On 8 November 2026, Debian released a fix for three Linux kernel flaws in Debian 13, the current version of the system, nicknamed Trixie. CERT-FR, the incident response centre attached to the French cybersecurity agency, relayed it on 13 November. If your website, your shop or your internal tools run on a Debian 13 server, the update needs to be applied. With one particularity: it only protects you once the server has been restarted.
Debian 13: what the update fixes
The three flaws carry the references CVE-2026-100080, CVE-2026-100081 and CVE-2026-100082. They affect the kernel, the core of the system that manages memory, files and access rights. According to Debian, they can allow an attacker to gain more rights than they should have, to bring the server down or to read information not meant for them.
The fixed kernel version is numbered 6.12.116-1. All earlier versions of Debian 13 are affected. Servers running Debian 12 or another distribution are not covered by this bulletin: they follow their own patch schedule.
Why the restart changes everything

For most software, installing the update is enough: the new version takes over at the next launch, often without anyone noticing. The kernel works differently. It is loaded when the server boots and stays in memory until the next restart.
In practice, your provider may install the patch on Monday while the server keeps running on the old, vulnerable kernel for weeks. On disk, everything is up to date. In memory, nothing has changed. This is a common oversight, because a restart takes the site down for a few minutes and it is tempting to put it off.
These few minutes can be planned: early in the morning, on a Sunday, or outside the hours when your shop receives the most orders. A short planned outage is better than a vulnerable server with no end date.
Three questions to ask your provider
You do not need to get into the technical details. These three questions are enough to tell you where you stand:
- Do our servers run Debian 13? A contract that mentions “Linux hosting” does not say. You need the list of machines and their operating system.
- Has the server restarted since the fix was installed, and which kernel is running today? The expected answer is a number equal to or higher than 6.12.116-1. A technician can check it with one command (
uname -vshows the version of the kernel in use). - Was the site tested after the restart? A server that restarts can come back with a service stopped. Displaying the home page is not enough: the contact form, customer login and payment must also work.
An answer such as “everything is up to date” does not answer any of these questions. Ask for facts: the server name, the kernel version, the restart date. It is the same advice as for the PHP2Shell flaw: the exact version number, in writing.
Maintenance that ends with a check
This patch will not be the last. Debian regularly publishes kernel updates, and each one raises the same question. Good practice fits in one sentence: every intervention ends with a check of the running kernel and a test of the key functions, recorded in a report.
This is how we work in our hosting and managed hosting service: patches installed, restart scheduled with you, then a check before closing the intervention.




