On 21 October, two critical PHP vulnerabilities (CVE-2026-109618 and CVE-2026-109619) allowed attackers to execute code on affected servers without logging in. We proactively took client websites offline, patched them, and had every website back online by 20:03 Belgian time that same evening. Our investigation afterwards covered more than 750 website environments. In the evidence we examined, we found no sign of a successful intrusion linked to this vulnerability, and no indication that personal data was accessed.
We have shared how we handled the incident and what our investigation found. Today, we’ll look at the layers behind that response, including one that runs on every server we manage: regular, automated vulnerability scanning.
The layers, from the outside in
Our default setup follows the principle of defence in depth. If one layer is compromised, the next one is still there.
| Layer | What it does |
| Edge (Dropsolid Edge Shield, add-on) | DDoS protection, web application firewall, CDN and bot control. Screens traffic before it reaches your application. |
| Infrastructure | A load balancer that only allows web traffic on ports 80 and 443, blocking disruptive Layer 3 and Layer 4 attacks by default. A network firewall, and proxy servers that block malicious traffic before it reaches the protected infrastructure. |
| Isolation | Each production environment runs on its own dedicated virtual machine, provisioned for a single client, within its own shielded virtual network. |
| Host | A host-based firewall, an intrusion prevention system that blocks suspicious IP addresses, the ClamAV antivirus engine, and vulnerability scanning by Aikido. Plus hardening best practices. |
| Patching | Each patch is evaluated for urgency and impact. Patches considered highly critical are deployed within one day. |
Every layer is actively monitored and adjusted as needed.
How the layers worked together in October
The PHP incident put several of these layers to work at once.
- Early warning: Aikido Security, our cybersecurity partner, gave us early warnings and access to advanced security tools.
- The edge: For clients with Edge Shield, we activated additional security rules and assessed whether they gave sufficient protection. At a certain point, it became clear that edge rules alone would not be enough.
- The decision: Following CERT-EU guidance to isolate publicly accessible applications until they were patched, we put websites into maintenance mode, including development and test environments.
- The patch: We tested and rolled out the official PHP patch, then brought websites back online step by step. Central management of all environments meant we could apply measures consistently, also outside production.
- The investigation: We examined traffic logs, server activity and files, scanned for malware, and compared our records against known attack indicators, looking back several months wherever records were available. A separate review checked the conclusions.
We found one attack attempt that took place before the patch was in place. Our analysis indicates that it did not succeed. No investigation can offer absolute certainty, so monitoring continues, and we will inform customers if new information affects them.
This is where vulnerability scanning fits in. A known weakness in PHP, the operating system or a container image is exactly what it looks for. It gives us regular, automated visibility on that layer for every server we manage, and its findings are evaluated and patched under our patch policy.
The layer no tool replaces: responsibility
Tools find issues and block traffic. People make the calls.
In October, our CPO Niels Aers and our Technical Director Dieter Blomme made the early decision to take sites offline and took responsibility when that decision was difficult. Our Operations Manager Laurence De Medts kept customers informed and connected throughout. The rest of the team patched, tested, investigated and kept customers up to date.
During an incident, customers should not have to coordinate an agency, a hosting company and a security investigation themselves. Because we bring together application expertise, our own platform and specialist security partners, we could defend proactively, investigate afterwards, and give customers one accountable team throughout. That means acting quickly, investigating thoroughly and being open about the findings.
The same care for your marketing data
Vulnerability scanning is not limited to websites. Your managed Mautic server is regularly scanned for known vulnerabilities in operating system packages, runtimes and container images too, on Essential, Pro and Enterprise.
That matters, because a marketing automation platform holds your contact lists. On top of scanning, every managed Mautic tier includes:
- Hosting in the EEA
- Major and minor updates, with your server kept patched
- A 7/4/12 backup retention schedule (7 days, 4 weeks, 12 months)
- Virus scanning
If you host Mautic Enterprise on your own server, that server is not covered by our scanning.
Layers, and one team behind them
The PHP incident did not have a single hero tool. It had an edge where we could add protective rules and gather detailed logs, a platform that let us act across every environment, a patch process that worked within a day, partners who warned us early, and people who made a hard call.
We know the downtime had an impact on our customers and their visitors, and we thank them for their patience and trust. Regular, automated vulnerability scanning on every server we manage is one more layer in that setup.
Want to see how your platform is protected? Read our Security documentation, or get in touch with our experts.
If your websites are hosted elsewhere, we're happy to show you what this looks like on the Dropsolid Platform.