GPU VulnDB

Database/Control plane, storage & DevOps

OpenTelemetry JS Prometheus exporter: a malformed request URI crashes the whole Node.js process

CVE-2026-44902Control plane, storage & DevOpscurated

Impact

The exporter's metrics endpoint - by default listening on 0.0.0.0:9464 - parses the request URL with no error handling, so a single request with an invalid URI raises an uncaught TypeError and terminates the Node.js process. It is not just the metrics server that dies: the entire service hosting the exporter goes with it, which is why this rates a straight remote availability kill with no authentication. On a fleet, anything scraped by Prometheus is generally reachable from the cluster network, so any tenant pod or anyone on the monitoring path can kill Node-based control-plane services, gateways or serving sidecars repeatedly, and a supervisor that restarts them just gets killed again.

Who can reach it

Anyone with network reach to the exporter's metrics port (default 0.0.0.0:9464). No authentication is required - the endpoint has none.

What to do

Upgrade @opentelemetry/exporter-prometheus (and any bundling package such as sdk-node or auto-instrumentations-node) to 0.217.0 or later and restart each affected service. Until the rebuild lands, bind the exporter to localhost or restrict port 9464 with a NetworkPolicy or firewall rule so only the Prometheus scraper can reach it.

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.