Five critical NGINX vulnerabilities put servers at risk of remote code execution

Udi Doron
Udi Doron
Nov 18, 2026 | 10 min read
Five critical NGINX vulnerabilities put servers at risk of remote code execution

Five critical vulnerabilities have been disclosed in NGINX, one of the world's most widely used web servers, potentially allowing remote attackers to crash servers and, under certain conditions, execute malicious code. F5 rates all five as critical (CVSS v3.1 score of 8.1), and each one affects a specific NGINX feature: HTTP/2, HTTP/3, the mail proxy, or proxying to dynamically resolved server names.

The vulnerabilities are tracked as CVE-2026-123007, CVE-2026-123633, CVE-2026-123634, CVE-2026-121810, and CVE-2026-113425. The two HTTP/2 flaws, CVE-2026-123633 and CVE-2026-123007, were independently discovered by researchers from Wiz, Ikotas Labs, Metnir, Rapid7, and Xint, who coordinated with F5 and the NGINX team to address them.

Echo customers using patched NGINX images are already protected, with fixes also available across affected packages.

Key takeaways

  • CVE-2026-123007, CVE-2026-123633, CVE-2026-123634, CVE-2026-121810, and CVE-2026-113425 are NGINX vulnerabilities that F5 rates as critical.
  • All five can crash NGINX worker processes and corrupt memory. Where ASLR is disabled or bypassed, they can also allow remote code execution.
  • Each depends on a specific feature being in use: HTTP/2, HTTP/3, the mail proxy, or proxying with variable server names alongside SSI or mirror subrequests.
  • Several of the flaws are long-standing. CVE-2026-113425 affects NGINX versions back to 0.5.33, CVE-2026-123634 back to 0.6.18, and CVE-2026-123007 back to 1.9.5, released in 2015.
  • All five are fixed in NGINX OSS 1.30.7 (stable) and 1.31.9 (mainline).

What happened?

On November 18, 2026, F5 and the NGINX team disclosed five critical vulnerabilities in NGINX, the open-source web server and reverse proxy used to handle traffic for websites, APIs, and enterprise applications.

NGINX frequently sits at the front of production infrastructure, accepting incoming connections and routing requests to backend applications. This makes it a particularly sensitive component: a vulnerability in NGINX can expose multiple services, even when the applications behind it contain no exploitable flaws.

All five vulnerabilities are memory-safety issues in NGINX worker processes that can lead to remote code execution (RCE). The baseline impact is a denial of service through worker crashes, and code execution requires ASLR to be disabled or bypassed. An attacker who successfully exploits one of these vulnerabilities could execute commands with the privileges of the NGINX worker process, access sensitive data, modify website content, or disrupt services.

Researchers demonstrated working exploits for the HTTP/2 vulnerabilities in controlled environments. At the time of disclosure, there were no confirmed reports of exploitation in the wild for any of the five.

Understanding the five NGINX vulnerabilities

Although all five vulnerabilities can lead to the same outcome, they originate from different memory-safety issues across several NGINX modules.

CVE-2026-123633: HTTP/2 use-after-free vulnerability

CVE-2026-123633 is a use-after-free vulnerability in the way NGINX manages HTTP/2 connections.

Under certain conditions, NGINX can incorrectly mark an active connection as reusable. When available connection slots become limited, the server may close that connection and free its memory, even though a request is still being processed.

The request can then continue accessing memory that has already been released. An attacker who manipulates this behavior may be able to corrupt memory and achieve remote code execution.

The vulnerability affects NGINX reverse-proxy configurations with HTTP/2 enabled when a worker process runs low on worker_connections, beginning with NGINX OSS version 1.19.7. Servers that only serve static files aren't affected.

CVE-2026-123007: HTTP/2 flow-control memory corruption vulnerability

CVE-2026-123007 is a memory corruption vulnerability in NGINX's HTTP/2 flow-control implementation.

