diff --git a/charts/kubezero-addons/templates/device-plugins/aws-neuron-ds.yaml b/charts/kubezero-addons/templates/device-plugins/aws-neuron-ds.yaml index ae092e96..e785cb25 100644 --- a/charts/kubezero-addons/templates/device-plugins/aws-neuron-ds.yaml +++ b/charts/kubezero-addons/templates/device-plugins/aws-neuron-ds.yaml @@ -12,11 +12,15 @@ spec: type: RollingUpdate template: metadata: + annotations: + scheduler.alpha.kubernetes.io/critical-pod: "" labels: name: neuron-device-plugin-ds spec: serviceAccount: neuron-device-plugin tolerations: + - key: CriticalAddonsOnly + operator: Exists - key: aws.amazon.com/neuron operator: Exists effect: NoSchedule @@ -40,6 +44,14 @@ spec: - inf1.2xlarge - inf1.6xlarge - inf1.24xlarge + - inf2.xlarge + - inf2.4xlarge + - inf2.8xlarge + - inf2.24xlarge + - inf2.48xlarge + - trn1.2xlarge + - trn1.32xlarge + - trn1n.32xlarge containers: - image: "{{ .Values.awsNeuron.image.name }}:{{ .Values.awsNeuron.image.tag }}" imagePullPolicy: IfNotPresent diff --git a/charts/kubezero-addons/values.yaml b/charts/kubezero-addons/values.yaml index 8df68ea9..37b817d4 100644 --- a/charts/kubezero-addons/values.yaml +++ b/charts/kubezero-addons/values.yaml @@ -53,7 +53,7 @@ sealed-secrets: aws-eks-asg-rolling-update-handler: enabled: false image: - tag: v1.7.0 + tag: v1.8.1 environmentVars: - name: CLUSTER_NAME @@ -145,7 +145,7 @@ awsNeuron: image: name: public.ecr.aws/neuron/neuron-device-plugin - tag: 1.9.3.0 + tag: 2.12.5.0 nvidia-device-plugin: enabled: false @@ -184,7 +184,7 @@ cluster-autoscaler: enabled: false image: - tag: v1.25.1 + tag: v1.26.4 autoDiscovery: clusterName: ""