chore(deps): update dependency dra-driver-nvidia-gpu to v25

[renovate]

This PR contains the following updates:

Package	Update	Change
dra-driver-nvidia-gpu	major	0.4.0-ubuntu24.04u2 → 25.12.0-ubuntu24.04u3
dra-driver-nvidia-gpu	major	0.4.0-ubuntu22.04u1 → 25.12.0-ubuntu22.04u3

---

Configuration

📅 Schedule: (UTC)

• Branch creation
  • At any time (no schedule defined)
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about these updates again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[aks-node-assistant]

AgentBaker Linux gate detective

Run: https://msazure.visualstudio.com/CloudNativeCompute/_build/results?buildId=168527801
Failed job/stage/task: $(System.Collections.Hashtable.job)
First failing step/test: $(System.Collections.Hashtable.first)

RCA: Known Ubuntu 22.04 HTTPSProxy PrivateDNS provisioning flake: vmssCSE returns VMExtensionProvisioningError with CSE ExitCode 50 and KubeletStartTime=n/a before kubelet starts. The failing scenario is unrelated to the GPU DRA driver update.

Confidence: High for the primary signature. Corroborated by timeline/status, focused failed logs, associated changes, and the flakiness wiki before publishing.

Strongest alternative: GPU DRA driver update caused provisioning failure; less likely because the failing scenarios are non-GPU Ubuntu 22.04 HTTPSProxy PrivateDNS and match prior watcher signature.

Recommended owner/action: E2E scenario/proxy/private-DNS owner: inspect CSE output/proxy/private DNS readiness and rerun after infra clears.

Wiki signature: $(System.Collections.Hashtable.sig) (source of truth)

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[aks-node-assistant]

AgentBaker Linux gate detective

Run: https://msazure.visualstudio.com/CloudNativeCompute/_build/results?buildId=169119692
Failed job/stage/task: e2e / Run AgentBaker E2E
Summary: E2E hit the known NetworkIsolated shared-cluster cleanup failure. abe2e-azure-networkisolated-v3-d6cc9 had a deleting/missing node resource group; cleanup failed with ResourceGroupDeletionBlocked / InUseNetworkSecurityGroupCannotBeDeleted because abe2e-networkisolated-securityGroup is still attached to the shared VNET subnet.

Likely cause/signature: stale NetworkIsolated shared cluster NSG in-use (networkisolated-shared-cluster-delete-blocked-inuse-nsg, repair #38506740). Confidence: High.
Alternative considered: PR #8733's GPU DRA component bump; less likely because failure is AKS shared-cluster deletion/NSG cleanup, not GPU provisioning.
Recommended owner/action: No PR action; E2E infra owner should resolve the stuck NSG/subnet association and harden cleanup/retry.
Wiki: https://dev.azure.com/msazure/09706533-03bf-4b43-9a9b-b49c75429646/_wiki/wikis/ed4a85e9-1085-4151-a39b-2753523eba2b?pagePath=%2FAKS%2FSIGs%20and%20Teams%2FAKS%20Components%2FSIG%3A%20Node%20Lifecycle%2FAI%20Agent%20Knowledge%2FAgentBaker%20Gate%20PR%20Pipeline%20Flakiness (networkisolated-shared-cluster-delete-blocked-inuse-nsg)

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.

[Copilot]

Copilot can't review bot-authored pull requests automatically. A user with Copilot access can request a review manually.

[aks-node-assistant]

Failed gate run

Run: https://msazure.visualstudio.com/CloudNativeCompute/_build/results?buildId=170310042
Failed job/stage/task: e2e / Run AgentBaker E2E

Detective summary (three-level RCA)

TL;DR: Infrastructure infra-flake. 40 kubenet scenarios timed out at 600s node-ready wait. Not caused by this PR. Confidence: high (≥95%).

L1 – Surface failure
waiting for node X to be ready failed (600.0s) across ~40 scenarios (including ubuntu2204pmcinstall, azurelinux3osguard, ubuntu2404npdbasic, etc.); ##[error] Script failed with exit code: 1. All failures in kubenet default sub-tests.

L2 – Corroboration
Node condition at failure: NetworkUnavailable=True, reason=NodeInitialization, message=Waiting for cloud routes; kubelet: cni plugin not initialized. Azure CNI scenarios all passed. Simultaneous failure with builds 170309966 (PR #8765) and 170309991 (PR #8766) in the same 3-minute window (01:13–01:16Z) confirms shared infra event. PMC install scenarios fail for the same reason, not due to any PMC install change.

L3 – Root-cause challenge

• ❌ PR-caused regression: ruled out — three unrelated PRs fail identically and simultaneously.
• ❌ PMC install issue: ruled out — PMC scenarios show same Waiting for cloud routes node condition; not a package install failure.
• ✅ E2E infra: kubenet CCM route-table provisioning failure — CCM could not update Azure route table for pod-CIDR routes; all kubenet nodes blocked in NodeInitialization.

Secondary signature: wireserver-nonhost-debug-pod-missing — wireserver validation failed with failed to get non host debug pod for wireserver validation; co-effect of shared cluster instability.

Recommended action: Re-run; no PR change required.

Wiki: e2e-kubenet-cloud-routes-waiting-cni-uninit · wireserver-nonhost-debug-pod-missing

[Copilot]

Copilot can't review bot-authored pull requests automatically. A user with Copilot access can request a review manually.

[Copilot]

Copilot can't review bot-authored pull requests automatically. A user with Copilot access can request a review manually.

[Copilot]

Copilot can't review bot-authored pull requests automatically. A user with Copilot access can request a review manually.

[Copilot]

Copilot can't review bot-authored pull requests automatically. A user with Copilot access can request a review manually.

[Copilot]

Copilot can't review bot-authored pull requests automatically. A user with Copilot access can request a review manually.

[Copilot]

Copilot can't review bot-authored pull requests automatically. A user with Copilot access can request a review manually.