HTTP/2 uses flow control to regulate how much data can be transmitted over a connection. NGINX maintains counters to track these limits, but specially crafted traffic can push the flow-control window past its allowed threshold.

According to F5, this can result in a heap buffer overflow, typically followed by a crash of the NGINX worker, allowing an attacker to corrupt memory and potentially execute arbitrary code.

This vulnerability is particularly concerning because it affects NGINX versions dating back to 1.9.5, released in 2015.

Researchers also demonstrated that exploitation does not require a vulnerable backend application or third-party module. An affected NGINX server serving ordinary static files can be vulnerable under the relevant conditions.

CVE-2026-123634: Resolver use-after-free vulnerability

CVE-2026-123634 is a use-after-free vulnerability in how NGINX's resolver processes responses.

It affects NGINX configured as a proxy when the proxied server name includes variables, and a Server Side Includes (SSI) or mirror directive sends a subrequest to a location that resolves the same server name. Under these conditions, a specially crafted HTTP request can trigger a use-after-free in a worker process.

An attacker could use this to crash the worker, corrupt its memory, or potentially execute code. The vulnerability affects NGINX OSS versions dating back to 0.6.18.

CVE-2026-121810: HTTP/3 use-after-free vulnerability

CVE-2026-121810 is a use-after-free vulnerability affecting NGINX configurations that use HTTP/3.

A remote, unauthenticated attacker can trigger the flaw to make an NGINX worker process restart, causing a denial of service, and may be able to execute code. Successful exploitation also depends on conditions beyond the attacker's control.

The vulnerability affects NGINX OSS versions beginning with 1.25.0.

CVE-2026-113425: Mail proxy use-after-free vulnerability

CVE-2026-113425 is a use-after-free vulnerability affecting NGINX configured as a mail proxy.

A remote, unauthenticated attacker can send traffic that triggers the flaw in a worker process, crashing it or corrupting its memory and disrupting service. Remote code execution is a potential impact.

This is the oldest of the five flaws, affecting NGINX OSS versions dating back to 0.5.33.

Which NGINX versions are affected?

Exposure depends on which NGINX features are in use, as each vulnerability requires a different configuration.

Vulnerability Affected feature First affected NGINX OSS version Fixed NGINX OSS versions
CVE-2026-123007 HTTP/2 1.9.5 1.30.7 stable / 1.31.9 mainline
CVE-2026-123633 HTTP/2 with proxying to upstreams 1.19.7 1.30.7 stable / 1.31.9 mainline
CVE-2026-123634 Proxy and resolver configurations 0.6.18 1.30.7 stable / 1.31.9 mainline
CVE-2026-121810 HTTP/3 1.25.0 1.30.7 stable / 1.31.9 mainline
CVE-2026-113425 Mail proxy 0.5.33 1.30.7 stable / 1.31.9 mainline

NGINX Plus is also affected, with fixes available in its supported release branches. NGINX Plus users should check F5's advisory for each CVE to confirm which release fixes it on their branch.

Organizations should check both the NGINX version and whether the affected features are enabled. Running an affected version does not necessarily mean a deployment is exploitable, but internet-facing systems meeting the vulnerable configuration requirements should be prioritized.

Why these vulnerabilities are particularly dangerous

The biggest concern is where NGINX typically operates.

Unlike vulnerabilities buried inside individual application dependencies, these flaws affect software that often serves as the entry point to production environments. NGINX may process requests for multiple applications, APIs, and services from a single deployment.

An attacker exploiting a vulnerable NGINX server could potentially compromise that shared infrastructure without first authenticating to any of the applications behind it.

The vulnerabilities also demonstrate how long-standing weaknesses can remain hidden in widely used software. Several of these flaws have existed for more than a decade, despite NGINX being deployed across millions of production environments.

