SOCLYDE logo
Current languageEN
Cybersecurity newscybersecurityLinux

NoisyTCP can crash Linux servers remotely

NoisyTCP can remotely crash Linux servers before login. Confirm a fixed kernel with your hosting provider and plan the restart and service checks.

Published on

By Soclyde

A Linux maintenance workstation in a paused production workshop

In summary

  • NoisyTCP affects kernel TCP processing before authentication.
  • A crash can take down the whole host, even through a containerized service.
  • Deploy the kernel fix and verify that it is actually running.

Explore next

Soclyde resources

Article contents

Hyunwoo Kim’s November 15, 2026 NoisyTCP (CVE-2026-99288) disclosure describes a Linux kernel flaw reachable before authentication. An attacker can remotely crash exposed TCP servers. For a small business, the concern is continuity of hosted applications rather than just the security of an SSH login.

A race in kernel connection handling

The researcher’s advisory describes concurrent TCP connection processing going wrong when incoming packets are spread across receive contexts. Services such as Nginx, Apache and SSH can be reached without unusual application settings.

The network conditions matter: multiple receive queues or software packet distribution across CPUs are required. Ask your hosting provider to assess the kernel exposure rather than checking only the web server version.

Containers do not preserve host availability

The Register highlights that a kernel crash can take down the entire host even when the targeted service is containerized. Several applications sharing that machine may therefore stop together.

Group services by host in your inventory: a website, proxy and internal tool are not necessarily independent environments. Recovery planning needs to account for that shared dependency.

Verify the kernel actually running

iTnews reports fixes in several stable branches and notes that observed crash times differ sharply between virtual machines and physical hardware. Those lab measurements do not establish a universal time to disruption.

Request the fixed package supplied by your distribution and evidence that it is active after maintenance. Downloading an update does not show that an older kernel has stopped running.

Avoid improvised network workarounds

iTnews explains the tradeoffs of temporary measures such as disabling SYN cookies: this reduces protection against some flooding attacks and does not cover every described path. Reducing reception to one queue can also affect performance.

Have a network administrator assess these options if patching immediately is impossible. Plan a restart window, application checks and a contact channel that does not rely on the affected host.

Linux maintenance credentials remain necessary during the intervention. Soclyde stores them in an encrypted vault on your devices, with secrets distinct from other business accounts.

That helps authorized staff find the access they need. A password vault does not repair the kernel or prevent this pre-login denial of service.

Key points

Confirm the distribution fix, its activation and application recovery after maintenance. The Soclyde generator remains useful for human credentials alongside patch management.

Frequently asked questions

Will a strong password stop this attack?

No. NoisyTCP can be triggered before authentication against a reachable TCP service under the network-processing conditions described by the researcher.

Will a container contain the outage?

Not necessarily. The flaw is in the host kernel, so a crash triggered through a containerized server can interrupt the host and its other workloads.

Has remote code execution been demonstrated?

The researcher distinguishes achievable denial of service from a harder memory-corruption path. The advisory does not demonstrate a general-purpose RCE exploit; that must not be confused with the remote crash.

References

Sources and references

Need advice?

Design your password strategy with Soclyde

Schedule a dedicated walkthrough with the team to see how local-first security adapts to your stack.

Talk with us

Keep reading