Add nvidia device plugin

kubernetes/apps/kube-system/kustomization.yaml

@@ -12,5 +12,6 @@ resources:
   - ./kubelet-csr-approver/ks.yaml
   - ./metrics-server/ks.yaml
   - ./node-feature-discovery/ks.yaml
+  - ./nvidia-device-plugin/ks.yaml
   - ./reloader/ks.yaml
   - ./spegel/ks.yaml

kubernetes/apps/kube-system/nvidia-device-plugin/app/helmrelease.yaml

@@ -0,0 +1,42 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/fluxcd-community/flux2-schemas/main/helmrelease-helm-v2beta2.json
+apiVersion: helm.toolkit.fluxcd.io/v2
+kind: HelmRelease
+metadata:
+  name: nvidia-device-plugin
+spec:
+  interval: 30m
+  chart:
+    spec:
+      chart: nvidia-device-plugin
+      version: 0.16.2
+      sourceRef:
+        kind: HelmRepository
+        name: nvdp
+        namespace: flux-system
+    metadata:
+      annotations:
+        reloader.stakater.com/auto: "true"
+  install:
+    remediation:
+      retries: 3
+  upgrade:
+    cleanupOnFail: true
+    remediation:
+      retries: 3
+      strategy: rollback
+  values:
+    podAnnotations:
+      configmap.reloader.stakater.com/reload: nvidia-helm-values
+    config:
+      name: nvidia-helm-values
+    runtimeClassName: "nvidia"
+    gfd:
+      enabled: true
+    nfd:
+      enabled: false
+    resources:
+      requests:
+        cpu: 100m
+      limits:
+        memory: 512Mi

kubernetes/apps/kube-system/nvidia-device-plugin/app/kustomization.yaml

@@ -0,0 +1,13 @@
+---
+# yaml-language-server: $schema=https://json.schemastore.org/kustomization
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  - ./helmrelease.yaml
+  - ./runtimeclass.yaml
+configMapGenerator:
+  - name: nvidia-helm-values
+    files:
+      - values.yaml=./resources/values.yml
+generatorOptions:
+  disableNameSuffixHash: true

kubernetes/apps/kube-system/nvidia-device-plugin/app/resources/values.yml

@@ -0,0 +1,12 @@
+---
+version: v1
+flags:
+  migStrategy: "single"
+  plugin:
+    deviceListStrategy: "envvar"
+    deviceIDStrategy: "uuid"
+sharing:
+  timeSlicing:
+    resources:
+      - name: nvidia.com/gpu
+        replicas: 3

kubernetes/apps/kube-system/nvidia-device-plugin/app/runtimeclass.yaml

@@ -0,0 +1,6 @@
+---
+apiVersion: node.k8s.io/v1
+kind: RuntimeClass
+metadata:
+  name: nvidia
+handler: nvidia

kubernetes/apps/kube-system/nvidia-device-plugin/ks.yaml

@@ -0,0 +1,18 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/fluxcd-community/flux2-schemas/main/kustomization-kustomize-v1.json
+apiVersion: kustomize.toolkit.fluxcd.io/v1
+kind: Kustomization
+metadata:
+  name: nvidia-device-plugin
+  namespace: flux-system
+  labels:
+    substitution.flux.home.arpa/enabled: "true"
+spec:
+  targetNamespace: kube-system
+  interval: 10m
+  path: "./kubernetes/apps/kube-system/nvidia-device-plugin/app"
+  prune: true
+  sourceRef:
+    kind: GitRepository
+    name: theshire
+  wait: true