Opens in a new tab

NGINX RCE Vulnerabilities: Critical Fixes

Unauthenticated Remote Code Execution in NGINX: CVE-2026-123633 and CVE-2026-123007

Two critical NGINX RCE vulnerabilities were disclosed on 18 November 2026, exposing HTTP/2 enabled servers to unauthenticated remote attacks. F5 has released patched NGINX Open Source and NGINX Plus builds, while a public proof of concept increases the urgency for affected organisations.

NGINX RCE vulnerabilities disclosed in HTTP/2

The flaws are tracked as CVE-2026-123007 and CVE-2026-123633. Both affect code within NGINX itself rather than optional third party modules, meaning an organisation can be exposed even if its website or application code is secure.

NGINX commonly sits at the front of web environments, receiving encrypted traffic and routing requests to websites, applications and APIs. It powers more than 225 million websites, and one vulnerable server may handle traffic for several services.

Both NGINX RCE vulnerabilities require HTTP/2 to be enabled. Crafted client traffic can reach the vulnerable processing paths before authentication, potentially allowing a remote attacker to corrupt memory, crash an NGINX worker or execute code on the underlying server.

CVE-2026-123007 received a CVSS v4.0 score of 9.2 from F5, placing it in the Critical category. F5 also assigned it a CVSS v3.1 score of 8.1, which is High under that scoring framework. The vendor states that there is no control plane exposure.

How the two NGINX HTTP/2 flaws work

CVE-2026-123007: Integer underflow and buffer overflow

CVE-2026-123007 is an integer underflow in NGINX’s HTTP/2 flow control logic. HTTP/2 uses flow control windows to determine how much data can be sent across a connection. NGINX can incorrectly increase the applicable window beyond expected limits when processing specially constructed traffic.

This behaviour can cause the send window calculation to underflow. NGINX may then derive a negative or excessively large length while constructing HTTP/2 frames, resulting in a heap buffer overflow and a worker segmentation fault.

Under certain conditions, the memory corruption can be developed into remote code execution. F5 notes that this is particularly relevant to systems without Address Space Layout Randomisation (ASLR), while researchers also confirmed code execution where ASLR could be bypassed.

The flaw dates back to NGINX Open Source 1.9.5. Researchers demonstrated exploitation against a server delivering ordinary static files, without a backend application or third party module, confirming that the exposure sits in the web server’s core HTTP/2 processing.

CVE-2026-123633: Connection use after free

CVE-2026-123633 is a use after free vulnerability associated with HTTP/2 connection handling. Research describes an error path involving NGINX’s attempt to send a GOAWAY frame when closing a connection.

If frame allocation fails, including under resource pressure, the error handling can free connection state that remains referenced by the calling code. Execution then continues with a reference to released memory. If that memory is reused in a controlled way, an attacker may corrupt process state and potentially execute code.

Technical details for CVE-2026-123633 are currently based primarily on researcher reporting and secondary vulnerability trackers. An accessible F5 advisory with a separate version table for this CVE was not available at disclosure, so organisations should monitor F5 for further clarification.

Affected NGINX products and fixed versions

The confirmed prerequisite for the NGINX RCE vulnerabilities is an NGINX deployment that terminates HTTP/2 traffic. Servers that do not have HTTP/2 enabled do not expose these specific vulnerable paths.

F5’s advisory for CVE-2026-123007 identifies the following affected and fixed releases:

  • NGINX Open Source 1.9.5 through 1.30.6 is vulnerable. Version 1.30.7 contains the fix.
  • NGINX Open Source 1.31.0 through 1.31.8 is vulnerable. Version 1.31.9 contains the fix.
  • NGINX Plus 37.0.0 through 37.0.7.1 is vulnerable. Build 37.0.8.1 contains the fix.
  • NGINX Plus 37.1.0 through 37.2.0.1 is vulnerable. Build 37.3.0.1 contains the fix.

F5 released these patched builds on 18 November 2026. Its advisory indicates that other non NGINX products listed in the vendor’s assessment are not vulnerable.

Research and third party tracking describe CVE-2026-123633 as affecting NGINX Open Source and NGINX Plus when HTTP/2 is enabled, with affected product trains similar to CVE-2026-123007. Until F5 publishes or confirms a dedicated version matrix, organisations should apply the listed fixed releases and verify any additional guidance from the vendor.

