NGINX Fixes HTTP/2 Flaw Allowing Unauthenticated Code Execution
$ grep -o 'CVE-[0-9-]*'
┌─ nginx-http2-flaw-code-execution.md :: hero
F5 has patched a decade-old vulnerability in NGINX, tracked as CVE-2026-123007 and dubbed "nginx2root", alongside four other flaws scored 9.2 out of 10 on CVSS 4.0. Present in NGINX Open Source since version 1.9.5 in September 2015, CVE-2026-123007 lets unauthenticated attackers execute code with NGINX worker permissions on servers with HTTP/2 enabled when ASLR is disabled or bypassed. Affected software includes NGINX Open Source 1.9.5 through 1.31.8 and NGINX Plus 37.x and R34 to R36. Fixed releases 1.31.9 and 1.30.7 are now available.
Key Facts
| Attribute | Value |
|---|---|
| Affected product and versions | NGINX Open Source 1.9.5 to 1.31.8 (including 1.31.7, 1.31.8, and 1.30.6); NGINX Plus 37.0.0 to 37.0.7.1, 37.1.0 to 37.2.0.1, and R34 to R36 |
| CVE IDs | CVE-2026-123007 ("nginx2root"), CVE-2026-123633, CVE-2026-123634, CVE-2026-121810, CVE-2026-113425 |
| CVSS | 9.2 out of 10 (CVSS 4.0, scored by F5 for all five flaws) |
| Exploited in the wild | No active attacks reported by F5, Wiz, or Rapid7 as of 18 November 2026; not listed in CISA's Known Exploited Vulnerabilities catalog |
| Fixed versions | NGINX Open Source 1.31.9 and 1.30.7; NGINX Plus 37.3.0.1 and 37.0.8.1 (CVE-2026-123633 fixed earlier in 37.2.0.1 and 37.0.7.1; no fix listed for R34 to R36) |
| Disclosure date | 18 November 2026 |
How Do the Five NGINX Flaws Work and Who Is Affected?
All five vulnerabilities can lead to code execution under NGINX worker permissions where ASLR is turned off or can be bypassed, provided a specific feature is enabled. CVE-2026-123007 affects every NGINX Open Source release from 1.9.5 through 1.31.8, as well as NGINX Plus 37.x and R34 to R36.
According to NGINX's security advisories, each of the five vulnerabilities is reachable only under a specific server setup:
- CVE-2026-123007: HTTP/2 turned on.
- CVE-2026-123633: HTTP/2 enabled with requests passed to a backend server while the worker is running low on worker_connections.
- CVE-2026-123634: A proxied server name constructed from variables, combined with a server-side include (SSI) or mirror subrequest to a location that looks up the same name.
- CVE-2026-121810: HTTP/3 turned on.
- CVE-2026-113425: NGINX configured as a mail proxy.
According to the Computer Incident Response Center Luxembourg (CIRCL), code executed via these flaws runs with the permissions of an NGINX worker process, meaning host root access requires a second privilege escalation step. In tests conducted by Xint, worker-level code gained root access on the host through a separate bug in the Linux kernel's SCTP code, even when NGINX was running inside a Docker container.
Technical requirements also vary by attack method. Rapid7 reported that its demonstration required about 128 GB of HTTP/2 traffic over two and a half minutes, meaning links slower than roughly 300 Mbit/s would exceed NGINX's default one-hour connection limit and fail. Rapid7's method also relies on a glibc memory-copy routine selected on x86-64 CPUs with avx512f and avx512vl features.
Executing code through CVE-2026-123007 grants the permissions of an NGINX worker process, and obtaining host root privileges requires a separate privilege escalation step such as a Linux kernel SCTP flaw.
Are These Vulnerabilities Being Exploited in the Wild?
No active exploitation of the NGINX vulnerabilities had been reported by F5, Wiz, or Rapid7 as of 18 November 2026, and CISA had not added CVE-2026-123007 to its Known Exploited Vulnerabilities catalog. However, proof-of-concept exploit code for CVE-2026-123007 has already been published on GitHub.
The authors of the GitHub proof-of-concept noted that it is suitable only for demonstrating the attack against a freshly started Docker container on the same machine. Even so, independent research teams confirmed that remote code execution is achievable under test conditions: Wiz bypassed ASLR and ran code on official NGINX 1.31.7 and 1.31.8 builds using HTTP/2 over TLS on 64-bit systems, while Xint reported executing code against test servers on AWS, GCP, and Azure, including over a public IP address, with only a reachable HTTP/2 port.
