- SMB/SME
- Enterprise
· Updated · ASC Editorial Team
NGINX2Root: A Public Exploit for the Web Server in Front of Everything
On 18 November 2026, F5 released fixes for five critical flaws in NGINX, the web server and reverse proxy that sits in front of a large share of the world's websites, APIs and applications. The most serious, CVE-2026-123007, is in NGINX's HTTP/2 code and goes back to version 1.9.5, released in 2015. Rapid7, whose researcher found it, has published a working exploit that can run code on an unpatched server without logging in. Researchers at Xint have named the set "NGINX2Root".
No attacks had been reported by 19 November, and none of the five flaws is on CISA's list of known exploited vulnerabilities. CERT-EU, the cyber security service for the EU institutions, advises updating as soon as possible, internet-facing servers first.
The public exploit has limits. According to Rapid7, it needs HTTP/2 turned on, depends on certain x86-64 processors, and needs about 128 GB of HTTP/2 traffic, around 17 minutes over a 1 Gbit/s link. It runs code as an NGINX worker process, not as root. Rapid7 says other routes to code execution may exist, though none are known. Those limits help you decide which servers to patch first. They aren't a reason to wait, because exploits tend to improve once they're public.
The other four flaws depend on particular features: HTTP/2 or HTTP/3 forwarding requests to back-end servers, the DNS resolver used in a particular way, or the mail proxy. All five are fixed in NGINX Open Source 1.30.7 and 1.31.9, and in NGINX Plus 37.0.8.1 and 37.3.0.1. There is no fix for NGINX Plus R34 to R36, so those need moving to a 37.x release.
This comes a week after the Apache HTTP/2 flaw, and the hard part is the same. NGINX is also built into container images and commercial products, where it may not appear as NGINX in an asset list. Check Point, for example, says its WAF as a Service is not affected and its WAF Agent Docker images aren't affected in their default settings, and it is preparing WAF Agent images based on NGINX 1.30.7. Palo Alto Networks says PAN-OS firewalls and Panorama are affected where HTTP/2 is turned on, as we explain in our article on its advisory. Fortinet says FortiOS, FortiProxy and FortiMail aren't affected, and it is still checking about 50 other products, including FortiWeb, FortiADC and FortiManager.
Three steps follow.
First, find every NGINX, including those in container images and in products you bought, and note which have HTTP/2 or HTTP/3 turned on.
Second, update to a fixed release, or your distribution's fixed package, starting with internet-facing servers. Rebuild container images too: restarting a container doesn't update the software inside it. If you can't update straight away, turning HTTP/2 off on exposed listeners removes the exposure to the two HTTP/2 flaws, including the one with the public exploit. Services that use gRPC need HTTP/2, so update those instead.
Third, watch for attempts. CERT-EU suggests checking NGINX's error log for worker processes that "exited on signal". Repeated crashes can mean someone is trying the exploit, so check what else the server was running at the time.
Do you know every place NGINX runs in your estate, including inside the products and images you didn't build yourself?
Get Insights by email
A short round-up of new articles every Monday morning. Nothing else.