chore(recipes): bump gpu-operator v26.7.0, DRA 0.5.0; hold driver 580

.settings.yaml

@@ -202,7 +202,7 @@ testing:
 
   # GPU CI runtime pins consumed through .github/actions/load-versions.
   # renovate: datasource=helm depName=gpu-operator depType=testing
-  gpu_operator_chart_version: 'v26.3.3'
+  gpu_operator_chart_version: 'v26.7.0'
   # Base image for the CI smoke-test snapshot-agent image. The agent binary is
   # static Go and detects GPUs driver-free via NFD/PCI, so it no longer needs a
   # CUDA base / nvidia-smi. Matches the NVIDIA static distroless base used for

api/aicr/v1/server.yaml

@@ -650,7 +650,7 @@ paths:
                   value:
                     version: "580.105.08"
                     enabled: true
-                    useOpenKernelModules: true
+                    kernelModuleType: auto
         "400":
           description: >
             Invalid request. Common causes: (1) no criteria provided — all

demos/dynamic.md

@@ -98,7 +98,7 @@ cat ./bundle/010-gpu-operator/cluster-values.yaml
 # # Generated by AICR
 # ---
 # driver:
-#   version: 580.173.02
+#   version: 595.91.07
 ```
 
 And `install.sh` applies it after the baked values, so it wins:
@@ -116,12 +116,12 @@ Three quick checks, the same ones the chainsaw suite enforces:
 
 ```shell
 # a) gone from the baked values.yaml (no match)
-grep -c 'version: 580' ./bundle/010-gpu-operator/values.yaml
+grep -c 'version: 595' ./bundle/010-gpu-operator/values.yaml
 # 0
 
 # b) present in cluster-values.yaml
 grep version ./bundle/010-gpu-operator/cluster-values.yaml
-#   version: 580.173.02
+#   version: 595.91.07
 
 # c) install.sh references cluster-values.yaml
 grep -q cluster-values.yaml ./bundle/010-gpu-operator/install.sh && echo wired
@@ -193,7 +193,7 @@ rm -rf ./b-oci
 oras pull "$REF" -o ./b-oci
 (cd ./b-oci && aicr verify .)
 cat ./b-oci/010-gpu-operator/cluster-values.yaml
-#   version: 580.173.02
+#   version: 595.91.07
 ```
 
 > ttl.sh is public; anything pushed is world-readable until it expires. Fine for a throwaway demo bundle, not for anything sensitive — push to your own registry by pointing `spec.bundle.output.target` (or `--output`) at it, e.g. `oci://registry.example.com/aicr-bundle:tag`. Transport is separate: `spec.bundle.registry.plainHTTP: true` only switches HTTP instead of HTTPS (enable it only when the registry is intentionally served over plain HTTP, such as a local test registry); leave it off for any HTTPS registry, and use `spec.bundle.registry.insecureTLS: true` to skip TLS verification against a self-signed HTTPS registry.

demos/query.md

@@ -38,7 +38,7 @@ aicr query \
   --selector components.gpu-operator.values.driver.version
 ```
 
-> `580.173.02`
+> `595.91.07`
 
 Subtree — full driver block:
 
@@ -53,8 +53,8 @@ enabled: true
 maxParallelUpgrades: 5
 rdma:
   enabled: false
-useOpenKernelModules: true
-version: 580.173.02
+kernelModuleType: auto
+version: 595.91.07
 ```
 
 ## Differentiation: Same Selector, Different Criteria

docs/integrator/aks-gpu-setup.md

@@ -376,7 +376,7 @@ required so the device plugin's volume-mounts allocation strategy still works
 for legitimately allocated pods, but it leaves the **volume-mounts
 device-request path open** (a pod that declares a `/dev/null`-backed mount whose
 destination is under `/var/run/nvidia-container-devices` can still select
-devices — the pinned toolkit v1.19.1 accepts the volume-mount device request
+devices — the pinned toolkit v1.20.0 accepts the volume-mount device request
 only when the mount *source* is `/dev/null`). This is the same posture as
 GPU-Operator-managed mode — not a regression — but it means full multi-tenant
 isolation additionally requires an admission policy restricting a

docs/integrator/automation.md

@@ -231,7 +231,7 @@ spec:
     # Helm chart from upstream
     - repoURL: https://helm.ngc.nvidia.com/nvidia
       chart: gpu-operator
-      targetRevision: v26.3.3
+      targetRevision: v26.7.0
       helm:
         valueFiles:
           # Values live under the numbered bundle dir (NNN-<component>/)

docs/integrator/data-flow.md

@@ -634,7 +634,7 @@ spec:
   sources:
     # Helm chart from upstream
     - repoURL: https://helm.ngc.nvidia.com/nvidia
-      targetRevision: v26.3.3
+      targetRevision: v26.7.0
       chart: gpu-operator
       helm:
         valueFiles:

docs/integrator/recipe-development.md

@@ -888,7 +888,7 @@ spec:
 # recipes/registry.yaml
 - name: gpu-operator
   helm:
-    defaultVersion: v26.3.3  # Changed from v26.3.2
+    defaultVersion: v26.7.0  # Changed from v26.3.3
 ```
 
