The October Inflection Point: Repairing a Crack in the Web’s Foundation

"PHP Vulnerability" text on PHP code script image.

Summary

The Internet community recently mobilized on short notice to apply an emergency security update that patched a critical vulnerability in the PHP programming language. That vulnerability threatened to leave 70% of the web at risk of takeover by outside attackers. This unusual event was notable for the scale of its impact, its urgent timing, and the publicity surrounding it. 

This article is intended to provide some background information and explain the steps that we at Network Services Group took to protect our hosting infrastructure and the web presence of our clients, while looking more closely at the changing Internet security landscape. 

What is PHP? 

PHP is a programming language that has been around since the mid 1990’s. It was designed to allow developers to add dynamic behavior to what were largely static websites of the era. Instead of just simple document retrieval, websites were now able to evolve into full web applications that could process information on the server — responding to user input, interacting with back-end databases, and generating content on demand. 

It wasn’t the first technology to allow that. I remember well building early web applications using technologies like Perl and C-based CGI scripts, ColdFusion, AOLserver, and the like. But those were clunky, proprietary, or experimental technologies with relatively limited adoption.

When PHP hit the scene, it offered a fully open-source language that integrated well with established open-source web servers and databases, and made developing complex web applications much easier than the other options at the time. 

Because of that, its popularity exploded right at the time that broadband Internet and e-commerce were really starting to take off. Countless applications and platforms were built on it as the Internet grew, to the point that the LAMP stack (Linux, Apache, MySQL, PHP) became the de facto hosting and development platform for the fast-growing web. 

The language at the time was relatively simple and lacked some advanced programming constructs, but it grew up with the web and eventually became much more feature-rich and formalized, allowing ever more sophisticated software to be built on top of it, both on and off the web. 

There are many other ways to develop web-based software these days, so PHP and the LAMP stack are nowhere near as ubiquitous as they used to be. But it had such a head start, and so much of the Internet has been built on it, that it’s still in extremely widespread use. Some estimates suggest that 70% of the web currently runs on top of PHP. 

At Network Services Group, we have been making extensive use of PHP for hosting web and Internet technologies since our founding in 2001. Web sites and web applications that we develop for our clients run on our NSG web content manager, which is powered by PHP. The NSG Mail system software that powers our email hosting platform is largely PHP-based, as is our DNS management system and several other key infrastructure components. 

So when rumors started spreading about a possible major vulnerability, I paid attention. 

The Announcement 

On Sunday, two days before the official announcement, word started rapidly spreading on discussion groups and mailing lists that a major vulnerability had been discovered, that emergency updates were being prepared, and that a video claiming proof of a working exploit had already been posted. 

An announcement soon appeared on the home page of the PHP project. It suggested that: 

  • A serious remote-code execution vulnerability was discovered that affected all PHP versions since 2012. 
  • A security update would be released on Tuesday at exactly 10:00 am. 
  • All administrators should be prepared to update any public-facing servers immediately at that time. 

The unusual urgency of the announcement and the warning about updating immediately suggested that this was so serious that, once the update was released, attackers would be able to figure out how to compromise any public PHP server within minutes. 

Soon after, notices started rolling in from other official sources, including government cybersecurity agencies around the world, and news articles about the situation started spreading into the mainstream. Many organizations preemptively took non-critical websites offline as a precaution. Without more technical details available, we had to take the warnings seriously and assume that this was indeed a major problem just waiting to be taken advantage of.

Our Response 

At that point, with a little more than a day to prepare, I ran a full audit of all of our servers and all of our customers’ servers. The goal was to identify which of them were serving PHP web applications and to come up with a plan to patch them as soon as possible.

Fortunately, our existing security policies already had us in a good position. We make a point of keeping all of our own servers running on current versions of Debian Linux. The same goes for customer-owned Linux servers that we help manage. All of them undergo regular security reviews and updates, so they were already all as up-to-date as could be expected. 

That means that, for all of those systems, we would be able to update them through official channels as soon as the Debian security team released updated packages. The Debian security team is among the best, fastest, and most transparent in the industry, and I was confident that they would have updated PHP packages available immediately following the disclosure embargo time. 

Still, I went ahead and applied all other pending updates to servers so that, when this PHP update dropped, we would be able to quickly install it without having to spend time downloading and installing other packages at the same time. And I did shut down some web applications that I knew didn’t really need to be available for the next couple of days.

Early on Tuesday morning, with an update plan in place, I kept an eye on security mailing lists and discussion forums for any more information that became available. 

At 9:45 am, 15 minutes before the official announcement, the Debian security team posted updated PHP packages and sent an announcement to their security mailing list. 

With all of our preparation work in place, and with Debian’s slightly early release of the updated packages, I'm happy to report that we were able to completely update our approximately 40 production servers before the exploit and patches were even made public, with effectively zero interruption or downtime. Along with other services, this meant that all of the modern websites we host were protected immediately. 

Plus, by lunchtime, we were able to update all of the customer-owned web, mail, and VPN servers that we support as well. 

The only major issue remaining was that we still host a handful of legacy customer sites, many inherited from other developers, that have not been updated in so long that they are unable to run on modern versions of PHP. 

This is certainly not ideal, and we had already taken steps to further isolate those legacy sites by running them and their old PHP versions in standalone containers that would not limit our ability to keep our main production environments up to date. 

The unsupported versions of PHP running in those containers are so old that the PHP organization has no plans to update them, even though they are also vulnerable to this new exploit. 

That left those particular sites in a serious pickle — either stay online with a vulnerable PHP version, or go offline until clients agreed to redevelop or replace them in a way that would be compatible with modern PHP. 

