Blog
AppSec Blog

NGINX2Root: Critical NGINX vulnerabilities need patching, but what’s the real risk?

 - 
noviembre 19, 2026

On November 18, 2026, critical vulnerabilities in NGINX were disclosed under the name NGINX2Root, with the potential to allow unauthenticated remote code execution (RCE). While the security risks are real and require urgent patching, the dramatic name overstates the direct impact. Here’s what’s actually vulnerable, how serious the threat is, and what organizations should do.

You information will be kept Private
Table of Contents

The NGINX2Root name, coined by security researchers at Xint, covers five memory-safety vulnerabilities affecting different NGINX configurations. The most widespread, CVE-2026-123007, affects installations with HTTP/2 enabled and has been present in the code since 2015. It carries a CVSS v4.0 score of 9.2 (Critical), and a working proof-of-concept exploit has been published.

Despite the dramatic name, NGINX2Root does not automatically give attackers root access. The five NGINX vulnerabilities may allow code execution with the privileges of an NGINX worker process. The demonstrated path to root privileges involves chaining one of these vulnerabilities with a separate Linux kernel flaw.

What to know about NGINX2Root

  • NGINX2Root refers to five memory-safety vulnerabilities reported in NGINX, all of which could potentially allow unauthenticated RCE under different conditions.
  • CVE-2026-123007 poses the broadest risk because it affects NGINX installations with HTTP/2 enabled, including servers that only deliver static content. It has a CVSS v4.0 score of 9.2 (Critical) and a CVSS v3.1 score of 8.1 (High).
  • A working lab exploit for CVE-2026-123007 is publicly available, but no exploitation in the wild has been confirmed as of November 19, 2026.
  • NGINX Open Source versions 1.30.7 (stable) and 1.31.9 (mainline), released November 18, address all five vulnerabilities. NGINX Plus also has fixed builds available.
  • Disabling HTTP/2 temporarily mitigates two of the five vulnerabilities (including CVE-2026-123007), but does not address the others.
  • Invicti has released a security check that helps identify vulnerable NGINX installations by examining their exposed version and confirming HTTP/2 support without attempting exploitation.

NGINX2Root: Background and disclosure timeline

NGINX is widely used as a web server, reverse proxy, and load balancer. It frequently sits directly in front of production applications and APIs, handling incoming traffic before requests reach the application itself. That position makes vulnerabilities in NGINX particularly significant. If an attacker can exploit a flaw in the web server, they may not need to find a vulnerability in the application behind it.

The timeline for NGINX2Root so far:

  • The NGINX2Root vulnerabilities were discovered during October and November 2026.
  • Xint identified five separate memory-safety issues, while several other research teams, including Wiz Research, Rapid7, Ikotas Labs, and Metnir, independently discovered the two HTTP/2 vulnerabilities.
  • Wiz reported CVE-2026-123633 on October 21 and CVE-2026-123007 on November 16. Rapid7 independently reported CVE-2026-123007 in November and subsequently published a working exploit.
  • On November 18, F5 and the NGINX project released coordinated security updates in versions 1.30.7 and 1.31.9. The releases address ten security vulnerabilities in total, including the five grouped under NGINX2Root.

The coordinated disclosure was accompanied by extensive technical research and demonstrations, which brought considerable attention to vulnerabilities that had remained undiscovered for years.

Does NGINX2Root actually give attackers root access?

The name NGINX2Root was coined by Xint and applies not to the direct impact of the five NGINX vulnerabilities but to a specific exploit chain that also requires a separate Linux kernel vulnerability.

Each of the five NGINX vulnerabilities can potentially provide remote code execution with the permissions of an NGINX worker process. This is already a serious security problem: depending on the environment, an attacker could access application data, interfere with requests, steal credentials accessible to the worker, or disrupt services.

However, NGINX workers normally run under restricted accounts rather than as root. To demonstrate full host compromise, Xint combined the NGINX vulnerabilities with a separate, reportedly unpatched use-after-free vulnerability in the Linux kernel’s SCTP implementation. The researchers reported successfully escalating privileges to root, including from inside a Docker container. The kernel vulnerability is separate from NGINX2Root’s five CVEs, and the full privilege-escalation chain has not been independently reproduced in public research. 