 **Adding components:**

docs/user/air-gap-mirror.md

@@ -112,7 +112,7 @@ metadata:
   name: aicr-images
 spec:
   images:
-    - name: nvcr.io/nvidia/gpu-operator:v26.3.3
+    - name: nvcr.io/nvidia/gpu-operator:v26.7.0
     - name: registry.k8s.io/nfd/node-feature-discovery:v0.19.0
     # ...
 ---
@@ -124,7 +124,7 @@ spec:
   charts:
     - name: gpu-operator
       repoURL: oci://ghcr.io/nvidia
-      version: v26.3.3
+      version: v26.7.0
     # ...
 ```
 
@@ -187,13 +187,13 @@ components:
   - name: aicr-images
     required: true
     images:
-      - nvcr.io/nvidia/gpu-operator:v26.3.3
+      - nvcr.io/nvidia/gpu-operator:v26.7.0
       - registry.k8s.io/nfd/node-feature-discovery:v0.19.0
       # ...
     charts:
       - name: gpu-operator
         url: oci://ghcr.io/nvidia/gpu-operator
-        version: v26.3.3
+        version: v26.7.0
         namespace: gpu-operator
       # ...
 ```

docs/user/api-reference.md

@@ -737,7 +737,7 @@ curl -X POST "http://localhost:8080/v1/bundle" \
     "apiVersion": "aicr.run/v1alpha2",
     "kind": "RecipeResult",
     "componentRefs": [
-      {"name": "gpu-operator", "type": "Helm", "chart": "gpu-operator", "source": "https://helm.ngc.nvidia.com/nvidia", "version": "v26.3.3", "namespace": "gpu-operator", "valuesFile": "components/gpu-operator/values.yaml"},
+      {"name": "gpu-operator", "type": "Helm", "chart": "gpu-operator", "source": "https://helm.ngc.nvidia.com/nvidia", "version": "v26.7.0", "namespace": "gpu-operator", "valuesFile": "components/gpu-operator/values.yaml"},
       {"name": "network-operator", "type": "Helm", "chart": "network-operator", "source": "https://helm.ngc.nvidia.com/nvidia", "version": "26.1.1", "namespace": "nvidia-network-operator", "valuesFile": "components/network-operator/values.yaml"}
     ],
     "deploymentOrder": ["gpu-operator", "network-operator"]

