Web Development & Digital Marketing Company in Dubai, UAE

Contact Us

A ten-year-old nginx flaw, and the one question to ask your hosting provider

Hosting, server updates and website maintenance, in house in Dubai since 2011. Rated 4.9 from 201 Google reviews.

Ask us to check your server

On 18 November 2026, the makers of nginx released security updates for five flaws in the software. Nginx is the web server that sits in front of a very large share of the world's websites, quite possibly including yours.

One of the five matters more than the others. It has been in the software since September 2015, and on an unpatched server it can let someone run their own code without logging in first.

Nobody has reported anyone actually using it yet. The fix is already available, and Ubuntu and Debian have both shipped updated packages. So this is not a panic. It is a deadline, and it comes with a single question worth asking today.

Server check

Not sure whether your website server is patched

Send us your website address and tell us who hosts it. Tomsher will tell you whether your server is running an affected version, who is responsible for updating it, and what to ask them for. A short written answer, and there is no obligation after it.

If your host has already patched and nothing needs doing, that is what we will tell you.

What nginx is, in one paragraph

When somebody opens your website, their browser does not talk to your website directly. It talks to a web server first, and that web server decides what to send back. Nginx is one of the two or three most widely used pieces of software for that job. If your site runs on a virtual private server, a cloud server or most managed hosting, there is a reasonable chance nginx is in front of it. You never see it, you never log into it, and in most businesses nobody can say who last updated it.


What was announced

Five flaws, published on nginx's own security advisories page. Here they are as nginx itself describes them.

CVEWhat nginx calls itSeverityVersions
CVE-2026-123007Memory corruption in HTTP/2 flow controlMajor1.9.5 to 1.31.8
CVE-2026-113425Use-after-free in mail proxy moduleMajor0.5.33 to 1.31.8
CVE-2026-123633Use-after-free in HTTP/2Medium1.19.7 to 1.31.8
CVE-2026-123634Use-after-free during response processing in resolverMedium0.6.18 to 1.31.8
CVE-2026-121810Use-after-free in HTTP/3Medium1.25.0 to 1.31.8

All five are fixed in the same two releases, 1.31.9 and 1.30.7.

One point on severity, because the coverage has been inconsistent. Some reports have described all five as scoring 9.2 out of 10. Nginx itself rates two of them as major and the other three as medium. Ubuntu scores the headline flaw at 9.2 out of 10 on the newer CVSS scale and 8.1 on the older one, and rates its own priority as high. The first flaw is serious by any measure. The other three are rated medium by the people who wrote the software, and it is worth knowing that before anyone takes a server offline in a hurry.

Why the first one is the one to care about

CVE-2026-123007 affects HTTP/2, a standard way browsers and servers talk to each other that is switched on by default on most modern setups. Security firm Wiz, which published its findings on 18 November, describes the outcome as being able to "Execute arbitrary code on the NGINX server without authentication".

In plain terms, someone who can reach your website over the internet could potentially get your server to run instructions they chose, without any password. That is about as serious as a web server problem gets.

Two things make it less severe than that sounds, and both are worth knowing.

F5, which owns nginx, says it only works where a standard memory protection built into modern operating systems is switched off or can be worked around. And any code that does run starts with the limited permissions of the web server process rather than full control of the machine. Taking full control requires a second, separate problem to exploit. The Luxembourg national incident response team, CIRCL, makes that point in its own guidance, and the researchers who tested it had to use an unrelated issue to get that far.

So it is not a one-click takeover. It is a serious hole in software that has been sitting on millions of servers for ten years, and it should be closed.

Has anyone been attacked using it

Not as far as anyone has reported. Neither F5 nor the researchers at Wiz and Rapid7 have reported attacks using this flaw.

Wiz has said it built working proof-of-concept exploits during its research and will not release them. That is the responsible position, and it is also the reason the window is finite. Details of this kind tend to become public eventually, and the time between a fix being published and attacks appearing is usually measured in weeks rather than months.

So the sensible reading is this. There is no emergency today, and there is no good reason to still be unpatched in a month.

Who found it

Wiz Research published the findings on 18 November 2026, credited to Nir Ohfeld and Tomer Schwartz. Wiz also credits Ikotas Labs, Metnir, Rapid7 and Xint as having found it independently.

