Limits on each distribution

MMWatchdog is tested on the target distributions. Most checks work the same on all of them. This page is for a person who runs one of them: it says what MMWatchdog cannot do there, why, what would change it, and what you can do now. It does not advise which distribution to choose. Each cause is named as it was measured, with the date and the source. A cause on MMWatchdog's own side says so. On ARM64, the tests run on Debian and Ubuntu only so far; the other distributions are tested on AMD64.

The page covers the limits that remain on a distribution. Which check answers which question on each distribution is in Security checks on each distribution.

On every distribution

While a fix has not been published, no check alerts: one example

  • Limited: while a distribution has not published a fix, no check of MMWatchdog alerts. apt_security_updates, dnf_security_updates, and zypper_security_patches stay healthy, because no fixed version waits to be installed. known_exploited_vulnerabilities_in_installed_packages fails only for a flaw on CISA's list of known exploited vulnerabilities.

  • Cause, one dated example: CVE-2026-98432, a flaw in the Linux kernel with a CVSS score of 9.8, which Ubuntu's CVE page rates critical and dates 2026-10-15. Each distribution's own advisory published its fixed kernel on these dates, read on 2026-11-14:

    Distribution Advisory Published
    Debian 13 DSA-6568-1 2026-10-16
    Debian 12 DLA-4841-1 2026-10-17
    Fedora 45 FEDORA-2026-e96bbfedfe (kernel 7.2.11, which contains the upstream fix of 7.2.10) 2026-10-16
    Rocky Linux 9 RLSA-2026:93412 2026-11-03
    AlmaLinux 10 ALSA-2026:93413 2026-11-09
    openSUSE Leap 16.1 openSUSE-SU-2026:22309-1 2026-11-09
    Ubuntu 24.04 and 26.04 USN-9022-1 and USN-9018-1 2026-11-10
    Arch Linux none: the tracker has no entry; Arch's kernel follows upstream releases none

    On 2026-11-14, the flaw was not on CISA's list, so on each distribution no check of MMWatchdog alerted until that distribution's fix was published: 18 days after Debian's on Rocky Linux 9, 24 days on AlmaLinux 10 and openSUSE Leap 16.1, and 25 days on Ubuntu. Canonical's post about the flaw on its own blog says that Ubuntu's fixes waited for its kernel certification, tests "on more than 2,000 certified servers and devices", and were in the -proposed pocket before that. One flaw is one example, not a measure of how fast any distribution ships fixes in general.

  • What would change it: an earlier fix from the distribution, or the flaw joining CISA's list, after which known_exploited_vulnerabilities_in_installed_packages fails while the node is affected, with no fix yet as the reason.

  • What you can do now: read your distribution's security notices for critical flaws, and apply a mitigation while the fix waits; for this flaw, Ubuntu's CVE page gave one.

Debian 12 and 13

No known limit beyond the one on every distribution, as of 2026-11-14.

Ubuntu 24.04 and 26.04

Some security fixes come only with Ubuntu Pro

  • Limited: apt_security_updates sees only the fixes that apt on the node can install. Many security fixes of the packages of Ubuntu's Universe component are in Ubuntu Pro's ESM Apps service only, so a node without Ubuntu Pro never gets them, and apt_security_updates stays healthy.
  • Cause: Ubuntu delivers those fixes through Ubuntu Pro, Canonical's subscription, not through its free archive.
  • What would change it: the fixes reaching Ubuntu's free archive.
  • What you can do now: add ubuntu_pro_security_fixes, which fails while a node misses such a fix, or attach the node to Ubuntu Pro.

A severe fix waited until 2026-11-10: one example

Ubuntu published its fix of CVE-2026-98432 on 2026-11-10, 25 days after Debian's. See While a fix has not been published, no check alerts.

Rocky Linux 9

A severe fix waited until 2026-11-03: one example

Rocky Linux 9 published its fix of CVE-2026-98432 on 2026-11-03, 18 days after Debian's. See While a fix has not been published, no check alerts.

No check for known exploited flaws

  • Limited: known_exploited_vulnerabilities_in_installed_packages is unknown.
  • Cause: the vulnerability data that Rocky Linux publishes through OSV.dev, measured on 2026-11-14 through OSV.dev's API and downloadable files:
    • It holds advisories only: every affected range ends with a fixed version, and no record says "affected, no fix yet". So the check could never report a flaw without a fix, and its findings would be the waiting fixes that dnf_security_updates already alerts on.
    • The data of Rocky Linux:9 had no record of three flaws on CISA's list: CVE-2024-6387 in openssh, CVE-2023-4911 in glibc, and CVE-2024-1086 in kernel. The data of AlmaLinux:9 had a record of each: ALSA-2024:4312, ALSA-2023:5453, and ALSA-2024:2394.
    • Rocky Linux 9.8 and 9.9 share the ecosystem Rocky Linux:9. Advisory RLSA-2026:85958 fixed CVE-2026-59997, CVE-2026-59998, and CVE-2026-60000 in openssh 9.9p1-13.el9_8.rocky.0.1 for Rocky Linux 9.8, and RLSA-2026:97866 lists the same flaws with the 9.9 fix. Asked about openssh-clients 9.9p1-13.el9_8.rocky.0.1, OSV.dev answered that RLSA-2026:97866 affects it, although the 9.8 fix is installed.
  • What would change it: data that records each flaw of each release, also the unfixed ones.
  • What you can do now: let dnf_security_updates alert on every waiting security update, and install the updates soon; keep the node on the current minor release, so that the fixes of the release reach it.