The distinction does not reduce the urgency of patching NGINX, but it does mean that any claims of automatic root compromise from any vulnerable NGINX installation are misleading.

How do the NGINX2Root vulnerabilities work?

All five vulnerabilities involve memory corruption, but they affect different parts of NGINX and require different conditions to exploit. The most significant is CVE-2026-123007, which affects HTTP/2 flow control. Unlike several of the other vulnerabilities, it does not require NGINX to proxy requests to a backend application.

CVE-2026-123007: Remote code execution through HTTP/2 flow-control corruption

HTTP/2 uses flow control to regulate how much data a server can send over a connection. The client advertises how much data it can receive, and the server maintains a corresponding send window. CVE-2026-123007 is an integer underflow vulnerability in the way NGINX manages this window.

Under specific conditions, an attacker can manipulate HTTP/2 flow-control state so that NGINX incorrectly increases the available send window beyond the protocol’s permitted limit. Subsequent arithmetic operations can then cause NGINX to interpret the window incorrectly, eventually passing an invalid length to a memory-copy operation. The resulting out-of-bounds memory access can crash an NGINX worker or expose process memory. Under suitable conditions, attackers can exploit the memory corruption to achieve remote code execution.

Researchers have demonstrated that the vulnerability can be exploited even on an NGINX server serving only static files, without a backend application or additional modules. This makes it relevant to ordinary web servers as well as reverse proxies.

The vulnerable code was introduced with HTTP/2 support in NGINX 1.9.5 in 2015. Consequently, installations running affected versions with HTTP/2 enabled should be considered potentially exposed, even if their application code is fully patched.

The other four NGINX2Root vulnerabilities

The remaining vulnerabilities also allow memory corruption that can potentially lead to code execution, but their prerequisites are more specific than just having HTTP/2 enabled:

CVE Vulnerability Required configuration
CVE-2026-123633 Use-after-free during HTTP/2 connection handling HTTP/2 enabled, with upstream proxying and specific connection-handling conditions
CVE-2026-123634 Use-after-free in the DNS resolver Dynamic upstream resolution and specific subrequest behavior, demonstrated with mirror
CVE-2026-121810 Use-after-free during HTTP/3 connection handling HTTP/3 enabled with upstream proxying
CVE-2026-113425 Use-after-free in the mail proxy module Mail proxy using proxy on and auth_http

These differences are important for temporary mitigation. Disabling HTTP/2 addresses exposure to CVE-2026-123007 and CVE-2026-123633, but does nothing to mitigate the other three vulnerabilities.

If immediate patching is not possible, each organization should therefore assess their actual NGINX configurations rather than assume that every installation has the same exposure.

Is NGINX2Root being actively exploited?

As of November 19, 2026, there are no confirmed reports of exploitation in the wild for the five NGINX2Root vulnerabilities.

A working public exploit does exist for CVE-2026-123007 – Rapid7 published proof-of-concept code demonstrating unauthenticated RCE against NGINX 1.31.8 running HTTP/2 over TLS. However, the published exploit is technically demanding and appears impractical for large-scale exploitation in its current form. It requires sending approximately 128 GB of HTTP/2 traffic over a single connection within a limited time and depends on specific processor and memory-copy behavior. Network bandwidth and connection time limits are the main practical constraints.

While Rapid7’s published PoC exploit is unsuitable for straightforward, large-scale exploitation, more effective techniques may be published in the future. Both Xint and Wiz have reported additional successful exploitation methods, including approaches that overcome some of the limitations described by Rapid7. Full technical details of those methods have not been publicly released.

The current situation calls for a measured response: there is no confirmed widespread attack campaign, but the vulnerabilities are real, working exploits have been demonstrated, and public disclosure increases the likelihood of further exploit development. Organizations should not wait for confirmed attacks before applying the fixes.

How to remediate NGINX2Root vulnerabilities

The recommended action is to upgrade affected NGINX installations to a fixed release and verify that the updated software is running.

