GPU VulnDB

Database/Container, Kubernetes & orchestration

RHACS: deploy-time policy enforcement bypassed by a crafted deployment config label

CVSS 8.5CVE-2026-10079Container, Kubernetes & orchestrationcurated

Impact

RHACS rewrites deployment identity metadata from the openshift.io/encoded-deployment-config label. A user who can create Deployments sets that label to "null", and ACS then treats the workload as having an empty UID, empty name, no labels and namespace "default". Deploy-time policy detection and enforcement visibility are bypassed, the workload is not persisted correctly in Central, and violation reporting and compliance correlation break for it. On a GPU cluster that relies on ACS admission policy to keep privileged pods, host mounts and unsigned images off accelerator nodes, this is a quiet way to land a workload that no gate inspected and that no report shows - the dashboard looks clean while the pod runs.

Who can reach it

Any authenticated Kubernetes or OpenShift user with permission to create Deployments in a namespace they can reach. No ACS console access and no node access needed.

What to do

Update Red Hat Advanced Cluster Security 4 to the fixed release in Red Hat's advisory and restart Central and the Sensors - a control-plane component restart, no node drain or reboot. The record names no standalone mitigation; in the interim, do not treat ACS deploy-time results as complete, and cross-check running workloads against the cluster API rather than against ACS inventory alone.

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.