane 2017-10-04 15:44:49 UTC Link ID OpenStack gerrit 361140
Lee Yarwood 2017-10-09 10:11:08 UTC Target Release --- 13.0 (Queens)
CC lyarwood
jingjing niu 2017-10-11 02:56:19 UTC Blocks 1188000
jingjing niu 2017-10-11 02:58:18 UTC Blocks 1188000
Sanjay Upadhyay 2017-10-30 12:31:09 UTC CC supadhya
Stephen Finucane 2017-12-06 16:35:03 UTC Status ASSIGNED ON_DEV
Stephen Finucane 2018-01-03 10:53:08 UTC Status ON_DEV POST
PnT Account Manager 2018-01-29 19:00:21 UTC CC pbandark
PnT Account Manager 2018-01-31 02:29:35 UTC CC vaggarwa
Joe Donohue 2018-02-08 19:05:12 UTC CC jdonohue
Stephen Finucane 2018-03-22 14:03:02 UTC Status POST MODIFIED
Fixed In Version nova-17.0.1-0.20180302144923.9ace6ed.el7ost
Lee Yarwood 2018-03-22 16:25:26 UTC Fixed In Version nova-17.0.1-0.20180302144923.9ace6ed.el7ost openstack-nova-17.0.1-0.20180302144923.9ace6ed.el7ost
errata-xmlrpc 2018-03-28 12:46:26 UTC Status MODIFIED ON_QA
Lee Yarwood 2018-03-29 09:16:08 UTC Keywords TechPreview
Lee Yarwood 2018-03-29 09:18:31 UTC Blocks 1561961
Lee Yarwood 2018-06-15 04:32:18 UTC Severity unspecified high
errata-xmlrpc 2018-06-27 12:17:51 UTC Status ON_QA RELEASE_PENDING
errata-xmlrpc 2018-06-27 13:31:22 UTC Status RELEASE_PENDING CLOSED
Resolution --- ERRATA
Last Closed 2018-06-27 09:31:22 UTC
errata-xmlrpc 2018-06-27 13:33:18 UTC Link ID Red Hat Product Errata RHEA-2018:2086
jingjing niu 2018-07-11 11:19:25 UTC Flags needinfo?(sfinucan)
Stephen Finucane 2018-07-19 10:22:29 UTC Doc Text Feature:

Added support for PCI device NUMA affinity policies. These are configured as part of the ``[pci]alias`` configuration option(s). There are three policies supported:

- ``required`` (must-have)
- ``legacy`` (must-have-if-available) (default)
- ``preferred`` (nice-to-have)

In all cases, strict NUMA affinity is provided if possible. The key difference between the policies is how much NUMA affinity one is willing to forsake before failing to schedule.

Reason:

These allow you to configure how strict your NUMA affinity should be on a per-device basis or, more specifically, per device alias. This is useful to ensure maximum resource utilization.

Result:

When the 'preferred' policy is configured for a PCI device, nova will now utilize CPUs on a different NUMA node from the NUMA node of the PCI device if this is all that is available. This will result in increased resource utilization with the downside of reduced performance for these instances.
Flags needinfo?(sfinucan)
Stephen Finucane 2018-07-19 10:24:34 UTC Flags needinfo?(sfinucan)
Erwan Gallen 2018-07-24 13:08:39 UTC CC egallen
Doc Type Enhancement Technology Preview
Laura Marsh 2018-08-07 13:14:10 UTC CC lmarsh
Flags needinfo?(sfinucan)
Laura Marsh 2018-08-07 16:56:18 UTC Doc Text Feature:

Added support for PCI device NUMA affinity policies. These are configured as part of the ``[pci]alias`` configuration option(s). There are three policies supported:

- ``required`` (must-have)
- ``legacy`` (must-have-if-available) (default)
- ``preferred`` (nice-to-have)

In all cases, strict NUMA affinity is provided if possible. The key difference between the policies is how much NUMA affinity one is willing to forsake before failing to schedule.

Reason:

These allow you to configure how strict your NUMA affinity should be on a per-device basis or, more specifically, per device alias. This is useful to ensure maximum resource utilization.

Result:

When the 'preferred' policy is configured for a PCI device, nova will now utilize CPUs on a different NUMA node from the NUMA node of the PCI device if this is all that is available. This will result in increased resource utilization with the downside of reduced performance for these instances.
This release adds support for PCI device NUMA affinity policies, which are configured as part of the '[pci]alias' configuration options. Three policies are supported:

'required' (must have)
'legacy' (default, must have, if available)
'preferred' (nice to have)

In all cases, strict NUMA affinity is provided, if possible. These policies allow you to configure how strict your NUMA affinity should be per PCI alias to maximize resource utilization. The key difference between the policies is how much NUMA affinity one is willing to forsake before failing to schedule.