Upgrade to a fixed NGINX version

The November 18 security releases address all five NGINX2Root vulnerabilities:

Product Fixed version
NGINX Open Source stable 1.30.7
NGINX Open Source mainline 1.31.9
NGINX Plus 37.0 branch 37.0.8.1
NGINX Plus 37.1/37.2 branch 37.3.0.1

Organizations running older NGINX Plus branches should consult the relevant F5 advisory and plan an upgrade to a supported fixed release.

Note that Linux distributions may also backport security fixes into older NGINX package versions. This means that the upstream version reported by nginx -v does not always indicate whether a distribution package is vulnerable. For example, Ubuntu has released fixes for several supported distributions while retaining older upstream NGINX version numbers. Administrators should consult their distribution’s security advisory and compare the complete installed package version against the vendor’s fixed package version.

Remember to restart NGINX after updating, or perform a supported live binary upgrade, and verify that no vulnerable worker processes remain active. Reloading the configuration alone does not update the running binary.

Apply temporary mitigations where patching is delayed

Where an immediate upgrade is not possible, temporary configuration changes can reduce exposure:

  • Disable HTTP/2 to mitigate CVE-2026-123007 and CVE-2026-123633. For NGINX 1.25.1 and later, use http2 off; in the relevant server blocks and remove any legacy http2 parameter from listen directives. For older versions, remove the http2 parameter from affected listeners.
  • Disable HTTP/3 on affected installations to mitigate CVE-2026-121810.
  • Review dynamic DNS resolution and subrequest configurations for CVE-2026-123634. Where possible, replace request-derived proxy destinations with fixed upstream groups and remove unnecessary subrequest features.
  • Restrict mail proxy exposure for CVE-2026-113425, particularly where auth_http is configured.

Again, these measures should be treated as temporary risk reduction, not replacements for patching.

Considering the widespread use of NGINX in various roles, organizations should also inventory NGINX installations beyond conventional web servers, including containers, Kubernetes ingress components, appliances, and NGINX-derived products. These may require their own vendor updates rather than a straightforward package upgrade.

How Invicti helps identify vulnerable NGINX installations

Invicti has released a security check for NGINX2Root to help organizations identify exposed NGINX installations that may be vulnerable to the critical HTTP/2 remote code execution flaw.

The check uses an active but non-exploitative detection method. It examines NGINX-generated responses to identify the server version and confirms that NGINX itself supports HTTP/2 over TLS. It then evaluates the observed version against the known vulnerable ranges. This approach provides actionable vulnerability findings without attempting to exploit the memory-corruption vulnerability.

The check is designed to avoid reporting vulnerabilities when the available evidence is ambiguous, thus helping security teams prioritize installations that warrant investigation and remediation. As with other version-based checks, findings should be evaluated alongside vendor security advisories, particularly where Linux distributions may have backported fixes.

By identifying potentially vulnerable NGINX deployments during dynamic application security testing, Invicti helps teams bring infrastructure-level exposure into their application security workflows and prioritize remediation alongside other security findings.

Final thoughts

NGINX2Root is another memorable name for another serious set of vulnerabilities, but its branding should not distract from the underlying technical facts.

The five NGINX vulnerabilities may enable unauthenticated remote code execution, but they have different prerequisites and do not automatically grant root access. The reported privilege escalation chain depends on an additional Linux kernel vulnerability, as yet undisclosed.

There is a big difference between a proof of concept that works in the lab and confirmed exploitation in the wild, but the absence of confirmed attacks does not reduce the urgency of patching, particularly as more practical exploitation techniques may emerge.

The immediate priorities are straightforward:

  • Identify affected NGINX installations
  • Determine which vulnerable features are enabled
  • Patch and restart to apply fixes
  • Use appropriate temporary mitigations where updates cannot be deployed immediately

Invicti’s new security check helps organizations identify potentially vulnerable HTTP/2-enabled NGINX installations without attempting exploitation, which supports a faster and more informed response to CVE-2026-123007 and other NGINX2Root vulnerabilities. 

For security teams, the practical lesson is familiar: focus on verified exposure and actionable remediation rather than the most dramatic interpretation of a vulnerability’s name.

Frequently asked questions

No items found.
Table of Contents