ce7cb702218693d08ea6d83464d2457c9250027c

vendor/infrastructure/nvidia/helm-releases.yaml

@@ -0,0 +1,123 @@
+apiVersion: helm.toolkit.fluxcd.io/v2
+kind: HelmRelease
+metadata:
+  name: gpu-feature-discovery
+  namespace: nvidia
+spec:
+  interval: 15m
+  releaseName: gpu-feature-discovery
+  targetNamespace: nvidia
+  chart:
+    spec:
+      chart: nvidia-device-plugin
+      version: 0.20.1
+      sourceRef:
+        kind: HelmRepository
+        name: nvidia-device-plugin
+        namespace: flux-system
+  install:
+    remediation:
+      retries: 3
+  upgrade:
+    remediation:
+      retries: 3
+  values:
+    runtimeClassName: nvidia
+    resources:
+      requests:
+        cpu: 5m
+        memory: 64Mi
+      limits:
+        memory: 128Mi
+    devicePlugin:
+      enabled: false
+    gfd:
+      enabled: true
+      securityContext:
+        privileged: false
+        allowPrivilegeEscalation: false
+        capabilities:
+          drop:
+            - ALL
+    nfd:
+      enabled: false
+---
+apiVersion: helm.toolkit.fluxcd.io/v2
+kind: HelmRelease
+metadata:
+  name: nvidia-device-plugin
+  namespace: nvidia
+spec:
+  interval: 15m
+  releaseName: nvidia-device-plugin
+  targetNamespace: nvidia
+  dependsOn:
+    - name: gpu-feature-discovery
+  chart:
+    spec:
+      chart: nvidia-device-plugin
+      version: 0.20.1
+      sourceRef:
+        kind: HelmRepository
+        name: nvidia-device-plugin
+        namespace: flux-system
+  install:
+    remediation:
+      retries: 3
+  upgrade:
+    remediation:
+      retries: 3
+  postRenderers:
+    - kustomize:
+        patches:
+          - target:
+              kind: DaemonSet
+              name: nvidia-device-plugin-mps-control-daemon
+            patch: |
+              $patch: delete
+              apiVersion: apps/v1
+              kind: DaemonSet
+              metadata:
+                name: nvidia-device-plugin-mps-control-daemon
+  values:
+    runtimeClassName: nvidia
+    resources:
+      requests:
+        cpu: 5m
+        memory: 64Mi
+      limits:
+        memory: 128Mi
+    nfd:
+      enabled: false
+---
+apiVersion: helm.toolkit.fluxcd.io/v2
+kind: HelmRelease
+metadata:
+  name: dcgm-exporter
+  namespace: nvidia
+spec:
+  interval: 15m
+  releaseName: dcgm-exporter
+  targetNamespace: nvidia
+  dependsOn:
+    - name: nvidia-device-plugin
+  chart:
+    spec:
+      chart: dcgm-exporter
+      version: 4.8.4
+      sourceRef:
+        kind: HelmRepository
+        name: nvidia-dcgm-exporter
+        namespace: flux-system
+  install:
+    remediation:
+      retries: 3
+  upgrade:
+    remediation:
+      retries: 3
+  values:
+    runtimeClassName: nvidia
+    nodeSelector:
+      feature.node.kubernetes.io/pci-10de.present: "true"
+    serviceMonitor:
+      enabled: false

vendor/infrastructure/nvidia/kustomization.yaml

@@ -0,0 +1,6 @@
+apiVersion: kustomize.config.k8s.io/v1beta1
+kind: Kustomization
+resources:
+  - namespace.yaml
+  - helm-releases.yaml
+  - network-policies.yaml

vendor/infrastructure/nvidia/namespace.yaml

@@ -0,0 +1,4 @@
+apiVersion: v1
+kind: Namespace
+metadata:
+  name: nvidia

vendor/infrastructure/nvidia/network-policies.yaml

@@ -0,0 +1,37 @@
+apiVersion: cilium.io/v2
+kind: CiliumClusterwideNetworkPolicy
+metadata:
+  name: host-egress-nvidia-dcgm-exporter
+spec:
+  nodeSelector: {}
+  egress:
+    - toEndpoints:
+        - matchLabels:
+            k8s:io.kubernetes.pod.namespace: nvidia
+            app.kubernetes.io/name: dcgm-exporter
+            app.kubernetes.io/instance: dcgm-exporter
+      toPorts:
+        - ports:
+            - port: "9400"
+              protocol: TCP
+---
+apiVersion: cilium.io/v2
+kind: CiliumNetworkPolicy
+metadata:
+  name: dcgm-exporter-ingress-vmagent
+  namespace: nvidia
+spec:
+  endpointSelector:
+    matchLabels:
+      app.kubernetes.io/name: dcgm-exporter
+      app.kubernetes.io/instance: dcgm-exporter
+  ingress:
+    - fromEndpoints:
+        - matchLabels:
+            k8s:io.kubernetes.pod.namespace: victoriametrics
+            app.kubernetes.io/name: vmagent
+            app.kubernetes.io/instance: main
+      toPorts:
+        - ports:
+            - port: "9400"
+              protocol: TCP