init commit

clusters/k3s-dgx/gpu-support/gpu-operator-helmrelease.yaml

@@ -0,0 +1,70 @@
+apiVersion: source.toolkit.fluxcd.io/v1
+kind: HelmRepository
+metadata:
+  name: nvidia
+  namespace: gpu-operator
+spec:
+  interval: 10m
+  url: https://nvidia.github.io/gpu-operator
+---
+apiVersion: helm.toolkit.fluxcd.io/v2
+kind: HelmRelease
+metadata:
+  name: gpu-operator
+  namespace: gpu-operator
+spec:
+  interval: 10m
+  chart:
+    spec:
+      chart: gpu-operator
+      version: "v23.9.1"
+      sourceRef:
+        kind: HelmRepository
+        name: nvidia
+        namespace: gpu-operator
+  values:
+    driver:
+      enabled: true
+      image: "nvcr.io/nvidia/driver"
+      version: "535.129.03"
+    operator:
+      defaultRuntime: nvidia-container-runtime
+    toolkit:
+      enabled: true
+      image: "nvcr.io/nvidia/k8s-device-plugin"
+      version: "v0.14.0"
+      env:
+        - name: CONTAINERD_CONFIG
+          value: /var/lib/rancher/k3s/agent/etc/containerd/config.toml
+        - name: CONTAINERD_SOCKET
+          value: /run/k3s/containerd/containerd.sock
+        - name: CONTAINERD_RUNTIME_CLASS
+          value: nvidia
+    devicePlugin:
+      enabled: true
+      image: "nvcr.io/nvidia/k8s-device-plugin"
+      version: "v0.14.0"
+      resources:
+        requests:
+          cpu: 100m
+          memory: 100Mi
+        limits:
+          cpu: 500m
+          memory: 500Mi
+    dcgm:
+      enabled: true
+      image: "nvcr.io/nvidia/dcgm-exporter"
+      version: "3.3.3-3.1.0-ubuntu22.04"
+      resources:
+        requests:
+          cpu: 100m
+          memory: 100Mi
+        limits:
+          cpu: 500m
+          memory: 500Mi
+    migManager:
+      enabled: true
+    gfd:
+      enabled: true
+    node-feature-discovery:
+      enabled: true