Trusence Every claim has a source
Last updated 21 November 2026 Search Türkçe
Security

Decade-old NGINX HTTP/2 flaw lets unauthenticated attackers run code on servers

Operators running NGINX with HTTP/2 enabled should patch to 1.31.9 or 1.30.7, or disable HTTP/2 until they can.

A cracked brick in a stone wall beside a brass key left in a keyhole, lit from the side on a dark background.
AI-generated illustration

A decade-old memory flaw tracked as CVE-2026-123007, nicknamed nginx2root, affects NGINX HTTP/2 handling in every open-source release from 1.9.5 onward, letting an unauthenticated remote party run code in a worker process when HTTP/2 is enabled. F5 shipped fixes in 1.31.9 and 1.30.7, plus 37.3.0.1 and 37.0.8.1 for NGINX Plus, and the same releases cover four companion flaws that carry a 9.2 CVSS 4.0 score. Researchers at Wiz and Xint showed the flaw can be exploited and, with a separate Linux SCTP kernel bug, used to reach root on the host even from inside a Docker container. Rapid7's public proof of concept needs roughly 128 GB of HTTP/2 traffic, so practical exposure is narrower than version counts suggest. No exploitation in the wild has been reported, and disabling HTTP/2 blocks the two HTTP/2 flaws until patching is done.

Why it matters

The flaw sits in every open-source NGINX release from 1.9.5 onward, so any server running those versions with HTTP/2 enabled is potentially reachable by an unauthenticated remote party. Chained with a separate Linux kernel SCTP bug, it can also reach root on the host from inside a Docker container, which widens the damage beyond the web server itself. Practical exposure is narrower than the version count suggests, since Rapid7's public proof of concept needs about 128 GB of HTTP/2 data and exploitation requires ASLR to be disabled or bypassed. No attacks in the wild have been reported so far.

Sources · 3