docs/user/cli-reference.md

@@ -1988,7 +1988,7 @@ The `--vendor-charts` flag pulls upstream Helm chart bytes into the bundle at bu
 my-bundle/
   001-gpu-operator/
     Chart.yaml                     # wrapper, declares the vendored subchart
-    charts/gpu-operator-v26.3.3.tgz # vendored upstream tarball
+    charts/gpu-operator-v26.7.0.tgz # vendored upstream tarball
     values.yaml                    # values nested under the subchart name
     cluster-values.yaml            # dynamic values, also nested
     install.sh                     # helm upgrade --install <name> ./<dir> ...
@@ -2017,10 +2017,10 @@ kind: BundleProvenance
 vendoredCharts:
   - name: gpu-operator
     chart: gpu-operator
-    version: v26.3.3
+    version: v26.7.0
     repository: https://helm.ngc.nvidia.com/nvidia
     sha256: abc123...
-    tarballName: gpu-operator-v26.3.3.tgz
+    tarballName: gpu-operator-v26.7.0.tgz
     pullerVersion: helm-cli v3.20.2
 ```
 

docs/user/container-images.md

@@ -42,7 +42,7 @@ _Rendering fidelity:_ `catalog-parity: charts are rendered with the shared recip
 | gatekeeper | helm | gatekeeper/gatekeeper | 3.22.2 | 3 |
 | gcp-driver-installer | manifest | — | — | 3 |
 | gke-nccl-tcpxo | manifest | — | — | 4 |
-| gpu-operator | helm | nvidia/gpu-operator | v26.3.3 | 15 |
+| gpu-operator | helm | nvidia/gpu-operator | v26.7.0 | 15 |
 | gpu-operator-ocp | manifest | — | — | 0 |
 | gpu-operator-ocp-olm | manifest | — | — | 0 |
 | grove | helm | grove-charts | v0.1.0-alpha.8 | 1 |
@@ -64,8 +64,8 @@ _Rendering fidelity:_ `catalog-parity: charts are rendered with the shared recip
 | nfd-ocp-olm | manifest | — | — | 0 |
 | nodewright-customizations | manifest | — | — | 5 |
 | nodewright-operator | helm | nodewright | v0.17.1 | 3 |
-| nvidia-dra-driver-gpu | helm | dra-driver-nvidia-gpu | 0.4.1 | 1 |
-| nvidia-dra-driver-gpu-ocp | helm | dra-driver-nvidia-gpu | 0.4.1 | 1 |
+| nvidia-dra-driver-gpu | helm | dra-driver-nvidia-gpu | 0.5.0 | 1 |
+| nvidia-dra-driver-gpu-ocp | helm | dra-driver-nvidia-gpu | 0.5.0 | 1 |
 | nvsentinel | helm | nvsentinel | v1.20.0 | 6 |
 | prometheus-adapter | helm | prometheus-community/prometheus-adapter | 5.3.0 | 1 |
 | prometheus-adapter-ocp | helm | prometheus-community/prometheus-adapter | 5.3.0 | 1 |
@@ -151,20 +151,20 @@ _No images extracted._
 ### gpu-operator
 
 - `docker.io/library/busybox:1.38.0@sha256:dc2d74b28e4cf8984fa52af1f39bc7c3d9c73760b41a74d629f5d11b1ab28616`
-- `nvcr.io/nvidia/cloud-native/dcgm:4.5.2-1-ubuntu22.04`
-- `nvcr.io/nvidia/cloud-native/gdrdrv:v2.5.2`
-- `nvcr.io/nvidia/cloud-native/k8s-cc-manager:v0.4.0`
-- `nvcr.io/nvidia/cloud-native/k8s-driver-manager:v0.11.0`
-- `nvcr.io/nvidia/cloud-native/k8s-mig-manager:v0.14.2`
-- `nvcr.io/nvidia/cloud-native/nvidia-fs:2.27.3`
-- `nvcr.io/nvidia/cloud-native/nvidia-sandbox-device-plugin:v0.0.3`
-- `nvcr.io/nvidia/cloud-native/vgpu-device-manager:v0.4.2`
+- `nvcr.io/nvidia/cloud-native/dcgm:4.6.0-1-ubuntu24.04`
+- `nvcr.io/nvidia/cloud-native/gdrdrv:v2.6`
+- `nvcr.io/nvidia/cloud-native/k8s-cc-manager:v0.4.3`
+- `nvcr.io/nvidia/cloud-native/k8s-driver-manager:v0.12.0`
+- `nvcr.io/nvidia/cloud-native/k8s-mig-manager:v0.15.0`
+- `nvcr.io/nvidia/cloud-native/nvidia-fs:2.29.4`
+- `nvcr.io/nvidia/cloud-native/nvidia-sandbox-device-plugin:v0.0.5`
+- `nvcr.io/nvidia/cloud-native/vgpu-device-manager:v0.5.0`
 - `nvcr.io/nvidia/driver:580.173.02`
-- `nvcr.io/nvidia/gpu-operator:v26.3.3`
-- `nvcr.io/nvidia/k8s-device-plugin:v0.19.3`
-- `nvcr.io/nvidia/k8s/container-toolkit:v1.19.1`
-- `nvcr.io/nvidia/k8s/dcgm-exporter:4.5.3-4.8.2-distroless`
-- `nvcr.io/nvidia/kubevirt-gpu-device-plugin:v1.5.0`
+- `nvcr.io/nvidia/gpu-operator:v26.7.0`
+- `nvcr.io/nvidia/k8s-device-plugin:v0.20.0`
+- `nvcr.io/nvidia/k8s/container-toolkit:v1.20.0`
+- `nvcr.io/nvidia/k8s/dcgm-exporter:4.6.0-4.8.3-distroless`
+- `nvcr.io/nvidia/kubevirt-gpu-device-plugin:v1.6.0`
 
 ### gpu-operator-ocp
 
@@ -281,11 +281,11 @@ _No images extracted._
 
 ### nvidia-dra-driver-gpu
 
-- `registry.k8s.io/dra-driver-nvidia/dra-driver-nvidia-gpu:v0.4.1`
+- `registry.k8s.io/dra-driver-nvidia/dra-driver-nvidia-gpu:v0.5.0`
 
 ### nvidia-dra-driver-gpu-ocp
 
-- `registry.k8s.io/dra-driver-nvidia/dra-driver-nvidia-gpu:v0.4.1`
+- `registry.k8s.io/dra-driver-nvidia/dra-driver-nvidia-gpu:v0.5.0`
 
 ### nvsentinel
 
@@ -450,7 +450,7 @@ Y = artifact attached, - = artifact absent, ? = could not probe.
 
   Image                                                           Sig  SBOM  Prov  Notes
   --------------------------------------------------------------  ---  ----  ----  -----
-  registry.k8s.io/dra-driver-nvidia/dra-driver-nvidia-gpu:v0.4.1  Y    -     -
+  registry.k8s.io/dra-driver-nvidia/dra-driver-nvidia-gpu:v0.5.0  Y    -     -
 
 Summary: 1/1 signed · 0/1 SBOM · 0/1 provenance
 ```