Docker, MySQL 8.4, MongoDB, and pgBackRest are not tested

  • Limited: docker_container, the MySQL 8.4 part of mysql, the official extra module mongodb, and pgbackrest are not tested on this distribution. They read the same programs as elsewhere and are expected to work.
  • Cause: the distribution's repositories have no Docker and no pgBackRest, and the tests run MySQL and MongoDB in Docker.
  • What would change it: tests that install these programs from their makers' own repositories.
  • What you can do now: nothing more than on another distribution; the checks report what they find.

AlmaLinux 10

A severe fix waited until 2026-11-09: one example

AlmaLinux 10 published its fix of CVE-2026-98432 on 2026-11-09, 24 days after Debian's. See While a fix has not been published, no check alerts.

No check for known exploited flaws

  • Limited: known_exploited_vulnerabilities_in_installed_packages is unknown.
  • Cause: the vulnerability data that AlmaLinux publishes through OSV.dev holds advisories only, measured on 2026-11-14: every affected range ends with a fixed version. So the check could never report a flaw without a fix, and its findings would be the waiting fixes that dnf_security_updates already alerts on.
  • What would change it: data that records each flaw of each release, also the unfixed ones.
  • What you can do now: let dnf_security_updates alert on every waiting security update, and install the updates soon.

Docker, MySQL 8.4, MongoDB, and pgBackRest are not tested

The same as on Rocky Linux 9.

Fedora 45

No check for known exploited flaws

  • Limited: known_exploited_vulnerabilities_in_installed_packages is unknown.
  • Cause, on MMWatchdog's side: the check is unknown on the whole Red Hat family, whose measured data is described under Rocky Linux 9. Fedora's own data was not measured.
  • What would change it: a measurement that shows Fedora's data records each flaw of each release, also the unfixed ones.
  • What you can do now: let dnf_security_updates alert on every waiting security update, and install the updates soon.

pgBackRest is not tested

  • Limited: pgbackrest is not tested on this distribution, and is expected to work.
  • Cause: Fedora's repositories have no pgBackRest.
  • What would change it: a test that installs pgBackRest from its makers' repository.
  • What you can do now: nothing more than on another distribution.

Arch Linux

No security-update check

  • Limited: MMWatchdog has no check of waiting security updates on Arch Linux, and known_exploited_vulnerabilities_in_installed_packages is unknown.
  • Cause: pacman has no security data to read. Measured on 2026-11-14: pacman's package databases, such as core.db, hold for each package its name, version, description, sizes, checksums, signature, and dependencies, and no field about security; pacman has no option about security. Arch Linux's security data lives apart, in the Arch Security Team's tracker at security.archlinux.org, which the tool arch-audit reads. The tracker's groups referenced 2,789 CVEs from 2021, but 45 from 2023, 13 from 2024, 79 from 2025, and 5 from 2026, so a check built on it would miss most recent flaws.
  • What would change it: pacman, or a maintained Arch Linux source, recording the flaws of current packages.
  • What you can do now: upgrade often, with pacman -Syu: a rolling release ships its fixes as upgrades of the whole system. If you want the tracker's signal anyway, run arch-audit -u, which lists the installed packages that have a tracked flaw and an upgrade, from a shell check.

No reboot check

  • Limited: reboot_after_updates is unknown.
  • Cause, on MMWatchdog's side: MMWatchdog knows how Debian, Ubuntu, the Red Hat family, and SUSE say that a node needs a reboot, and has no reader for Arch Linux.
  • What would change it: a reader for a signal that Arch Linux gives.
  • What you can do now: reboot after an upgrade that brings a new kernel.

distribution_release_security_support is unknown too, which is no limit: a rolling release has no end of support.

openSUSE Leap 16.1

A severe fix waited until 2026-11-09: one example

openSUSE Leap 16.1 published its fix of CVE-2026-98432 on 2026-11-09, 24 days after Debian's. See While a fix has not been published, no check alerts.

No check for known exploited flaws

  • Limited: known_exploited_vulnerabilities_in_installed_packages is unknown.
  • Cause: the vulnerability data that openSUSE publishes through OSV.dev, measured on 2026-11-14: the ecosystem openSUSE:Leap 16.1 held 3 advisories, while openSUSE:Leap 16.0 held 1,242. The data of 16.0 does not fit 16.1, whose packages are other builds. It also holds advisories only: every affected range ends with a fixed version.
  • What would change it: OSV.dev's data for Leap 16.1 filling in, or a reader for SUSE's own OVAL files at ftp.suse.com/pub/projects/security/oval/, which cover Leap 16.1, also the flaws without a fix, in another format and about 10 MB to download for each run.
  • What you can do now: let zypper_security_patches alert on every waiting security patch, and install the patches soon.

pgBackRest with PostgreSQL 18 is not tested

  • Limited: pgbackrest is not tested on this distribution.
  • Cause: openSUSE Leap 16.1's pgBackRest 2.53.1 does not support its PostgreSQL 18, measured on 2026-11-14 in the test.
  • What would change it: a newer pgBackRest in openSUSE Leap.
  • What you can do now: use a pgBackRest that supports your PostgreSQL.

MariaDB's client against MySQL 8.4 is not tested

  • Limited: a mysql check of a MySQL 8.4 server through MariaDB's client is not tested; MySQL's own client is.
  • Cause: openSUSE Leap 16.1's MariaDB client 12.3 lacks the plugin caching_sha2_password, which MySQL 8.4 logins use, measured on 2026-11-14 in the test.
  • What would change it: that plugin in openSUSE's MariaDB client.
  • What you can do now: install MySQL's own client on nodes that check a MySQL 8.4 server.

This page is docs/reference/limits-on-each-distribution.md in the MMWatchdog repository.