A flaw was found in ml-metadata. The statically linked gRPC stack is outdated, vulnerable to known HTTP/2 denial‑of‑service vulnerabilities. An in‑cluster attacker with network access to the MLMD pod could exploit this by sending specially crafted HTTP/2 requests, causing the pod to crash and all pipeline runs in that namespace to fail. The weakness is a resource exhaustion flaw (CWE‑770) that permits denial of service.

This issue affects Red Hat OpenShift AI (RHOAI). The vulnerable component is the ml‑metadata service running within the MLMD pod, which listens on port 8080. No specific version range is provided, so all RHOAI installations that include ml‑metadata are potentially impacted until a patch or updated gRPC stack is deployed.

The CVSS score is 7.5, indicating moderate to high severity, and the vulnerability is currently not listed in the CISA KEV catalog and has no EPSS score available. An attacker consistent with the in‑cluster profile would need connectivity to the pod’s 8080 port, so enforcing network policies to limit which pods may contact the MLMD service mitigates the risk. If the restrictions are bypassed, the denial of service can propagate to every pipeline in the affected namespace, effectively halting business processes.