Feature: New GPU operator version v24.6.2

swan-cern/Chart.lock

@@ -7,6 +7,6 @@ dependencies:
   version: 0.4.3
 - name: gpu-operator
   repository: oci://registry.cern.ch/kubernetes/charts
-  version: v22.9.1
-digest: sha256:4600d0fa64afa906aaa7570a138dd796a87b20edd18cb96829f0d8e08372c4f3
-generated: "2024-10-01T09:39:24.176258854Z"
+  version: v24.6.2
+digest: sha256:02b128aa07fd796252255ac4f09708fef9b99aabd0d01d4f2c29d28a2e1aca55
+generated: "2024-10-16T10:39:56.250436142+02:00"

swan-cern/Chart.yaml

@@ -16,6 +16,6 @@ dependencies:
     version: 0.4.3
     repository: https://fluent.github.io/helm-charts
   - name: gpu-operator
-    version: v22.9.1
+    version: v24.6.2
     repository: oci://registry.cern.ch/kubernetes/charts
     condition: gpu-operator.enabled

swan-cern/templates/gpu/nvidia-time-slicing-config.yaml → swan-cern/templates/gpu/nvidia-device-plugin-config.yaml

@@ -2,9 +2,12 @@
 apiVersion: v1
 kind: ConfigMap
 metadata:
-  name: nvidia-time-slicing-config
+  name: nvidia-device-plugin-config
   namespace: {{ .Release.Namespace }}
 data:
+  default: |-
+    version: v1
+    flags: {}
   slice-4: |-
     version: v1
     sharing:
@@ -24,4 +27,3 @@ data:
         - name: nvidia.com/gpu
           replicas: 10
 {{- end }}
-

swan-cern/values.yaml

@@ -274,26 +274,27 @@ gpu-operator:
   enabled: true
   validator:
     repository: registry.cern.ch/kubernetes
-    version: v22.9.1
+    version: v24.6.2
   operator:
     repository: registry.cern.ch/kubernetes
-    version: v22.9.1
+    version: v24.6.2
     defaultRuntime: containerd
     initContainer:
       repository: registry.cern.ch/kubernetes
       version: 11.4.2-base-ubi8
   driver:
     repository: registry.cern.ch/kubernetes
     image: nvidia-gpu-driver
-    version: "v550.54.15-6.6.13-200.fc39.x86_64"
+    version: "v550.54.15-6.8.4-200.fc39.x86_64"
     imagePullPolicy: Always
     manager:
       repository: registry.cern.ch/kubernetes
       version: v0.6.0
     licensingConfig:
       configMapName: "nvidia-grid-license"
-    kernelModuleConfig:
-      name: kernel-module-params
+      nlsEnabled: false
+    nfd:
+      enabled: false
   vgpuManager:
     repository: registry.cern.ch
     driverManager:
@@ -305,12 +306,15 @@ gpu-operator:
     driverManager:
       repository: registry.cern.ch
       version: v0.5.1
+  gdrcopy:
+    repository: registry.cern.ch/kubernetes
+    version: v2.4.1-1
   vgpuDeviceManager:
     repository: registry.cern.ch
     version: v0.2.0
   toolkit:
     repository: registry.cern.ch/kubernetes
-    version: v1.11.0
+    version: v1.16.2-ubuntu20.04
   devicePlugin:
     repository: registry.cern.ch/kubernetes
     version: v0.13.0
@@ -324,7 +328,7 @@ gpu-operator:
       name: nvidia-dcgm-exporter-metrics
   gfd:
     repository: registry.cern.ch/kubernetes
-    version: v0.7.0
+    version: v0.16.2-ubi8
   mig:
     strategy: mixed
   migManager:
@@ -334,17 +338,19 @@ gpu-operator:
       name: nvidia-mig-config
   nodeStatusExporter:
     repository: registry.cern.ch
-    version: v22.9.1
+    version: v24.6.2
+  kataManager:
+    repository: registry.cern.ch/kubernetes
+    version: v0.2.1
   sandboxDevicePlugin:
     repository: registry.cern.ch/kubernetes
     version: v1.2.1
+  ccManager:
+    repository: registry.cern.ch/kubernetes
+    version: v0.1.1
   devicePlugin:
     repository: registry.cern.ch/kubernetes
     version: v0.12.2-ubi8
     config:
-      name: nvidia-time-slicing-config
-  node-feature-discovery:
-    image:
-      repository: registry.cern.ch/kubernetes/node-feature-discovery
-      tag: v0.10.1
-
+      name: nvidia-device-plugin-config
+      default: "default"