Five newly disclosed NGINX vulnerabilities could allow unauthenticated attackers to execute code on vulnerable web servers. The flaws, collectively known as NGINX2Root, were disclosed on November 18, 2026. Security updates are now available, but administrators must verify their NGINX versions, enabled protocols and deployment configurations to establish their actual exposure.
The most concerning issue, CVE-2026-123007, affects NGINX installations with HTTP/2 enabled. Researchers have demonstrated remote code execution under vulnerable conditions, while a separate Linux kernel vulnerability was used in laboratory research to escalate privileges to root.
The five vulnerabilities have an 8.1 CVSS v3.1 score and a 9.2 CVSS v4.0 score in F5’s assessment. Exploitation requires vulnerability-specific configuration conditions.
- What Is NGINX2Root and Why Does It Matter?
- Five NGINX Vulnerabilities: Which Configurations Are Affected?
- Important: NGINX2Root Does Not Automatically Mean Root Access
- Which NGINX Versions Contain the Security Fixes?
- Does Cloudflare Protect NGINX Servers Against NGINX2Root?
- What Should VPS and Web Server Administrators Do Now?
- Basic Local Verification Commands
- Can HTTP/2 Be Disabled as a Temporary Mitigation?
- Are NGINX Servers Already Being Attacked?
- NGINX2Root: Frequently Asked Questions
- What Happens Next?
What Is NGINX2Root and Why Does It Matter?
NGINX2Root is the name given by security researchers at Xint to a group of memory-safety vulnerabilities affecting NGINX, one of the world’s widely deployed web servers and reverse proxies.
The issues involve HTTP/2 flow control, connection management, HTTP/3 processing, DNS resolution and mail proxy authentication. They can cause memory corruption or use-after-free conditions that may allow an attacker to execute code within the privileges of an NGINX worker.
In its technical disclosure, Xint described the vulnerabilities and demonstrated exploitation under controlled conditions. Other researchers, including teams at Wiz and Rapid7, independently identified serious HTTP/2-related weaknesses.
The significance extends beyond websites directly served by NGINX. The software also operates as a reverse proxy, caching layer, load balancer and gateway to applications running on other infrastructure.
Five NGINX Vulnerabilities: Which Configurations Are Affected?
Not every NGINX deployment is vulnerable to every issue. The conditions depend on which protocol modules are enabled and how requests are handled.
| CVE | Affected Component | Required Condition | Potential Impact |
|---|---|---|---|
| CVE-2026-123007 | HTTP/2 flow control | HTTP/2 enabled on a vulnerable listener | Memory corruption and potential RCE |
| CVE-2026-123633 | HTTP/2 connection handling | HTTP/2 with upstream proxying and specific connection-draining conditions | Use-after-free and potential RCE |
| CVE-2026-121810 | HTTP/3 / QUIC | HTTP/3 enabled with affected upstream processing | Use-after-free and potential RCE |
| CVE-2026-123634 | DNS resolver | Specific dynamic proxy and subrequest configurations | Use-after-free and potential RCE |
| CVE-2026-113425 | Mail proxy | Proxy authentication using the affected mail module | Use-after-free and potential RCE |
Sources: NGINX security advisories, F5 and CERT-EU Security Advisory 2026-022. These descriptions summarize required conditions, not complete exploitability tests.
The vulnerability receiving particular attention is CVE-2026-123007. It affects the HTTP/2 flow control implementation and has been present in affected NGINX code since HTTP/2 support was introduced in 2015.
CERT-EU describes the issue as an integer overflow that can produce an out-of-bounds heap write. Researchers demonstrated remote code execution, making the issue especially relevant to internet-facing HTTPS servers with HTTP/2 enabled.
Important: NGINX2Root Does Not Automatically Mean Root Access
The five NGINX vulnerabilities can lead to code execution with the privileges of the NGINX worker. Xint separately demonstrated a chain involving a Linux kernel SCTP vulnerability to obtain root privileges, including from a container. This additional privilege escalation is not an automatic result of exploiting NGINX and depends on a separate host vulnerability.
Which NGINX Versions Contain the Security Fixes?
The NGINX project released patched Open Source versions on November 18, 2026. Administrators should update to:
| Product | Fixed Release | Notes |
|---|---|---|
| NGINX Open Source stable | 1.30.7 | Stable branch security release |
| NGINX Open Source mainline | 1.31.9 | Mainline branch security release |
| NGINX Plus R37 | 37.3.0.1 | Commercial release family |
| NGINX Plus R37 earlier track | 37.0.8.1 | Specified fixed build |
Linux distributions may backport fixes into older-looking package versions. Therefore, administrators should compare their exact installed package build against the operating-system vendor’s security advisory rather than relying exclusively on the upstream version number.
The authoritative upstream list is maintained on the official NGINX security advisories page.
Does Cloudflare Protect NGINX Servers Against NGINX2Root?
Cloudflare published an engineering analysis on November 20 explaining how its HTTP reverse proxy changes the exposure of origin servers.
When a hostname is properly proxied, Cloudflare terminates client HTTP/2 and HTTP/3 connections at its edge and creates separate connections to the origin. The attacker’s original protocol control frames are not simply forwarded to NGINX.
Simplified architectural diagram. This protection depends on the attack path passing through the Cloudflare HTTP proxy.
Cloudflare says this prevents the direct client-controlled HTTP/2 and HTTP/3 attack sequences from reaching properly protected origins.
