High [CVE-2026-94623] Denial of Service via NIXL multi-prompt assertion failure
This high-severity Red Hat Linux advisory covers CVE-2026-94623; related products: Red Hat AI Inference Server, Red Hat Enterprise Linux AI (RHEL AI) 3, Red Hat OpenShift AI (RHOAI).
Aggregated and source-linked by VulniPulse. Data sources, validation and limitations.
VulniPulse record published Record updated
Android app · Google Play
Monitor future Red Hat Linux CVEs from your phone.
Choose a whole vendor or a precise platform, then receive matching security advisories by phone notification, email, or both. Coverage follows 32 official vendor sources and 160+ reviewed platform categories.
Summary
Denial of Service via NIXL multi-prompt assertion failure. Red Hat rates this important (CVSS 7.5).
Weakness: CWE-617.
Affected products named by the advisory: Red Hat AI Inference Server; Red Hat Enterprise Linux AI (RHEL AI) 3; Red Hat OpenShift AI (RHOAI).
- 0.29.0
Official advisory · high-confidence parse· fetched 8 days ago·verify at source
Fixed versions
No fixed release is recorded yet. That does not prove no patch exists — confirm against the vendor advisory.
Official advisory · high-confidence parse· fetched 8 days ago·verify at source
Mitigation checklist
- Restrict access to the inference serving port to trusted internal networks, or run vLLM in standard standalone mode without disaggregated NIXL KV-cache transfer. To limit network reachability on host deployments using firewalld, restrict traffic to authorized client addresses: ``` firewall-cmd --permanent --zone=trusted --add-source=<TRUSTED_NETWORK_CIDR> firewall-cmd --permanent --zone=public --remove-port=8000/tcp firewall-cmd --reload ``` In Red Hat OpenShift AI environments, apply a Kubernetes NetworkPolicy to the inference namespace to block ingress traffic from untrusted pods and external routes. Caveats: Limiting network reachability blocks requests from untrusted external clients. Disabling disaggregated NIXL cache transfer requires reconfiguring workloads to run on standalone nodes, which may decrease distributed throughput for large-scale serving. Warning: Reloading firewall rules or modifying cluster deployment manifests can disrupt ongoing connections and will require restarting running serving pods.
Official advisory · high-confidence parse· fetched 8 days ago·verify at source
Discussion(0)
No comments yet. Share field notes, upgrade gotchas, or questions — verify against the vendor advisory before acting on community advice.
Sign in to join the discussion.