-with-ci-artifacts-dra e2e consistently failing since ~2026-06-02

[mboersma]

The pull-cluster-api-provider-azure-conformance-with-ci-artifacts-dra presubmit job has been failing consistently across unrelated PRs since around 2026-06-02.

Which test broke

A single spec fails while the other 98 of 99 conformance specs pass and the cluster itself comes up healthy:

[FAIL] [sig-node] [DRA] [FeatureGate:DRAExtendedResource] [Feature:DynamicResourceAllocation]
  must run pods with extended resource on dra nodes and device plugin nodes
  [KubeletMinVersion:1.35] [Serial]
  [FAILED] start pod1: Timed out after 300.000s.

This is an alpha feature-gated test (KubeletMinVersion:1.35), so the regression is almost certainly upstream (CI k8s build / DRA extended-resource feature), not in CAPZ — no CAPZ change touches this flavor.

When it broke

• Last green run: 2026-06-01
• First failure of the regression: 2026-06-02, failing consistently since (intermixed across multiple unrelated PRs).

Failure pattern on TestGrid: https://testgrid.k8s.io/sig-cluster-lifecycle-cluster-api-provider-azure#capz-pr-conformance-k8s-ci-dra-main

Example failed runs (different PRs, same failure)

• PR Let ASO v2.17.0 manage its own CRDs #6304, 2026-06-02: https://prow.k8s.io/view/gs/kubernetes-ci-logs/pr-logs/pull/kubernetes-sigs_cluster-api-provider-azure/6304/pull-cluster-api-provider-azure-conformance-with-ci-artifacts-dra/2061891096803282944
• PR [WIP] Default spot and custom-vnet e2e templates to Azure Linux 4 #6330, 2026-06-08: https://prow.k8s.io/view/gs/kubernetes-ci-logs/pr-logs/pull/kubernetes-sigs_cluster-api-provider-azure/6330/pull-cluster-api-provider-azure-conformance-with-ci-artifacts-dra/2064087176764723200

Since it's failing on unrelated PRs, it shouldn't block merges for changes that don't touch this area, but the job needs attention (likely a temporary skip of the DRAExtendedResource [Serial] test, or tracking the upstream fix).