However, the protection is conditional. An attacker who can connect directly to the origin IP address may bypass the proxy. Certain resolver-related configurations and mail services are also outside the protection provided by Cloudflare’s HTTP proxy.
Administrators should therefore restrict origin access appropriately and install the NGINX patches regardless of whether Cloudflare is deployed.
What Should VPS and Web Server Administrators Do Now?
The recommended response is to identify vulnerable installations, verify the relevant configurations, install supported updates and confirm that the running server process has been replaced.
- Identify the running NGINX version. Check the installed binary and package information. If the server uses vendor packages, review the distribution-specific security notice for backported fixes.
- Inspect relevant configuration. Determine whether HTTP/2, HTTP/3, dynamic proxying, custom DNS resolvers or NGINX mail proxy features are active. Review configuration locally and avoid exposing sensitive output.
- Install the appropriate fixed release. Use the official NGINX distribution channel or your operating-system vendor’s patched packages. Test the application stack before production deployment.
- Replace affected running processes. Follow the vendor’s update instructions. Validate configuration and ensure that patched NGINX workers are running; merely changing files on disk is not sufficient.
- Review suspicious activity. Examine worker crashes and unexpected child processes, together with system and security telemetry. A crash alone is not proof of successful compromise.
- Confirm recovery readiness. Maintain tested backups of server configuration and application data. Ensure recovery procedures can work independently of a potentially compromised production server.
Basic Local Verification Commands
On a Linux server, administrators can begin with the following local checks:
nginx -v nginx -V sudo nginx -t sudo systemctl status nginx
The first command displays the NGINX version. The second provides build information, while the configuration test checks for syntax errors. The service status command helps confirm whether NGINX is active under systemd.
These commands are diagnostic, not a complete vulnerability assessment. They do not determine whether all vendor patches are installed or whether a specific CVE is exploitable.
Teams planning production changes should also maintain independent recovery procedures. Our VPS backup and disaster recovery guide explains why server snapshots alone may not be sufficient after a serious infrastructure incident.
Can HTTP/2 Be Disabled as a Temporary Mitigation?
CERT-EU confirms that disabling HTTP/2 on affected listeners can mitigate CVE-2026-123007 and CVE-2026-123633 while administrators prepare updates.
But this does not address the other three NGINX2Root vulnerabilities, including issues affecting HTTP/3, DNS resolution and mail proxying.
Any protocol configuration change should be tested before deployment, particularly on high-traffic production services where disabling a protocol can affect performance or compatibility.
Are NGINX Servers Already Being Attacked?
Researchers have publicly demonstrated exploitability, and proof-of-concept research has increased the information available to security teams and potential attackers.
However, laboratory exploitation, internet exposure estimates and confirmed real-world compromise are three different things.
CERT-EU’s initial advisory reported no confirmed exploitation in the wild. Cloudflare’s November 20 analysis similarly did not identify confirmed attacks involving the five flaws. These statements describe the evidence available at the time of those reports, not a guarantee that exploitation cannot occur.
Censys estimated approximately 8.8 million potentially exposed NGINX services on November 19 using version and protocol observations. That figure should not be interpreted as 8.8 million confirmed exploitable installations, since package backports and deployment details can change actual exposure.
The broader security lesson also applies to other internet-facing infrastructure. Servers.News recently covered a critical Brotli vulnerability affecting Node.js and Envoy , another example of how vulnerabilities in shared infrastructure components can affect multiple application stacks.
NGINX2Root: Frequently Asked Questions
Does this mean my WordPress website is vulnerable?
Not necessarily. WordPress is an application, while NGINX is a web server that may serve or proxy requests to WordPress. Exposure depends on the installed NGINX package, enabled features and server configuration. A WordPress core update does not replace an NGINX security update.
Does my server need updating if I use Cloudflare?
Yes. Cloudflare’s HTTP proxy can block important client-controlled protocol attack paths when the origin is correctly protected. It does not remove the vulnerabilities from NGINX or guarantee protection against every configuration involved in NGINX2Root.
Is NGINX 1.18.0 automatically vulnerable?
Not necessarily. Some Linux distributions apply security backports while retaining older upstream version numbers. Check the exact package revision and the operating-system vendor’s security advisory before drawing a conclusion.
Can an attacker always obtain root access?
No. The NGINX vulnerabilities can allow execution with worker-process privileges. The root-level demonstration involved chaining a separate Linux kernel flaw and does not describe the outcome of every NGINX exploitation scenario.
What Happens Next?
The NGINX security fixes are available, and the immediate priority is applying appropriate vendor updates to exposed installations.
Security teams should continue monitoring vendor advisories, newly published detection guidance and reliable reports about real-world exploitation.
Future disclosures concerning the separate Linux kernel vulnerability may also change the risk assessment for organizations operating affected hosts and containers.
Bottom line: NGINX2Root is a serious web-server security issue, but effective response starts with confirming the actual configuration rather than assuming every NGINX installation is exploitable. Apply verified patches, protect origin access and review relevant security telemetry.
Editorial sources: NGINX Security Advisories; CERT-EU Security Advisory 2026-022; Xint’s NGINX2Root technical research; F5 security guidance; Cloudflare’s November 20 engineering analysis; and independent reporting by BleepingComputer. Exploit demonstrations and internet exposure estimates are distinguished from confirmed real-world compromise.