Disclosure and patch timeline

The NGINX RCE vulnerabilities were found independently by several research teams, including Wiz Research, Ikotas Labs, Metnir, Rapid7 and Xint. The coordinated disclosure unfolded over several weeks:

  • 21 October 2026: Wiz reported the use after free issue later assigned CVE-2026-123633. F5 confirmed that another team had previously reported the same flaw.
  • 10 November 2026: F5 supplied a proposed CVE-2026-123633 patch to the researchers for review.
  • 16 November 2026: Wiz reported the HTTP/2 flow control issue later assigned CVE-2026-123007.
  • 17 November 2026: F5 confirmed that CVE-2026-123007 would be addressed in the planned security release.
  • 18 November 2026: F5 released patched builds and its CVE-2026-123007 advisory. Technical details for both vulnerabilities were disclosed publicly on the same day.

Current exploitation status

A public proof of concept for CVE-2026-123007, attributed to a Rapid7 researcher, was available on GitHub at disclosure. It requires HTTP/2 and targets the vulnerable flow control behaviour, reducing the work needed for others to investigate or attempt exploitation.

As of 18 November 2026, there was no authoritative confirmation of widespread exploitation in the wild, and F5 had not named any threat actor using the flaws. No concrete indicators of compromise, such as malicious IP addresses, domains or file hashes, had been published.

The absence of confirmed attacks does not remove the risk. Public technical details, a proof of concept and the internet facing role of NGINX give defenders a limited window in which to identify and update exposed systems.

Why these NGINX flaws matter

A compromised NGINX process can provide access to traffic, configuration, credentials or connected internal services, depending on how the server is deployed and what privileges it holds. An attacker could steal data, alter websites, interrupt services or use the server as a foothold for further activity.

The impact may extend beyond one application because a single reverse proxy or ingress server can route traffic to multiple systems. The long affected version range also means older, overlooked installations may remain exposed.

Actions for organisations using NGINX

Organisations should identify internet facing NGINX instances, confirm whether HTTP/2 is enabled and compare installed versions with F5’s affected ranges. Container images, reverse proxies, load balancers and ingress deployments should all be included.

  • Upgrade NGINX Open Source to 1.30.7 or 1.31.9, as appropriate.
  • Upgrade NGINX Plus to 37.0.8.1 or 37.3.0.1, as appropriate.
  • If patching cannot be completed immediately, temporarily disable HTTP/2 on externally reachable instances.
  • Review worker crashes and unusual HTTP/2 connection activity, while recognising that no definitive compromise indicators are currently available.
  • Track subsequent F5 advisories for version specific clarification concerning CVE-2026-123633.

Disabling HTTP/2 is a temporary risk reduction measure, not a substitute for installing the fixed releases. Patch verification should confirm both the running binary version and the version embedded in any container or packaged deployment.

Originally reported by wiz.io.

Share this bulletin

About the Author

Rob McBride Headshot - CyPro Partner and leading cyber security expert

Rob McBride

Partner

  • CISSP
  • ACA Chartered Accountant
  • MPhil
  • BSc
  • SOC 2
  • ISO 27001

Rob McBride

Rob is a Founding Partner at CyPro and a highly experienced CISO. Beginning his career with a successful tenure at Deloitte, Rob has since amassed a wealth of experience, notably serving as a cyber security advisor to the UK government and spearheading cloud security transformations for several global banks.

At CyPro, Rob leads the managed service business line, working extensively across multiple sectors including telecommunications, technology, higher education, travel, and retail. He is passionate about equipping small and medium-sized businesses (SMBs) with robust cyber security strategies to fuel their growth.

View Profile
Back to Bulletins

Related CyPro Services

  • Managed Detection and Response (MDR)

    Managed Detection and Response (MDR) is an end-to-end managed service designed to help organisations detect, analyse and respond to cyber threats quickly and effectively. It...
    View Service
CyPro Cookie Consent

Hmmm cookies...

Our delicious cookies make your experience smooth and secure.

Privacy PolicyOkay, got it!

We use cookies to enhance your experience, analyse site traffic, and for marketing purposes. For more information on how we handle your personal data, please see our Privacy Policy.

Schedule a Call