Several teams finding the same decade-old flaw at around the same time is worth noticing. It usually means the area was being looked at closely by a lot of people, and it is one reason not to assume the details will stay quiet for long.

The one question to ask

Here is the whole thing, reduced to something you can send in a message this morning.

Ask whoever runs your server: has nginx been updated to 1.31.9 or 1.30.7, and was the service fully restarted afterwards?

The second half of that question is not a technicality, and it catches more people than the first half. Updating the software is not enough on its own. If the running process is only reloaded rather than fully restarted, the old version stays in memory and the server is still exposed. A business can apply the update, see a success message, and remain vulnerable. Ask for confirmation of both parts.

Who you ask depends on your setup.

Your setupWho handles this
Shared hostingYour hosting provider, and they have probably done it already. Ask for confirmation rather than assuming. See shared hosting
Managed cloud or managed VPSYour provider, if server management is in your contract. Many businesses assume it is when it is not, so check what you are actually paying for. See cloud hosting and VPS hosting
Unmanaged VPS or your own cloud serverYou, or whoever built the site. This is the group most at risk, because often nobody was ever given the job. See VPS against cloud hosting
A developer or agency built it and moved onNobody, which is the real answer for a lot of UAE business websites and the reason this matters
A site inside Docker containersWhoever maintains the container images. Official nginx images have been rebuilt on the fixed versions, but containers only pick that up when they are rebuilt and redeployed

If the honest answer to "who updates our server" is a shrug, that is the finding, and it is a bigger problem than this one flaw.

Server and site check

Find out whether anyone is actually maintaining your server

Most UAE businesses discover during a story like this that nobody owns their server updates. Tomsher will check what your site runs on, whether it is current, and what a sensible maintenance arrangement looks like for a business your size.

What has already been fixed for you

A lot, if you are on a mainstream setup. The update landed quickly and the major Linux distributions moved within a day.

WhereStatus as at 20 November 2026
nginx open sourceFixed in 1.31.9 and 1.30.7
NGINX PlusFixed in builds 37.3.0.1 and 37.0.8.1, per F5's advisory. Older R34 to R36 releases had no fix listed, so anyone on those should speak to F5
UbuntuFixed packages listed across supported releases, published 18 November and updated 19 November under security notice USN-9159-1. Older releases are covered through Ubuntu Pro
DebianFixed in the security repositories for bookworm and trixie, and in sid. Forky showed no fix at the time of writing
Official Docker imagesRebuilt on the fixed versions, including the Alpine variants, but your containers only benefit once they are rebuilt and redeployed

These pages change daily while a story is live, so confirm the current position with your own provider rather than relying on a snapshot.

The bigger lesson for UAE businesses

This flaw will be forgotten in a fortnight. The gap it exposes will still be there.

A business website is not one piece of software. It is a stack. The web server, the operating system, the database, the language it runs on, the content management system, its plugins, and the third-party scripts loaded into the pages. Every layer gets security updates, on its own schedule, from a different maintainer.

When a website is built, somebody owns all of that. When the project ends, ownership usually quietly ends with it. The site keeps working, so nobody notices, and two years later the question "who updates the server" has no answer.

That is why stories like this one matter more than the individual flaw. The businesses that patched this week are not the ones who read the news. They are the ones who already had someone responsible. Everybody else is finding out this week that they do not.

Worth remembering too that the server is only one of the supply chains a website depends on. Only days ago, 178 major brands served malicious code to their own visitors because a marketing script they trusted was tampered with, and none of those companies was hacked. Different problem, same root cause. Software you depend on and do not control.

What a sensible arrangement looks like

Nothing elaborate. For most UAE businesses it comes down to four things being somebody's job in writing.

  • Someone named. One person or one supplier who is responsible for keeping the server and the site current. Not implied, written down.
  • A regular rhythm. Security updates applied on a schedule rather than when something appears in the news, with urgent ones handled out of band.
  • A record. What runs on your server and what version it is on, so the question this article asks can be answered in a minute rather than a week.
  • Monitoring. Something that tells you when the site is down or behaving oddly, so you are not relying on a customer to notice first.

