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

@@ -0,0 +1,9 @@
+version: v1
+flags:
+  migStrategy: "none"
+  failOnInitError: true
+  nvidiaDriverRoot: "/"
+  plugin:
+    passDeviceSpecs: false
+    deviceListStrategy: envvar
+    deviceIDStrategy: uuid

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

@@ -22,6 +22,14 @@ spec:
     remediation:
       retries: 3
       strategy: rollback
-  valuesFrom:
-    - kind: ConfigMap
-      name: nvidia-helm-values
+  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

@@ -6,8 +6,8 @@ resources:
   - ./helmrelease.yaml
   - ./runtimeclass.yaml
 configMapGenerator:
-  - name: nvidia-helm-values
+  - name: nvidia-plugin-configs
     files:
-      - values.yaml=./resources/values.yml
+      - config.yaml=./config/config.yml
 generatorOptions:
   disableNameSuffixHash: true

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

@@ -1,10 +0,0 @@
-runtimeClassName: "nvidia"
-gfd:
-  enabled: true
-nfd:
-  enabled: false
-resources:
-  requests:
-    cpu: 100m
-  limits:
-    memory: 512Mi

kubernetes/bootstrap/talos/apps/helmfile.yaml

@@ -10,8 +10,6 @@ helmDefaults:
 repositories:
   - name: cilium
     url: https://helm.cilium.io
-  - name: nvdp
-    url: https://nvidia.github.io/k8s-device-plugin
 
 releases:
   - name: cilium
@@ -20,15 +18,9 @@ releases:
     version: 1.15.4
     values: ["../../../apps/kube-system/cilium/app/resources/values.yml"]
     wait: true
-  - name: nvidia-device-plugin
-    namespace: kube-system
-    chart: nvdp/nvidia-device-plugin
-    version: 0.14.5
-    values: ["../../../apps/kube-system/nvidia-device-plugin/app/resources/values.yml"]
-    wait: true
   - name: spegel
     namespace: kube-system
     chart: oci://ghcr.io/spegel-org/helm-charts/spegel
     version: v0.0.21
     values: ["../../../apps/kube-system/spegel/app/resources/values.yml"]
-    wait: true
+    wait: true