When the 'preferred' policy is configured for a PCI device, nova uses CPUs on a different NUMA node from the NUMA node of the PCI device, if it is available. This results in increased resource utilization, but performance is reduced for these instances.
Laura Marsh 2018-08-07 16:57:05 UTC Doc Text This release adds support for PCI device NUMA affinity policies, which are configured as part of the '[pci]alias' configuration options. Three policies are supported:

'required' (must have)
'legacy' (default, must have, if available)
'preferred' (nice to have)

In all cases, strict NUMA affinity is provided, if possible. These policies allow you to configure how strict your NUMA affinity should be per PCI alias to maximize resource utilization. The key difference between the policies is how much NUMA affinity one is willing to forsake before failing to schedule.

When the 'preferred' policy is configured for a PCI device, nova uses CPUs on a different NUMA node from the NUMA node of the PCI device, if it is available. This results in increased resource utilization, but performance is reduced for these instances.
This release adds support for PCI device NUMA affinity policies, which are configured as part of the '[pci]alias' configuration options. Three policies are supported:

'required' (must have)
'legacy' (default; must have, if available)
'preferred' (nice to have)

In all cases, strict NUMA affinity is provided, if possible. These policies allow you to configure how strict your NUMA affinity should be per PCI alias to maximize resource utilization. The key difference between the policies is how much NUMA affinity one is willing to forsake before failing to schedule.

When the 'preferred' policy is configured for a PCI device, nova uses CPUs on a different NUMA node from the NUMA node of the PCI device, if it is available. This results in increased resource utilization, but performance is reduced for these instances.
Laura Marsh 2018-08-07 17:01:19 UTC Doc Text This release adds support for PCI device NUMA affinity policies, which are configured as part of the '[pci]alias' configuration options. Three policies are supported:

'required' (must have)
'legacy' (default; must have, if available)
'preferred' (nice to have)

In all cases, strict NUMA affinity is provided, if possible. These policies allow you to configure how strict your NUMA affinity should be per PCI alias to maximize resource utilization. The key difference between the policies is how much NUMA affinity one is willing to forsake before failing to schedule.

When the 'preferred' policy is configured for a PCI device, nova uses CPUs on a different NUMA node from the NUMA node of the PCI device, if it is available. This results in increased resource utilization, but performance is reduced for these instances.
This release adds support for PCI device NUMA affinity policies, which are configured as part of the “[pci]alias” configuration options. Three policies are supported:

“required” (must have)
“legacy” (default; must have, if available)
“preferred” (nice to have)

In all cases, strict NUMA affinity is provided, if possible. These policies allow you to configure how strict your NUMA affinity should be per PCI alias to maximize resource utilization. The key difference between the policies is how much NUMA affinity you're willing to forsake before failing to schedule.

When the “preferred” policy is configured for a PCI device, nova uses CPUs on a different NUMA node from the NUMA node of the PCI device, if it is available. This results in increased resource utilization, but performance is reduced for these instances.
Stephen Finucane 2018-08-21 15:45:38 UTC Flags needinfo?(sfinucan) needinfo?(sfinucan)
Stephen Finucane 2018-11-07 17:22:10 UTC Blocks 1647536
Stephen Finucane 2018-11-07 17:23:47 UTC Summary [RFE] Optional NUMA affinity for SR-IOV devices [RFE] Optional NUMA affinity for PCI devices
Lee Yarwood 2018-11-16 16:02:45 UTC Blocks 1650606
Stephen Finucane 2019-01-11 11:50:00 UTC CC vinayak.ram
Vinayak 2019-01-16 14:02:28 UTC Status CLOSED MODIFIED
Resolution ERRATA ---
Flags needinfo?(sfinucan)
Keywords Reopened
Stephen Finucane 2019-02-05 10:42:20 UTC Status MODIFIED CLOSED
Resolution --- ERRATA
Flags needinfo?(sfinucan)
Last Closed 2018-06-27 13:31:22 UTC 2019-02-05 10:42:20 UTC
Antonios Dakopoulos 2019-02-06 13:57:23 UTC CC adakopou
Stephen Finucane 2019-08-02 11:09:38 UTC Comment 13 is private 1 0
Russell Bryant 2019-09-09 15:45:54 UTC CC rbryant
Bertrand 2019-10-02 16:04:57 UTC Blocks 1757886
Lee Yarwood 2019-11-22 10:43:00 UTC Blocks 1775575
Lee Yarwood 2019-11-22 10:47:09 UTC Blocks 1775576
Karrar Fida 2020-01-16 20:06:32 UTC Blocks 1783354
Blocks 1791991

Back to bug 1446311