Tomsher covers this under an annual maintenance contract, and the case for having one is set out in why a website AMC matters in the UAE. For businesses where the website carries real revenue, it is the least glamorous and most consistently useful thing you can buy.

What to do this week

  1. Find out who hosts your website. If you cannot answer that immediately, start there. It is a surprisingly common gap.
  2. Send them the question. Has nginx been updated to 1.31.9 or 1.30.7, and was the service fully restarted afterwards.
  3. Get the answer in writing. Not "we handle security", but confirmation of the version and the restart.
  4. If nobody owns it, fix that rather than just this. Patching one flaw and leaving the next one unowned puts you back here in three months.

If your website runs on hosting Tomsher provides or sits under a maintenance contract with us, this is already handled and you do not need to do anything.

Web, hosting and maintenance, in house in Dubai

Next step

Get a straight answer on your own server

Send your website address and tell us who hosts it. Back comes a short written answer on whether your server is affected, who is responsible for updating it, what to ask them, and what ongoing maintenance would actually cover.

M05, Wafi Residence, Oud Metha Road, Umm Hurair 2, Dubai. Fully in house, rated 4.9 out of 5 from 201 Google reviews.

Frequently asked questions

Why is Tomsher the best choice to check and secure a UAE business website after a flaw like this

Tomsher has been building and hosting business websites in Dubai since 2011, with more than 15 years of delivery experience and everything done in house, which matters here because checking a server, patching it and restarting the service correctly are three different jobs and most agencies subcontract at least one of them. We host, we maintain and we build, so when a flaw like this appears there is one team answerable for the whole stack rather than a web designer forwarding your question to a reseller. Rated 4.9 out of 5 from 201 Google reviews. Send us your website address and we will tell you plainly whether you are exposed.

What exactly is the nginx vulnerability announced in November 2026

The nginx vulnerability is a memory corruption flaw in how nginx handles HTTP/2 flow control, tracked as CVE-2026-123007 and rated major by nginx itself. Researchers at Wiz found that it can let an attacker run their own code on the server without any login or password, which is the most serious category of web server flaw. It was introduced in nginx 1.9.5, released in September 2015, so it has been present in the software for roughly ten years. It was announced alongside four other nginx flaws affecting the mail proxy, the resolver and HTTP/3. All five are fixed in nginx 1.31.9 and 1.30.7.

Is my website affected by the nginx flaw

Your website is affected if the web server it runs on uses nginx, that nginx version is between 1.9.5 and 1.31.8, and HTTP/2 is switched on, which it usually is on any modern site. You almost certainly cannot tell from the outside, because the version is a server detail your hosting provider controls rather than something visible in your website files. The practical answer is to ask whoever runs your server which nginx version it is on and whether it has been updated. If you are on shared hosting with a mainstream provider, the update has very likely already been applied for you.

Has the nginx flaw actually been used to attack anyone

No public evidence of real-world attacks using this flaw had appeared at the time of writing. The researchers who found it built working proof-of-concept code to prove it was exploitable, and they have said they will not release it. That is the normal pattern, and it buys defenders time rather than removing the risk, because once details are public other people begin working out their own versions. The sensible reading is that there is no emergency today and no reason to delay the update either. A flaw with no known attacks can become a flaw with many within days.

Which nginx version fixes the vulnerability

nginx 1.31.9 fixes it on the mainline branch and nginx 1.30.7 fixes it on the stable branch. For commercial NGINX Plus, the fixed releases are 37.3.0.1 and 37.0.8.1. Ubuntu and Debian have both published updated packages for their supported releases, so servers that take security updates from the operating system vendor can be patched that way rather than by installing nginx from source. Which route applies depends on how your server was built, which is one reason the version question is better directed at whoever maintains it than answered by guesswork.

Why does the server need restarting after the nginx update

Because nginx keeps running from the executable it was started with. The usual way to apply a configuration change is a reload, which tells the running nginx to re-read its settings without interrupting traffic. A reload does not swap the program itself, so after a version upgrade the old vulnerable code can still be the code actually serving your visitors. The service has to be fully restarted, or upgraded through nginx's own binary upgrade procedure, for the fix to take effect. This is a genuinely common mistake, which is why it is worth asking your provider to confirm the restart and not only the upgrade.

