SYNCTECH
Security · · by the Synctech team

NoisyTCP Linux kernel flaw CVE-2026-99288: what server owners do

NoisyTCP Linux kernel flaw CVE-2026-99288: what server owners do

A Linux kernel flaw named NoisyTCP (CVE-2026-99288) lets an attacker on the internet crash a Linux server without logging in. The researcher who found it published the details on 16 November 2026. If your company runs its own Linux server or VPS, ask whoever manages it to install your Linux vendor's kernel update when it is out, then reboot.

What does the flaw do?

The bug is in how the kernel looks up TCP connections while it handles many incoming packets at the same time. A crafted stream of packets can make the kernel read a connection that has just been freed and reused, and the server stops with a kernel crash.

The researcher says it works against ordinary services in their default setup, such as Nginx, Apache and SSH. On virtual servers the crash took between 5 seconds and 2 minutes to trigger in his tests, while a quiet physical server took about 14 hours. When a web server runs in a container, the whole host goes down, along with every other container on it. A second, harder path could lead to memory corruption.

The code has been in the kernel since 2015, so almost every Linux distribution is affected. Red Hat rates it Important, and Ubuntu rates it Medium. As of 17 November 2026, Ubuntu's page still lists its releases as needing evaluation, so updated kernel packages may take some days to arrive.

Who needs to act?

Companies with their own Linux server, cloud VPS or virtual machine that faces the internet should act first. The flaw is in the Linux kernel, so it does not apply to the Windows or macOS system on office PCs and Macs. Some office devices are built on Linux, so it is worth asking their makers whether a firmware update will follow.

What to do

  1. Find out who looks after each Linux server or VPS you own. For a cloud VPS, this may be you, a developer or your IT vendor.
  2. Ask them to watch for the kernel update from the distribution's security page (Ubuntu, Red Hat, AlmaLinux, Debian and others) and install it when it arrives.
  3. Plan a reboot after the update. The new kernel only runs after a restart.
  4. The researcher lists temporary workarounds for servers that cannot be updated quickly. One of them turns off SYN cookies, which weakens protection against flood attacks, so only an experienced administrator should decide whether to use them.
  5. Ask the makers of any Linux-based office device, such as a firewall or storage box, whether it is affected, and apply their update when it is out.

How SyncTech can help

If you would rather not manage a server yourself, our business web hosting plans keep the server updates on our side.

Source and credit

We are sharing this news from the oss-security mailing list, NIST, Red Hat and Canonical. This summary is written by Synctech in our own words.

Related announcements

Questions? Open a ticket or WhatsApp +6016-7618518.