GPU VulnDB

Database/Container, Kubernetes & orchestration

Kyverno: crafted forEach mutate policy crash-loops the background controller and blocks admission

CVSS 7.7CVE-2026-41485Container, Kubernetes & orchestrationcurated

Impact

An unchecked type assertion in the legacy engine's forEach mutation handler lets a crafted Policy or ClusterPolicy crash Kyverno's cluster-wide background controller into a persistent CrashLoopBackOff. The same bug makes the admission controller drop connections and block every operation on resources the policy matches. Depending on the webhook failure policy, that can stop pod creation for everyone on the cluster, including GPU job launches. The crash loop lasts until the policy is deleted. CEL-based policies are not affected.

Who can reach it

Authenticated user with permission to create a Kyverno Policy (namespaced) or ClusterPolicy. No other access is needed.

What to do

Upgrade Kyverno to 1.17.2 or 1.16.4, which rolls its controller pods. Nodes are not affected. If you are hit before upgrading, delete the offending policy to recover, and limit who can create Policy and ClusterPolicy objects.

References

Related entries

All Container, Kubernetes & orchestration entries

This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.