Key Article Takeaways
- Patched isn’t the same as rebooted. If freebsd-version -kru shows different versions, a fixed kernel is sitting on disk while the old one is still running.
- Write severity tiers before you need them. When the debate happens during an incident, it usually ends with “later.”
- Measure time from advisory to patched-and-rebooted, for each tier. Auditors, customers and cyber-insurers increasingly ask for this number.
- EoL doesn’t flag unpatched CVEs; it hides them. After support ends, scanners only say “EoL operating system.” When an upgrade can’t land on the project’s schedule, LTS keeps your patch process running until it can.
Improve the way you make use of FreeBSD in your company.
Find out more about what makes us the most reliable FreeBSD development organization, and take the next step in engaging with us for your next FreeBSD project.
FreeBSD Support FreeBSD DevelopmentAdditional Resources
Here are more interesting articles on FreeBSD that you may find useful:
- FreeBSD: Why Enterprise FreeBSD Deployments Need Long-Term Support
- FreeBSD Release Cycles and Enterprise Requirements: Why LTS Matters
- FreeBSD: The Hidden Cost of Upgrading in Production
- Why Is Everyone Stuck on FreeBSD 13?
- FreeBSD’s Release Cycle: What It Means for Your Upgrade Schedule
A CVE lands in your inbox with your platform's name in it, and the clock starts. Are we affected? On which systems? Is there a fix for the releases we actually run? How fast can it be deployed without breaking anything? That is a lot for a Tuesday morning. Patch cadence is the process that answers those questions on a schedule. This article covers what to check, what the project has changed to publish vulnerability data faster, how to build a patch process on top of it, and where that process breaks down.
Start With What You Can Check Today
FreeBSD splits security response into two channels, and you need to plan for both. The base system is covered by Security Advisories (SAs) and Errata Notices (ENs) from the FreeBSD Security Team. The third-party packages are covered by the ports vulnerability database. The commands to be monitored:
$ freebsd-version -kru # installed kernel, running kernel, userland
$ freebsd-update fetch # pending base system patches
$ pkg audit -F # fetch vuln database, audit installed packages
The first command matters more than it looks like: a patched kernel that hasn't been booted is an unpatched kernel; if the three values are different, you need to fix that problem. Subscribe to the security-notifications mailing list so advisories arrive instead of being discovered, and put pkg audit -F in your monitoring rather than in a human's memory. That way, a host that starts reporting a vulnerable package is an alert, not a manual discovery.
Make sure you understand what an advisory is telling you. An SA names the affected releases, the corrected versions, any workaround, and the patch. On the other hand, an EN covers serious non-security defects through the same machinery. If you run jails, remember the shape of the system: every jail shares the host kernel, so a kernel SA is patched once on the host, while each jail's base system and package set need their own monitoring in the loop.
One property of this whole system needs to be clearly understood: advisories and patches are only issued for supported releases. Keep that in mind; we'll come back to it.
What FreeBSD Is Building to Publish Faster
FreeBSD is in the process of modernizing how vulnerability data is published, so systems can consume it as easily as humans. The legacy VuXML database that pkg audit reads is going to be converted to OSV, the Open Source Vulnerabilities standard used across the open source ecosystem. FreeBSD also plans to support OSV JSON in pkg(8) itself. These tooling updates will allow FreeBSD to pull vulnerability data from the global databases, including the NIST CPE dictionary. Alongside it, the project is building SBOM generation into the toolchain, with SPDX-compatible output for the base system.
Part of this work is commissioned through the Sovereign Tech Agency, which is investing over €600,000 in FreeBSD infrastructure, security, and regulatory compliance, and the FreeBSD Foundation's security engineering work runs in parallel with that, including standards engagement on the EU Cyber Resilience Act.
For an operator, this is a concrete improvement. Machine-readable advisories in a cross-ecosystem format mean your vulnerability scanners, dashboards, and compliance tooling can treat FreeBSD like a first-class citizen next to the rest of your fleet, and CVE-to-package matching gets faster and requires less manual effort on both the publishing side and the system administration side.
Building the Patch Process
Vulnerability feeds alone are not a process. Best practices define five key components to a security patch cadence:
Inventory first. You cannot patch what you don't know is there. An accurate list of systems, what releases and installed package sets they have, and when their EoLs are, is the foundation you need to start from. Quality SBOM tooling makes it easier to keep an accurate catalogue.
Automate detections. Get your advisories by mail or RSS feed, ensure pkg audit is in your monitoring, and the latest OSV feeds into whatever scanner you already rely on. The goal is that "are we affected?" is answered by automation within hours of publication, rather than by your team scrambling to find the answer.
Tier the response. Not every CVE deserves an emergency window; a remotely exploitable issue in a listening service is a problem for right now. Whereas a local denial-of-service in a component you don't ship is a problem for the next release window. Write the tiers down before the incident, have a rubric your team can understand and stick to when the pressure is on, because severity debates held during the incident always resolve to "later."
Stage the rollout. Patches go to a canary system before the rest of the fleet, because a bad patch deployed everywhere at once is a fleet-wide outage. Keep the canary window short and well-defined: hours for the emergency tier, a day or two otherwise, so staging doesn't become deferral. Make sure you have enough time to get useful feedback before the rest of the rollout. Finally, make sure you close the loop with the reboot and verify with freebsd-version -kru to show disagreeing versions.
Measure it. Track time from advisory publication to patched-and-rebooted, per tier. Auditors, customers, and cyber-insurance questionnaires increasingly ask for this number, and a measured value beats an estimate. Security patches are going to be a consistent part of our lives for the foreseeable future. They are no longer an unexpected event.
Where the Process Breaks
Now back to the property from the first section. Every piece of the stack in this article: the advisories, freebsd-update, the OSV feeds, the fix behind every alert, all have the same precondition: a supported release. When a release reaches end-of-life, the advisories stop, the patches stop, ports and packages move on without you or disappear entirely. pkg audit will continue to reporting vulnerabilities that no longer have fixes. The process doesn't degrade at EoL; it terminates entirely.
EoL dates arrive faster than your product schedules move: a minor release goes EoL just three months after its successor ships, which typically happens every 6 months. FreeBSD 14.4 ends on December 31, 2026, and 15.1 follows on March 31, 2027. We covered why certified appliances and long-lived deployments can't always take that upgrade on schedule in our enterprise LTS article. The security consequences of that article are the point here. The gap between "the release we run" and "the releases that get advisories" is not a compliance abstraction. It is a growing list of specific, named, unpatched CVEs on production systems. The worst part of this? Your security scanner likely won’t flag that your EoL release is not patched against the latest CVE; it will just say “EoL operating system”.
Extending Security Patching Beyond End-of-Life
Klara’s LTS for FreeBSD keeps the security and support process running past the project's EOL dates. Security Advisories and Errata Notices continue for the releases your fleet is running on. Each advisory is triaged against your installed base and workload, so "are we affected?" arrives answered and scored to your environment; fixes are backported by engineers who work on FreeBSD as authors, and changes ship with SBOM-backed documentation that drops straight into the compliance reporting above. Eligible bug fixes go upstream, keeping you aligned with the project rather than drifting onto a private fork, and every engagement carries a migration path engineered in parallel, so LTS is the bridge to your next release, not a permanent detour.
Getting started uses the inventory you already built: the list of systems, the releases they run, and each release's EoL date next to the product lifecycle sitting on top of them. Where those dates disagree, your patch process has an expiration date. Bring that list to us at klarasystems.com, and keep the loop running on every system you own, not just the recently installed ones.





