Security alert · High
NoisyTCP: Linux kernel TCP flaw lets remote attackers crash servers
· 3 minutes read · Centoculus Security Team
Known since
Affected: Linux kernel TCP stack (flaw about 11 years old), nearly all distributions; affected versions not given in the source
What happened
On 16 November 2026, details of a Linux kernel flaw called NoisyTCP were made public. According to the report, the bug is in the kernel's TCP implementation and has been there for about 11 years. Attackers can use it to crash systems that run common network services such as Nginx, Apache and SSH in their default configurations. A crash takes the whole server offline, not just one service.
The report we have does not include a CVE ID, a severity score or a list of affected or fixed kernel versions. We have no information that the flaw is being exploited in attacks. CISA does not list it as exploited.
Who is affected
- The report says the flaw affects nearly all Linux distributions, because it is in the kernel's TCP code. We have no list of the distributions that are or are not affected.
- Servers that accept TCP connections from the network are at risk. The report names web servers running Nginx or Apache and servers with SSH open, all in their default configurations. That covers most web and hosting servers.
- The impact described is a crash (denial of service). The report does not mention data theft or code execution.
The kernel version your distribution ships decides whether you are affected. Your distribution's security advisory is the reliable source for this.
What to do now
- Check your distribution's security tracker or mailing list (Debian, Ubuntu, RHEL/AlmaLinux/Rocky, SUSE and others) for a kernel update that fixes NoisyTCP. Once a CVE ID is assigned, you can search for it there.
- When a fixed kernel is available, install it with your package manager, for example
apt update && apt upgradeordnf update kernel*. - Reboot the server so it runs the new kernel. Installing the package does not replace the kernel that is already running. If you use live patching (for example Canonical Livepatch, kpatch or KernelCare), check whether a live patch for this flaw has been published.
- After the reboot, check that the server is running the new kernel and that Nginx, Apache, SSH and your other services have started.
- Until you can update, limit which hosts can reach services that do not need to be public, such as SSH. You can do this with your firewall or your provider's network filters. Public web servers stay exposed, so the kernel update is the real fix.
- Keep your server monitoring on, so you are alerted quickly if a server stops responding or reboots unexpectedly.
How to check whether you are affected
- Run
uname -rto see which kernel is running. Compare it with the fixed version in your distribution's advisory once one is published. - Compare the running kernel with the installed kernel packages, for example with
dpkg -l | grep linux-imageorrpm -q kernel. If a newer kernel is installed but not running, the server still needs a reboot. - To look for unexpected crashes or reboots, check
last -x rebootand the kernel logs from the previous boot, for examplejournalctl -k -b -1. Look for kernel panics or oops messages.
Sources
More on Linux kernel
If you run Linux servers, also check
Related posts
Alerts ·
WordPress plugin vulnerabilities: 8 plugins, including WPvivid Backup
Vulnerabilities in eight WordPress plugins were disclosed on 20 November 2026, among them WPvivid Backup, NextGEN Gallery and Optimole. The most serious, in Mang Board, lets unauthenticated visitors delete board posts.
Alerts ·
Drupal core: critical installer flaw can allow site takeover without login
Drupal fixed a critical flaw in the built-in installer (install.php). Under specific conditions it lets attackers without a login reinstall a site and take it over. Update to 10.6.22, 11.3.22 or 11.4.13, and block the installer.
Alerts ·
VMware vCenter Server: critical authentication bypass fixed after zero-day attacks
Broadcom has fixed a critical authentication bypass in VMware vCenter Server (VMSA-2026-0009). Mandiant says it was used in zero-day attacks from early September 2026. vCenter 8.0, 9.0 and 9.1 are affected. ESX hosts are not.