While no real-world attacks were recorded at the time of disclosure, researchers demonstrated working code execution against test servers on public cloud platforms and a proof-of-concept is publicly available on GitHub.
How Can Administrators Check Exposure and Spot Signs of Attack?
Administrators can confirm HTTP/2 exposure by reviewing NGINX configuration files for http2 directives and testing the origin server directly with openssl s_client using -alpn h2. To detect potential intrusion, CIRCL recommends monitoring for unexpected worker child processes, worker crashes, heavy HTTP/2 traffic with sparse logs, and unusual outbound connections.
HTTP/2 is active wherever a listen directive includes http2 or where http2 on is configured. Running openssl s_client with -alpn h2 returns ALPN protocol: h2 if the server accepts HTTP/2. Organisations using a content delivery network (CDN) must test the NGINX origin server directly rather than the CDN edge, because an origin that accepts only HTTP/1.1 cannot be exploited via HTTP/2.
When reviewing logs and system telemetry, CIRCL advises looking for NGINX workers spawning unexpected programmes, workers exiting abnormally or leaving core dumps, long-lived HTTP/2 connections carrying heavy incoming traffic with few logged requests, and unusual outbound connections or DNS lookups. None of these indicators proves compromise on its own.
A clean NGINX access log does not rule out exploitation because HTTP/2 control frames do not normally register as standard logged requests.
What Should You Do?
- Upgrade to F5's official patched releases: NGINX Open Source 1.31.9 (for 1.31.0 to 1.31.8) or 1.30.7 (for 1.9.5 to 1.30.6, including older branches such as 1.28), or pull the official 1.31.9 or 1.30.7 Docker images (including Alpine variants). For NGINX Plus, upgrade to 37.3.0.1 or 37.0.8.1; no fix was listed for R34 to R36.
- Perform a full NGINX restart or a live binary upgrade after updating, because a configuration reload alone leaves the old executable running and active connections attached to old worker processes.
- Check Ubuntu and Debian distribution advisories closely before assuming a host is patched or unpatched: as of 18 November 2026, Ubuntu (16.04 to 26.10) and Debian (bookworm, trixie, forky, and sid) had not shipped fixed packages, and distribution packages can include backported fixes without changing the version shown by nginx -v.
- Apply temporary mitigations where immediate upgrades are not possible: disable HTTP/2 (set http2 off; and remove http2 from listen directives on NGINX 1.25.1 and later, or remove http2 from listen on 1.9.5 through 1.25.0) to block CVE-2026-123007 and CVE-2026-123633, disable HTTP/3 where used for CVE-2026-121810, avoid SSI and mirror directives alongside variable-built proxied server names for CVE-2026-123634, and restrict mail proxy listeners to trusted networks for CVE-2026-113425.
- Block the SCTP kernel module from loading on Linux hosts that do not use SCTP by adding install sctp /bin/false to a .conf file in /etc/modprobe.d/, noting that this neither unloads an already loaded module nor disables SCTP built directly into the kernel.
- Review systems for CIRCL's indicators of suspicious worker activity; organisations in Malaysia can report suspected cybersecurity incidents to Cyber999 (MyCERT).
Common Questions
Does reloading NGINX apply the security fix after an upgrade?
No. A simple configuration reload leaves the old NGINX executable running, and active connections can remain attached to unpatched worker processes. After upgrading to fixed versions 1.31.9, 1.30.7, 37.3.0.1, or 37.0.8.1, administrators must restart NGINX fully or perform a live binary upgrade.
Are servers behind a CDN automatically safe from CVE-2026-123007?
Not necessarily. Organisations behind a CDN must test the NGINX origin server itself using openssl s_client with -alpn h2. An origin server that only accepts HTTP/1.1 cannot be attacked through HTTP/2, whereas an origin accepting HTTP/2 remains exposed if unpatched.
Does turning off HTTP/2 block all five newly patched NGINX flaws?
No. According to CIRCL, disabling HTTP/2 mitigates CVE-2026-123007 and CVE-2026-123633, but it does not block the remaining three vulnerabilities. Administrators must also disable HTTP/3 for CVE-2026-121810, apply vendor workarounds for CVE-2026-123634 and CVE-2026-113425, and upgrade as soon as possible.
Sources and Methodology
This dispatch is based solely on the reporting in The Hacker News published on 18 November 2026. No external documents, vendor advisories, or third-party trackers were independently reviewed or verified outside the provided source text.
$ ls -t news/ | head -5
$ ls ./tags
- #Vulnerabilities