Critical NGINX HTTP/2 flaw allows remote code execution after ten years
A decade-old vulnerability in NGINX's HTTP/2 support, CVE-2026-123007, enables unauthenticated attackers to execute code on servers. Fixed versions 1.31.9 and 1.30.7 are now available.
F5 has released emergency patches for NGINX to address a critical security flaw that has existed in the software for ten years. The vulnerability, identified as CVE-2026-123007 and nicknamed "nginx2root," allows unauthenticated attackers to execute arbitrary code on servers with HTTP/2 enabled. Fixed versions 1.31.9 and 1.30.7 were made available on November 18, 2026, urging immediate action for infrastructure teams.
What happened
The flaw resides in NGINX’s handling of HTTP/2 traffic and has been present in every open-source release since version 1.9.5, which launched in September 2015. Security researchers from Wiz and Xint demonstrated that the vulnerability can be exploited without login credentials. In their tests, they successfully bypassed Address Space Layout Randomization (ASLR), a common memory protection technique, to run code on affected systems. Xint researchers noted that access to a reachable HTTP/2 port was sufficient to trigger the exploit.
While the initial code execution occurs with the permissions of the NGINX worker process, escalating to root access requires a secondary vulnerability. CIRCL, the Computer Incident Response Center Luxembourg, reported that attackers could combine this flaw with a separate bug in the Linux kernel’s SCTP code to gain full host control, even within Docker containers. Despite the severity, no active exploitation has been reported by F5, Wiz, or Rapid7, and the issue is not yet listed in CISA’s Known Exploited Vulnerabilities catalog.
In addition to CVE-2026-123007, the new releases address four other high-severity flaws, all scoring 9.2 out of 10 on the CVSS 4.0 scale. These additional vulnerabilities also permit code execution under specific conditions, such as when HTTP/3 is enabled or when NGINX is used as a mail proxy. Every open-source version from 1.9.5 through 1.31.8 is affected, including recent security releases from October 2026.
How it works
The primary vulnerability exploits how NGINX processes HTTP/2 frames. When HTTP/2 is enabled via the listen directive or the http2 on setting, the server becomes susceptible to memory corruption attacks. Researchers found that by sending specially crafted HTTP/2 traffic, they could manipulate memory structures in the NGINX worker process. This manipulation allows them to overwrite return addresses or function pointers, redirecting execution flow to malicious code.
Successful exploitation often depends on specific system configurations. For instance, the attack demonstrated by Rapid7 requires approximately 128 GB of HTTP/2 traffic and relies on glibc memory-copy routines available on x86-64 CPUs with AVX-512 features. On slower network links, the attack may fail due to connection timeouts. However, Xint’s proof-of-concept worked over public IPs on major cloud providers like AWS, GCP, and Azure, indicating that network speed is not always a barrier.
Key details
- Affected Versions: All NGINX Open Source releases from 1.9.5 to 1.31.8, and NGINX Plus releases R34 to R36 and certain 37.x builds.
- Fixed Versions: NGINX Open Source 1.31.9 and 1.30.7; NGINX Plus 37.3.0.1 and 37.0.8.1.
- Prerequisites: HTTP/2 must be enabled; some variants require HTTP/3, mail proxy usage, or specific backend configurations.
- Impact: Remote code execution with worker process permissions; potential for root escalation via kernel bugs.
- Mitigation: Upgrade immediately or disable HTTP/2 and HTTP/3 if upgrading is not possible.
- Detection: Monitor for abnormal worker exits, core dumps, and unusual outbound DNS lookups.
Why it matters
For software engineers and DevOps teams, this vulnerability highlights the risks of long-standing dependencies in critical infrastructure. NGINX powers a significant portion of the web, and many organizations may be running vulnerable versions without realizing it, especially if they rely on distribution packages that lag behind upstream releases. Ubuntu and Debian had not shipped fixed packages at the time of the announcement, meaning system administrators must verify package versions carefully rather than relying solely on version numbers reported by nginx -v.
The complexity of the exploit does not diminish its threat. While some attack vectors require high bandwidth or specific CPU features, others are more accessible. The availability of proof-of-concept code on GitHub increases the likelihood of opportunistic attacks. Furthermore, the fact that HTTP/2 control frames do not appear in standard access logs makes detection difficult, requiring deeper monitoring of process behavior and network traffic.
What you can do
- Upgrade NGINX: Move to version 1.31.9 or 1.30.7 for Open Source, or the corresponding fixed NGINX Plus versions.
- Restart Properly: Perform a full restart or live binary upgrade after patching; a simple reload may leave old vulnerable workers running.
- Disable HTTP/2: If immediate patching is not feasible, disable HTTP/2 by removing
http2fromlistendirectives or settinghttp2 off;in server blocks. - Block SCTP: Add
install sctp /bin/falseto/etc/modprobe.d/to prevent loading the SCTP kernel module, mitigating escalation paths. - Verify Configuration: Use
openssl s_client -alpn h2to confirm whether your server still accepts HTTP/2 connections after changes. - Monitor Logs: Look for signs of compromise such as unexpected program launches by NGINX workers or heavy inbound traffic with few logged requests.