For the sake of security, I made the call just before 10:00 to put those sites in maintenance mode and temporarily disconnect them from the Internet. My hope was that, once more information about the vulnerability was made public, we could find a way to mitigate the danger of those old PHP versions. I started looking through the newly-released code to see if there was any feasible way to bring those old sites back online without putting them at serious risk. 

Fortunately, before I had to dig too far, one of the core Debian and PHP developers took the initiative to backport the fixes to several older PHP versions and release unofficial updates on his own. The resulting packages are not official PHP releases, but they were made and released by a long-time trusted maintainer from the community, so I consider them the next best thing. 

I rebuilt all of our legacy hosting containers to use those unofficial PHP updates, and with that I was able to relaunch those legacy sites. All of them were safely brought back online by mid-afternoon. 

Aftermath 

While the vulnerabilities in this case turned out not to be as easily exploitable as the warnings implied, they are not just theoretical. While there has not yet been a widespread takeover of unpatched servers, researchers were able to use the vulnerability to achieve remote code execution on one of them a few days later and prove that it was exploitable in practice. 

The U.S. Cybersecurity and Infrastructure Agency has confirmed that exploits have been observed on the open Internet and added them to their Known Exploited Vulnerabilities catalog as well. 

Takeaways 

There's a reason we approach security the way we do. 

When you run servers at our scale, it's not enough to just get them working for the short term. They need to be managed in a way that optimizes for standardization and long-term maintainability. In our case, that includes: 

  • Building on a trusted platform. We run Debian Linux on all of our servers. As a non-profit, open-source project that provides the foundation for a large number of other derivative Linux distributions, it is well known for its technical quality and stability. And it avoids the profit motives that lead commercial operating system vendors to make decisions that benefit themselves over the needs of their customers. 

Plus, its security team has proven time and time again to be among the best, fastest to respond, and transparent in the entire industry. 

  • Keeping things current. The longer you let a system fall behind, the more difficult it is to make changes and updates when necessary. We make a point of running regular security reviews and updates on all servers, and updating them to new major operating system versions in a timely manner. 
  • Standardizing hosting environments. It is important that configuration across servers is consistent and predictable as much as possible. If hosting systems are configured manually, they can diverge over time, and surprises can creep in. Keeping them standardized means each individual server is less unique, and we can be far more nimble about moving workloads, recovering from failures, and keeping things up to date. 

We manage this in large part by building our hosting environments and applications into installable Debian packages. That means that instead of manually configuring all the software components of a new server, we can run a single command like apt install nsghosting or apt install mailsystem and get a well-tested environment ready to go. Those environments then keep themselves updated and consistent across the fleet as part of our regular security review process. 

All of this means that we were able to quickly and confidently update all of our servers and know exactly what to expect. 

Web sites have a lifecycle. You should plan for that. 

The extra workarounds to keep those legacy sites running just make it clearer that web sites and web applications have a lifecycle of their own and should not be considered “one and done” projects. 

By necessity, a web site operates inside of a constantly changing ecosystem. Programming languages used by sites evolve. Supporting components or services release new versions or even get abandoned by their authors. New types of external threats are discovered. Underlying web protocols change. New browser features gain adoption. Best practices evolve in response to the rest of the Internet environment. 

But too often we see web development companies treat web sites as a single deliverable with no plan in place to manage security updates or respond to changing conditions or technologies. In the best case, that means that you're missing out on better ways to do things and your site becomes stale and dated. In the worst case, it becomes a security nightmare as a site relies on technology that is no longer supported or updated against new threats. 

When we develop sites, we build them with current best practices on currently supported technologies. But we all know that those things change over time. If you have a website, you should plan to budget for ongoing review and development as part of its lifecycle. We can assist with annual reviews and recommendations for things that may need to be brought “up to code” so that you don't end up with a legacy site that only keeps running with band-aids, workarounds, and hopes. 

The security landscape is changing. 

This long-standing PHP vulnerability was discovered using AI-based tools that analyze software in ways that were not feasible until recently. 

And that's not just a PHP problem. Software projects across the board are scrambling to respond to a huge influx of AI-assisted bug reports in recent months. This includes the Linux kernel itself, which has been fielding an extraordinary number of AI-assisted bug reports and releasing updates at a rapid-fire rate. 

Many of the issues being discovered this way are extremely obscure and can only be exploited in wildly improbable conditions. But some, as we've seen, do have the potential for wider concern. And the increased load being placed on volunteer software developers is becoming a real issue. 

October 2026 seems to have become a bit of an inflection point. There have been a surprising number of high-profile vulnerabilities discovered this month alone, and the industry is rushing to adapt. 

As hosting providers, this is prompting us to adjust our practices as well. It means less tolerance for allowing sites to run older code. It means reviewing hosting platforms and finding ways to further limit the exposure of even well-trusted software to the public Internet. 

It also means we will probably have more conversations with clients about the importance of keeping up with their website lifecycles. 

Network Services Group Can Help. 

We occupy a sweet spot in the hosting and development landscape for small and medium businesses, right between self-hosting all on your own and trading your digital autonomy to big-tech corporations that don’t even know who you are. We bring decades of experience and expertise to make sure things are done right, yet are local enough to know our customers and work closely with them to safely accomplish their goals. 

If you need a website or application developed, please feel free to reach out so we can discuss the right approach for your organization. Especially if you're our neighbor here in the Southeast Michigan area. 

If you are an existing client with a website that hasn't been updated in way too long, we'd like to hear from you too. We can help get you back on a solid, modern foundation and plan your online presence for the years ahead. 

Stay safe, everyone, and may your experiences with technology be positive and productive. 

Autumn Mahoney 

Director of Hosting & Development