From be006fe792cce4c1010cfc5ef677efe330ef11e0 Mon Sep 17 00:00:00 2001 From: Kevin Quinn Date: Thu, 23 Apr 2026 13:45:47 +0100 Subject: [PATCH 1/3] TELCODOCS-2272-RN422 NUMA-aware scheduler 500 node scalability release note Co-Authored-By: Claude Opus 4.6 --- modules/rn-ocp-release-notes-new-features.adoc | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/modules/rn-ocp-release-notes-new-features.adoc b/modules/rn-ocp-release-notes-new-features.adoc index 13d5a30303b9..c96b979a1ae2 100644 --- a/modules/rn-ocp-release-notes-new-features.adoc +++ b/modules/rn-ocp-release-notes-new-features.adoc @@ -184,6 +184,15 @@ Detailed information. [id="ocp-release-notes-scale-and-perform_{context}"] == Scalability and performance +// TELCODOCS-2272 +[id="ocp-release-notes-numa-scheduler-scalability_{context}"] +== NUMA-aware scheduler supports clusters with up to 500 nodes + +With this release, you can scale the NUMA-aware secondary scheduler to support clusters with up to 500 nodes. The scheduler defaults to a `Burstable` quality of service (QoS) profile, which reduces baseline resource consumption while allowing the scheduler to scale up during peak loads. If your cluster requires eviction protection for the scheduler pod, you can opt in to a `Guaranteed` QoS profile by annotating the `NUMAResourcesScheduler` custom resource. + +For clusters with 200 or more nodes, you can tune `cacheResyncPeriod`, `podsFingerprinting`, and `infoRefreshMode` settings to balance scheduling accuracy against API server load. + +For more information, see xref:scalability_and_performance/cnf-numa-aware-scheduling.adoc#cnf-topology-aware-scheduler-scalability_numa-aware[Topology-aware scheduler scalability]. // [id="ocp-release-notes-support-sigstore_{context}"] // == Support for sigstore bring your own PKI (BYOPKI) image validation From ae6e8d9c311b839f51c49c7677f6570365266a3a Mon Sep 17 00:00:00 2001 From: Kevin Quinn Date: Thu, 23 Apr 2026 14:52:44 +0100 Subject: [PATCH 2/3] TELCODOCS-2272-RN422-numa Fix release note format and xref Co-Authored-By: Claude Opus 4.6 --- modules/rn-ocp-release-notes-new-features.adoc | 13 +++++-------- 1 file changed, 5 insertions(+), 8 deletions(-) diff --git a/modules/rn-ocp-release-notes-new-features.adoc b/modules/rn-ocp-release-notes-new-features.adoc index c96b979a1ae2..71c7759bcfd2 100644 --- a/modules/rn-ocp-release-notes-new-features.adoc +++ b/modules/rn-ocp-release-notes-new-features.adoc @@ -185,14 +185,11 @@ Detailed information. == Scalability and performance // TELCODOCS-2272 -[id="ocp-release-notes-numa-scheduler-scalability_{context}"] -== NUMA-aware scheduler supports clusters with up to 500 nodes - -With this release, you can scale the NUMA-aware secondary scheduler to support clusters with up to 500 nodes. The scheduler defaults to a `Burstable` quality of service (QoS) profile, which reduces baseline resource consumption while allowing the scheduler to scale up during peak loads. If your cluster requires eviction protection for the scheduler pod, you can opt in to a `Guaranteed` QoS profile by annotating the `NUMAResourcesScheduler` custom resource. - -For clusters with 200 or more nodes, you can tune `cacheResyncPeriod`, `podsFingerprinting`, and `infoRefreshMode` settings to balance scheduling accuracy against API server load. - -For more information, see xref:scalability_and_performance/cnf-numa-aware-scheduling.adoc#cnf-topology-aware-scheduler-scalability_numa-aware[Topology-aware scheduler scalability]. +NUMA-aware scheduler supports clusters with up to 500 nodes:: ++ +With this release, you can scale the NUMA-aware secondary scheduler to support clusters with up to 500 nodes. The scheduler defaults to a `Burstable` quality of service (QoS) profile, which reduces baseline resource consumption while allowing the scheduler to scale up during peak loads. If your cluster requires eviction protection for the scheduler pod, you can opt in to a `Guaranteed` QoS profile by annotating the `NUMAResourcesScheduler` custom resource. For clusters with 200 or more nodes, you can tune `cacheResyncPeriod`, `podsFingerprinting`, and `infoRefreshMode` settings to balance scheduling accuracy against API server load. ++ +For more information, see xref:../scalability_and_performance/cnf-numa-aware-scheduling.adoc#cnf-topology-aware-scheduler-scalability_numa-aware[Topology-aware scheduler scalability]. // [id="ocp-release-notes-support-sigstore_{context}"] // == Support for sigstore bring your own PKI (BYOPKI) image validation From 29f5ad2ec9a805c67909d76afebe893a22255ceb Mon Sep 17 00:00:00 2001 From: Kevin Quinn Date: Thu, 23 Apr 2026 16:05:30 +0100 Subject: [PATCH 3/3] updates to RN to keep it simple --- modules/rn-ocp-release-notes-new-features.adoc | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/rn-ocp-release-notes-new-features.adoc b/modules/rn-ocp-release-notes-new-features.adoc index 71c7759bcfd2..01c877bebc3a 100644 --- a/modules/rn-ocp-release-notes-new-features.adoc +++ b/modules/rn-ocp-release-notes-new-features.adoc @@ -187,7 +187,7 @@ Detailed information. // TELCODOCS-2272 NUMA-aware scheduler supports clusters with up to 500 nodes:: + -With this release, you can scale the NUMA-aware secondary scheduler to support clusters with up to 500 nodes. The scheduler defaults to a `Burstable` quality of service (QoS) profile, which reduces baseline resource consumption while allowing the scheduler to scale up during peak loads. If your cluster requires eviction protection for the scheduler pod, you can opt in to a `Guaranteed` QoS profile by annotating the `NUMAResourcesScheduler` custom resource. For clusters with 200 or more nodes, you can tune `cacheResyncPeriod`, `podsFingerprinting`, and `infoRefreshMode` settings to balance scheduling accuracy against API server load. +With this release, you can scale the NUMA-aware secondary scheduler to support clusters with up to 500 nodes. The scheduler defaults to a `Burstable` quality of service (QoS) profile, which reduces baseline resource consumption while allowing the scheduler to scale up during peak loads. + For more information, see xref:../scalability_and_performance/cnf-numa-aware-scheduling.adoc#cnf-topology-aware-scheduler-scalability_numa-aware[Topology-aware scheduler scalability].