GPU VulnDB

Database/Control plane, storage & DevOps

openstack-network-exporter: missing HTTP header timeouts let an unauthenticated attacker exhaust file descriptors

CVSS 5.3CVE-2026-83536Control plane, storage & DevOpscurated

Impact

The exporter's HTTP server is configured without ReadHeaderTimeout or the other basic timeouts, so an unauthenticated attacker can open many connections and trickle header bytes until file descriptors run out - a slowloris. The exporter then stops answering Prometheus scrapes, which means the network metrics and the alerts built on them go silent for that host while everything else keeps running. That is a monitoring blind spot rather than a compromise: nothing is disclosed or modified, and the vendor rates availability impact low. It is worth closing because the exporter normally listens on a management or metrics network where anyone with a foothold can reach it unauthenticated, and silent monitoring is how a real fault goes unnoticed.

Who can reach it

Anyone who can open TCP connections to the exporter's metrics port - in practice anyone on the management or metrics network, or a workload that can route to it. No authentication required.

What to do

Apply the Red Hat OpenStack Platform 18.0 update for openstack-network-exporter when it ships and restart the exporter; the record does not name a fixed version yet. In the meantime, bind the exporter to the metrics network only and restrict its port to the Prometheus scrapers with a host firewall rule, or front it with a proxy that enforces header timeouts. Either way this is a service restart, not node work.

References

Related entries

All Control plane, storage & DevOps 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.