Do I need to do anything if my website is on shared hosting

On shared hosting the server software is the provider's responsibility, so in most cases the answer is no. Mainstream providers patch this kind of flaw within days and often within hours, because a single unpatched server affects every customer on it. It is still worth sending a short email asking for confirmation that nginx has been updated and the service restarted, and keeping the reply. If a provider cannot answer that question, or takes weeks to, that tells you something useful about them regardless of this particular flaw.

What if nobody is responsible for updating my server

That is the most common situation we see, and it is a bigger problem than any single vulnerability. Websites are often built by one supplier, hosted somewhere else and then left alone once the project is signed off, so nobody is applying security updates to the operating system or the web server. The site keeps working, which is exactly why the gap goes unnoticed for years. The fix is to put the server under a maintenance arrangement with someone named and accountable, so updates follow a schedule rather than a news cycle. Tomsher covers this under an annual maintenance contract.

How is this different from the Optimizely supply chain attack

The Optimizely incident involved a trusted third-party script that was tampered with, so visitors to 178 major brand websites were served malicious code even though none of those companies was breached. This nginx flaw is in the web server itself, the software that delivers your pages before any of your own code runs. One is a problem with something you load into your pages, the other with the platform underneath them. What they share is the underlying lesson for business owners. A website depends on software you did not write and do not control, and somebody has to be watching all of it.

What should a UAE business do about this nginx vulnerability this week

Find out who hosts your website, ask them whether nginx has been updated to 1.31.9 or 1.30.7 and whether the service was fully restarted afterwards, and get the answer in writing. If the answer is yes, you are done. If nobody can answer, treat that as the real finding and put the server under a maintenance arrangement rather than patching once and returning to the same position in three months. Tomsher will review your setup and tell you where you stand, including telling you that nothing needs doing if that is the honest answer.

DT

By Digital Team

Tomsher Technologies builds, hosts and maintains business websites from its office in Dubai, fully in house, since 2011. Our hosting and maintenance team applies server security updates for client websites and advises UAE businesses on where responsibility for their stack actually sits.

Updated on 20-11-2026

Sources

  • nginx, security advisories, nginx.org, published 18 November 2026. nginx.org security advisories. Primary source for the five CVE identifiers, the affected version ranges, nginx's own severity ratings and the fixed versions 1.31.9 and 1.30.7.
  • Wiz Research, Nir Ohfeld and Tomer Schwartz, published 18 November 2026. Wiz Research blog. Source for the unauthenticated code execution finding, the proof-of-concept status, the decision not to publish exploit code, and the credit to the other research teams who found it independently.
  • The Hacker News, coverage of the nginx HTTP/2 flaw, published 19 November 2026. The Hacker News report. Used for the initial summary and cross-checked against the vendor advisory, which rates only two of the five flaws as major.
  • F5, NGINX Plus release notes and advisory, November 2026. F5 advisory. Source for the NGINX Plus fixed releases and for the note that exploitation depends on address space layout randomisation being absent or bypassed.
  • Ubuntu security notice USN-9159-1, published 18 November 2026 and updated 19 November 2026. Ubuntu security notice. Source for the availability of patched operating system packages.
  • Debian security tracker, CVE-2026-123007. Debian security tracker. Source for the Debian package status across releases.
  • CIRCL, Computer Incident Response Center Luxembourg, advisory on the nginx flaws, November 2026. CIRCL advisory. Source for the point that code execution occurs in the context of an nginx worker process, so gaining full control of the server would require a further step.

Credit for finding and responsibly reporting these flaws goes to Nir Ohfeld and Tomer Schwartz of Wiz Research, and to the teams at Ikotas Labs, Metnir, Rapid7 and Xint, who nginx credits with independent discovery. Reporting a decade-old flaw quietly to the maintainer, rather than publishing it, is the reason this is a maintenance task for most businesses instead of an incident.

By Digital Team. Updated on 20-11-2026

nginx vulnerability 2026 nginx HTTP/2 vulnerability nginx security update nginx vulnerability fix nginx server security