AWS eks documentation change
Summary
Updated Neuron DRA driver recommendations and support details: clarified compatibility with EKS compute options (managed node groups, self-managed nodes, Karpenter, EKS Auto Mode), added topology-aware EFA allocation feature, expanded devicegroup attribute explanations, removed Bottlerocket unsupported note, and adjusted prerequisites.
Security assessment
The changes involve operational clarifications about supported platforms and configurations for Neuron device management. No security vulnerabilities, patches, or security-specific features are mentioned. Updates focus on feature availability, compatibility matrices, and usage examples without any security advisories or threat mitigation content.
Diff
diff --git a/eks/latest/userguide/device-management-neuron.md b/eks/latest/userguide/device-management-neuron.md index 93824a621..129f7da13 100644 --- a//eks/latest/userguide/device-management-neuron.md +++ b//eks/latest/userguide/device-management-neuron.md @@ -17 +17,3 @@ AWS Trainium and AWS Inferentia are purpose-built machine learning chips designe -It’s recommended to use the Neuron DRA driver for new deployments on EKS clusters running Kubernetes version 1.34 or later. The Neuron DRA driver provides topology-aware allocation, connected device subset scheduling, Logical NeuronCore (LNC) configuration, and UltraServer multi-node allocation without requiring custom scheduler extensions. The Neuron device plugin remains supported. +It’s recommended to use the Neuron DRA driver for new deployments on EKS clusters running Kubernetes version 1.34 or later with EKS managed node groups or self-managed node groups. The Neuron DRA driver provides topology-aware allocation, connected device subset scheduling, Logical NeuronCore (LNC) configuration, and UltraServer multi-node allocation without requiring custom scheduler extensions. + +The Neuron DRA driver is not supported with Karpenter or EKS Auto Mode. Use the Neuron device plugin with Karpenter and EKS Auto Mode. The Neuron device plugin also remains supported for EKS managed node groups and self-managed nodes. @@ -24,2 +26,2 @@ Minimum Kubernetes version | 1.34 | All EKS-supported Kubernetes versions -Karpenter and EKS Auto Mode | Not supported | Supported -EKS-optimized AMI support | AL2023 | AL2023, Bottlerocket +EKS Compute | Managed node groups, self-managed nodes | EKS Auto Mode, Karpenter, managed node groups, self-managed nodes +EKS-optimized AMI support | AL2023 (Neuron), Bottlerocket | AL2023 (Neuron), Bottlerocket @@ -29,0 +32 @@ Attribute-based selection | Filter devices by instance type, driver version, an +Topology-aware EFA allocation | DRA-native topology-awareness | Automatic topology-awareness (EKS-optimized AL2023 AMIs only) @@ -37,2 +39,0 @@ Detailed information about the Neuron DRA driver is available in the [Neuron DRA -Using the Neuron DRA driver with Bottlerocket is not currently supported. - @@ -41 +42 @@ Using the Neuron DRA driver with Bottlerocket is not currently supported. - * An Amazon EKS cluster running Kubernetes version 1.34 or later. + * An Amazon EKS cluster running Kubernetes version 1.34 or later with EKS managed node groups or self-managed node groups. @@ -157 +158 @@ The following example requests 4 connected Neuron devices: -The supported `matchAttribute` values for connected subsets are `resource.aws.com/devicegroup1_id`, `resource.aws.com/devicegroup4_id`, `resource.aws.com/devicegroup8_id`, and `resource.aws.com/devicegroup16_id`. +The supported `matchAttribute` values for connected subsets are `resource.aws.com/devicegroup1_id`, `resource.aws.com/devicegroup4_id`, `resource.aws.com/devicegroup8_id`, and `resource.aws.com/devicegroup16_id`. The number in the `devicegroup` attribute name corresponds to the number of Neuron devices in the connected topology group. For example, `resource.aws.com/devicegroup1_id` identifies a single Neuron device, `resource.aws.com/devicegroup4_id` identifies a group of 4 connected devices, and `resource.aws.com/devicegroup8_id` and `resource.aws.com/devicegroup16_id` identify groups of 8 and 16 connected devices respectively. Choose the `matchAttribute` that matches the device `count` in your request so that the allocated devices belong to the same connected topology group. For more information on these attributes, see the [Neuron DRA driver documentation](https://awsdocs-neuron.readthedocs-hosted.com/en/latest/containers/neuron-dra.html). @@ -189,0 +191,4 @@ The following example requests all Neuron devices with LNC set to 1: +### Allocate Neuron devices with aligned EFA interfaces + +See [Topology-aware EFA and GPU/Neuron device allocation](./device-management-efa.html#efa-dra-topology-aware) + @@ -250,0 +256,2 @@ You can verify that Neuron devices are accessible by running the `neuron-ls` too + requests: + aws.amazon.com/neuron: 1