Merge branch 'cherry-pick_4080ddb808f88c1c01678f4746634812cdef11e0' i…

…nto 'release-23.9' Cherry-picks for 23.9.2 release See merge request nvidia/kubernetes/gpu-operator!1034
NVIDIA · Mar 5, 2024 · 8ea691d · 8ea691d
2 parents eb72305 + eb9f644
commit 8ea691d
Show file tree

Hide file tree

Showing 4 changed files with 16 additions and 10 deletions.
diff --git a/bundle/manifests/gpu-operator-certified.clusterserviceversion.yaml b/bundle/manifests/gpu-operator-certified.clusterserviceversion.yaml
@@ -194,11 +194,11 @@ spec:
     - name: container-toolkit-image
       image: nvcr.io/nvidia/k8s/container-toolkit@sha256:59a3875e7a37eb370385e654184efa3a1b193c9ea352165818496b19cbe14aa4
     - name: driver-image
-      image: nvcr.io/nvidia/driver@sha256:028cc6cf5944c0df4fbee8fe0f45654c09996ac13b5b37fb803d79ae2d37d378
-    - name: driver-image-525
-      image: nvcr.io/nvidia/driver@sha256:91f240e026be5e7cae1cc41e795b9a95b4beb85dd2b8f5a772ab2e1576a9e0e8
+      image: nvcr.io/nvidia/driver@sha256:7481a3697783dcdca9ae78e7b548a6900e86ea33ab49ec14f0ba55db2fdb1a2e
+    - name: driver-image-535
+      image: nvcr.io/nvidia/driver@sha256:abda1ac56371d55917b96ff330109980f468e133c9d5705da0ef87429f14ccd7
     - name: driver-image-470
-      image: nvcr.io/nvidia/driver@sha256:060c1005b09e4e3ba587c21104c22c5c2dceab690356e2097c5671233ad72a61
+      image: nvcr.io/nvidia/driver@sha256:56c79482582cdfbc58d3134e8672637c5bf05f328880f76898f526143d04c6af
     - name: device-plugin-image
       image: nvcr.io/nvidia/k8s-device-plugin@sha256:bb81ae8d034d1e1c6a342091a5f1dda5884dd43ab47e96624487a133d1edd4b3
     - name: gpu-feature-discovery-image
@@ -906,11 +906,11 @@ spec:
                   - name: "DEVICE_PLUGIN_IMAGE"
                     value: "nvcr.io/nvidia/k8s-device-plugin@sha256:bb81ae8d034d1e1c6a342091a5f1dda5884dd43ab47e96624487a133d1edd4b3"
                   - name: "DRIVER_IMAGE"
-                    value: "nvcr.io/nvidia/driver@sha256:028cc6cf5944c0df4fbee8fe0f45654c09996ac13b5b37fb803d79ae2d37d378"
-                  - name: "DRIVER_IMAGE-525"
-                    value: "nvcr.io/nvidia/driver@sha256:91f240e026be5e7cae1cc41e795b9a95b4beb85dd2b8f5a772ab2e1576a9e0e8"
+                    value: "nvcr.io/nvidia/driver@sha256:7481a3697783dcdca9ae78e7b548a6900e86ea33ab49ec14f0ba55db2fdb1a2e"
+                  - name: "DRIVER_IMAGE-535"
+                    value: "nvcr.io/nvidia/driver@sha256:abda1ac56371d55917b96ff330109980f468e133c9d5705da0ef87429f14ccd7"
                   - name: "DRIVER_IMAGE-470"
-                    value: "nvcr.io/nvidia/driver@sha256:060c1005b09e4e3ba587c21104c22c5c2dceab690356e2097c5671233ad72a61"
+                    value: "nvcr.io/nvidia/driver@sha256:56c79482582cdfbc58d3134e8672637c5bf05f328880f76898f526143d04c6af"
                   - name: "DRIVER_MANAGER_IMAGE"
                     value: "nvcr.io/nvidia/cloud-native/k8s-driver-manager@sha256:27c44f4720a4abf780217bd5e7903e4a008ebdbcf71238c4f106a0c22654776c"
                   - name: "MIG_MANAGER_IMAGE"

diff --git a/deployments/gpu-operator/templates/clusterpolicy.yaml b/deployments/gpu-operator/templates/clusterpolicy.yaml
@@ -91,6 +91,12 @@ spec:
       env: {{ toYaml .Values.validator.plugin.env | nindent 8 }}
       {{- end }}
     {{- end }}
+    {{- if .Values.validator.cuda }}
+    cuda:
+      {{- if .Values.validator.cuda.env }}
+      env: {{ toYaml .Values.validator.cuda.env | nindent 8 }}
+      {{- end }}
+    {{- end }}
     {{- if .Values.validator.driver }}
     driver:
       {{- if .Values.validator.driver.env }}

diff --git a/deployments/gpu-operator/values.yaml b/deployments/gpu-operator/values.yaml
@@ -135,7 +135,7 @@ driver:
   usePrecompiled: false
   repository: nvcr.io/nvidia
   image: driver
-  version: "535.129.03"
+  version: "550.54.14"
   imagePullPolicy: IfNotPresent
   imagePullSecrets: []
   startupProbe:

diff --git a/validator/main.go b/validator/main.go
@@ -250,7 +250,7 @@ func main() {
 		&cli.BoolFlag{
 			Name:        "with-workload",
 			Aliases:     []string{"l"},
-			Value:       false,
+			Value:       true,
 			Usage:       "indicates to validate with GPU workload",
 			Destination: &withWorkloadFlag,
 			EnvVars:     []string{"WITH_WORKLOAD"},