Skip to main content
Back to blog

Nginx and PHP Security Alerts: Two in Two Days, Both Dealt With

Technology In Healthcare · 7 min read · Paul Chapman
Updated. Verified. Taken care of. A glass shield illustrates the completed website security updates.

Two security alerts about the software that sits underneath websites arrived on consecutive days this week. If one of them has reached your inbox, the questions are the usual ones: does this affect our website, has it been fixed, and do we need to do anything?

For practices on our platform, the short answers are yes, yes and no. Here is what happened, what our team did about it, and why you did not need to hear from us first.

Two alerts, two different pieces of software

Neither alert was about a practice website as such. Both concerned software that sits underneath a very large share of the world’s websites, whoever built them.

WEDNESDAY 18 NOVEMBER

Nginx, the web server.

Nginx is the software that receives a visitor’s request and hands back the page. Researchers disclosed five flaws in it, and fixed versions were released the same day.

THURSDAY 19 NOVEMBER

PHP, the language behind the pages.

PHP is the software that builds each page before it is sent. The PHP project released new versions fixing a batch of vulnerabilities, seven of them rated critical.

The Nginx flaws

The Nginx issues have been given the name “NGINX2Root”. The most serious, CVE-2026-123007, could let an attacker run their own code on a server without needing a login. It sits in the part of Nginx that handles HTTP/2, the modern way browsers talk to websites, and the faulty code has been there since Nginx first supported HTTP/2 in 2015. A working demonstration of the attack was published by security researchers within a day of the announcement.

NHS England issued cyber alert CC-4897 on 19 November, asking affected organisations to apply the updates as soon as possible. The fixes themselves are listed in the official Nginx security advisories.

One detail made this alert harder than most. According to the researchers who found it, the attack travels inside the encrypted connection and does not appear in a server’s normal visitor logs. It is not something you can reliably filter out or watch for. The dependable answer is to install the fixed version.

The PHP release

If this sounds familiar, it is. We wrote about the PHP security alert in October, and this was a second PHP security release within a month. The PHP team announced it a day ahead, then published the fixed versions on 19 November. NHS England followed with cyber alert CC-4900 the same afternoon.

The most talked-about fix, CVE-2026-129289, is another flaw that could allow an attacker to run code on a server without logging in. It only applies to servers set up in a particular way, and the PHP team itself noted that the critical issues need unusual or non-default configurations. That is reassuring, but it is not a reason to wait. A fix that is installed protects you whatever the configuration turns out to be.

Why does this matter for a practice website?

Your website is where patients look for opening hours, appointment routes, prescription requests and urgent notices. If the server underneath it could be taken over by a stranger, the pages your patients trust could be changed, or the site could be taken offline at the moment people need it.

That is the worst case, not a description of what happened. A vulnerability is a weakness that could be used. An alert is not proof that anyone has used it. The job is to close the weakness before that question ever arises.

What our team did

Our hosting does not update itself. When a fix like this is released, somebody has to install it on each server, restart the service and check that the new version is the one actually answering requests. That somebody is our Chief Technology Officer, Sebastian Sulinski, and our technical team.

  1. 18 NOVEMBER · NGINX

    Five flaws disclosed. Fixes released the same day.

    The fixed version of Nginx became available, and went straight onto our list.

  2. 19 NOVEMBER · PHP

    A second alert, one day later.

    The PHP release landed in the afternoon. The fixed version of PHP joined the same programme of work.

  3. THE SAME WEEK · COMPLETE

    Both updates in place across our servers.

    Seb and the team brought every server that runs our practice websites up to the fixed versions, with no action needed from any practice.

Two things worked in our favour. The first is habit: we follow the security announcements for the software we run, rather than waiting for a practice to forward an NHS alert.

The second is how our hosting is arranged. Websites that sit behind Cloudflare’s network were not exposed to the Nginx attack at all, because a visitor’s connection ends at Cloudflare and never reaches our server directly. We did not treat that as a reason to leave anything unpatched. A layer of protection in front of a server is welcome. It is not a substitute for fixing the server.

What does your practice need to do?

If your website is on our Core platform: nothing. There is nothing to install, no setting to change and no need to raise a ticket. The updates were applied at server level, so they cover every practice website we host.

If your IT team, ICB or health board asks for supplier confirmation against either NHS alert, send the request to our support team. We will answer the actual question so that you do not have to interpret a technical advisory.

Our confirmation covers the websites we host. It does not cover other systems your practice uses, such as clinical systems or services from other suppliers. Your IT team can check those against the two NHS alerts.

With another website provider?

Ask the same four questions. Twice.

Was our service affected? Has the fix been applied, and when? Has it been verified on the running service? Is there anything our practice needs to do? Ask them once for Nginx and once for PHP, and ask for the answers in writing.

See the four questions in full

Custom-built does not mean exempt

We have never used WordPress, and we say so often. That removes a whole category of risk: there are no third-party plugins or themes on your website waiting for somebody to update them.

It did not exempt us from this week. Nginx and PHP sit beneath almost every website, ours included. What a custom platform changes is who has to act, and how quickly they can. There is one platform, run by the people who built it, on servers we manage ourselves. When a fix is released, there is no queue of plugin authors to wait for and no question about whose job it is.

Two alerts in two days, and you did not have to chase

Weeks like this are a fair test of a website provider. Not because anything went wrong, but because the work that keeps a website safe is mostly invisible until an alert makes it visible. Seb and the team spent this week doing that work so that practice managers could spend theirs on patients.

If you would like to know more about how your practice website is looked after, read how we approach website security, or get in touch with our team.

Information checked: 20 November 2026. This article covers the Nginx vulnerabilities disclosed on 18 November 2026 (CVE-2026-123007 and four related issues), the PHP security release of 19 November 2026 (including CVE-2026-129289), and our Core website platform.

Reader responses

Have your say

Every response is read by our team, and the best are published below. Your email is optional and never shown.