The growing use of AI-assisted vulnerability research makes this especially relevant. Researchers can increasingly identify and validate complex memory-safety flaws that previously required substantial manual effort. As vulnerability discovery accelerates, organizations need a way to turn new findings into protection just as quickly.

How to protect against the NGINX vulnerabilities

Organizations running NGINX should take the following steps:

  1. Identify affected deployments. Inventory NGINX instances, including container images and application dependencies, and determine which use HTTP/2, HTTP/3, the mail proxy, or proxying with variable server names.
  2. Upgrade to a patched release. For NGINX OSS, update to version 1.30.7 or 1.31.9, or a later release, which fix all five vulnerabilities. NGINX Plus users should apply the F5 security updates for their release branch. Restart NGINX after upgrading, since a reload keeps the old binary running.
  3. Reduce exposure if upgrading isn't immediately possible. For the two HTTP/2 vulnerabilities, on NGINX 1.25.1 or later, configure http2 off; and remove HTTP/2 parameters from relevant listen directives. On older versions, remove the http2 parameter from listen directives. Test compatibility, reload NGINX, and verify that old workers have exited. Clients will fall back to HTTP/1.1, which breaks gRPC routes that use grpc_pass. The other three vulnerabilities depend on HTTP/3, the mail proxy, or proxying with variable server names alongside SSI or mirror subrequests. If you don't need those features, disabling them removes the exposure. Check F5's advisories for any other interim mitigations.
  4. Review potentially suspicious activity. Unexpected NGINX worker crashes or unusual commands executed by NGINX processes may warrant investigation, although these are not confirmed indicators of exploitation.
  5. Verify the software running in production. Confirm that patched versions have reached deployed container images and services, rather than relying only on the availability of an upstream fix.

How Echo protects customers against the NGINX vulnerabilities

Echo has already delivered patched NGINX software to customers, addressing these critical vulnerabilities without requiring teams to manage the underlying remediation themselves.

For all five vulnerabilities, Echo's advisory platform confirms that fixes have been completed across the same 11 package entries, covering NGINX, OpenResty, and the NGINX Ingress Controller, including FIPS variants.

This broader coverage matters because organizations rarely consume NGINX in isolation. They also depend on ingress controllers, application platforms, and other software that incorporates NGINX. Patching the underlying web server alone does not automatically ensure every downstream image is secure.

Echo continuously monitors vulnerabilities affecting the open-source software its customers depend on, validates the necessary fixes, and delivers hardened, patched software through the sources customers already use.

Instead of researching affected versions, rebuilding images, and coordinating individual patching efforts, teams can consume maintained software with security fixes already incorporated.

For organizations that need to remain on specific software versions, Echo's backporting capabilities also help deliver security fixes without requiring disruptive version upgrades.

The result is a shorter path from vulnerability disclosure to protection, with less work falling on security and engineering teams.

The growing importance of faster vulnerability remediation

The disclosure of these NGINX vulnerabilities comes just one day after Java4Shell, another serious memory-safety vulnerability affecting widely used infrastructure.

Both incidents highlight a growing challenge for enterprise security teams. As researchers uncover complex vulnerabilities in foundational software, organizations must rapidly determine whether they're affected, identify available fixes, and ensure those fixes reach production.

Having a patch available is an important first step. But organizations are only protected once that patch reaches the software they're actually running.

With Echo, that remediation work is handled at the software source, so teams can continue using the technologies they depend on without carrying the burden of patching every newly discovered vulnerability.

Already using Echo? Customers consuming the relevant patched NGINX software are protected against the remediated vulnerabilities.

Not using Echo yet? Get a demo to learn how Echo delivers continuously maintained, hardened open-source software across your supply chain.

Frontier models can now find and exploit zero days in your code

Find out what to do when there's no CVE, scanner alert or fix available yet.

What are the 7 blind spots in your vulnerability scans?

Discover when "0 vulnerabilities" doesn't actually mean you're clean.

Read now →

Ready to eliminate vulnerabilities at the source?