Summary

Fix the AWS UAT config so the Prometheus PVC binds to the cluster's default StorageClass instead of a hardcoded gp3 name that doesn't exist on an aicr-bundle-deployed EKS cluster.

Motivation / Context

On the AWS UAT cluster, deployment readiness checks blocked indefinitely: the Prometheus StatefulSet was never created. The prometheus-operator was healthy but refused to reconcile with:

ReconciliationFailed: storage class "gp3" does not exist
→ StatefulSetNotFound (DESIRED 1, READY 0)

spec.bundle.scheduling.storageClass was pinned to a literal gp3. An aicr-bundle-deployed EKS cluster has no StorageClass by that name — the aws-ebs-csi-driver component provisions a gp3-backed default named ebs-csi-default-sc (annotated is-default-class=true). A class literally named gp3 only exists on clusters using the AWS managed EBS CSI addon. The stale comment ("Matches the UAT cluster") was incorrect.

Fixes: N/A
Related: N/A

Type of Change

• Bug fix (non-breaking change that fixes an issue)

Component(s) Affected

• Other: tests/uat (AWS UAT test config)

Implementation Notes

Set storageClass: "". The injection path only writes a value when non-empty (pkg/cli/bundle.go → pkg/bundler/bundler.go), so an empty value is a true no-op. The EKS overlay (recipes/overlays/eks.yaml) already defines the Prometheus volumeClaimTemplate (50Gi, RWO, emptyDir: null) without a storageClassName, so the PVC binds to the cluster default SC. Persistent storage is preserved (not dropped to emptyDir) and the config no longer depends on how EBS CSI was installed.

GCP UAT config is unchanged — GKE genuinely ships a class named premium-rwo.

Testing

# Verified on the live AWS UAT cluster: created the missing default-equivalent SC,
# operator reconciled, prometheus-kube-prometheus-prometheus StatefulSet 1/1 Running,
# PVC bound. The config change makes this durable across cluster recreates.

Risk Assessment

• Low — Isolated change to a single UAT test config value, easy to revert.

Rollout notes: N/A — test harness config only; no production code path.

Checklist

• Tests pass locally (make test with -race)
• Linter passes (make lint)
• I did not skip/disable tests to make CI green
• I added/updated tests for new functionality
• I updated docs if user-facing behavior changed
• Changes follow existing patterns in the codebase
• Commits are cryptographically signed (git commit -S)