examples/recipes/aks-training.yaml

@@ -52,7 +52,7 @@ componentRefs:
     chart: gpu-operator
     type: Helm
     source: https://helm.ngc.nvidia.com/nvidia
-    version: v26.3.3
+    version: v26.7.0
     valuesFile: components/gpu-operator/values-aks-training.yaml
     dependencyRefs:
       - nfd
@@ -151,7 +151,7 @@ componentRefs:
     chart: dra-driver-nvidia-gpu
     type: Helm
     source: oci://registry.k8s.io/dra-driver-nvidia/charts
-    version: 0.4.1
+    version: 0.5.0
     valuesFile: components/nvidia-dra-driver-gpu/values.yaml
     overrides:
       controller:

examples/recipes/eks-gb200-ubuntu-training-with-validation.yaml

@@ -61,9 +61,9 @@ validation:
   deployment:
     constraints:
       - name: gpu-operator.version
-        value: "== v26.3.3"
+        value: "== v26.7.0"
         severity: warning
-        remediation: "Update GPU Operator to v26.3.3"
+        remediation: "Update GPU Operator to v26.7.0"
     checks:
       - expected-resources
 
@@ -100,7 +100,7 @@ componentRefs:
     chart: gpu-operator
     type: Helm
     source: https://helm.ngc.nvidia.com/nvidia
-    version: v26.3.3
+    version: v26.7.0
     valuesFile: components/gpu-operator/values-eks-training.yaml
     expectedResources:
       - kind: Deployment
@@ -129,7 +129,7 @@ componentRefs:
     chart: dra-driver-nvidia-gpu
     type: Helm
     source: oci://registry.k8s.io/dra-driver-nvidia/charts
-    version: 0.4.1
+    version: 0.5.0
     valuesFile: components/nvidia-dra-driver-gpu/values.yaml
     dependencyRefs:
       - gpu-operator

examples/recipes/eks-training.yaml

@@ -41,7 +41,7 @@ componentRefs:
     chart: gpu-operator
     type: Helm
     source: https://helm.ngc.nvidia.com/nvidia
-    version: v26.3.3
+    version: v26.7.0
     valuesFile: components/gpu-operator/values-eks-training.yaml
     dependencyRefs:
       - cert-manager