F5 Fixes Six Critical BIG-IP Flaws: How to Find Exposed Virtual Servers and Pick the Right Hotfix

F5 Fixes Six Critical BIG-IP Flaws: How to Find Exposed Virtual Servers and Pick the Right Hotfix

Five iRule bugs and one APM policy flaw, all rated 9.8, can crash the traffic engine and may allow remote code execution. Exposure depends on each virtual server's configuration, and the right fix depends on the exact build.

A BIG-IP admin console locked away on a private network feels safe. With F5's latest flaws, that does not matter much, because the weak spot sits in the part of the device that handles your users' live traffic.

Today is November 21, 2026. On November 18, F5 released emergency engineering hotfixes for six critical BIG-IP vulnerabilities, all scored 9.8 out of 10 under CVSS v3.1 and 9.3 under CVSS v4.0. F5 also fixed a seventh, high-severity bug that can knock out SAML sign-ins. BIG-IP sits in front of many company applications, remote-access portals and DNS services, and attackers already exploited a different BIG-IP flaw in September. This post explains what was fixed, who is exposed and how to work out the right fix for each device.

What BIG-IP does and where the flaws live

BIG-IP is F5's platform for spreading application traffic across servers, enforcing access rules and providing remote access. Two terms help:

  • A virtual server is the address and port on BIG-IP that receives traffic for one service, such as a website or a VPN portal.
  • An iRule is a small script attached to a virtual server that inspects or changes traffic as it passes through.

All six critical flaws affect the Traffic Management Microkernel (TMM), the process that handles every connection going through BIG-IP. On a virtual server with a vulnerable setup, a request from someone who has not logged in can make TMM crash. Traffic stops while it restarts. F5 also lists remote code execution, meaning an attacker running their own code on the device, as a possible result.

This is the data plane, the part that handles real user traffic. A customer opening your app or an employee opening the remote-access portal reaches TMM before any login happens. So a private admin console, or an access policy that asks for a password, does not protect against these bugs.

The six critical flaws

Five involve iRules. Each is a stack-based buffer overflow, a memory error where data spills past the space set aside for it, which can crash a process and may let an attacker steer it:

  • CVE-2026-121805: an iRule that processes subscriber information in client traffic.
  • CVE-2026-116953: an iRule used for generic message routing.
  • CVE-2026-120292: needs an HTTP profile plus a message-routing iRule on the same virtual server.
  • CVE-2026-120285: needs an HTTP profile plus an iRule using an IP-address function on the same virtual server.
  • CVE-2026-120722: an iRule that processes DNS traffic. F5's DNS advisory names BIG-IP DNS, BIG-IP CNF and BIG-IP Next for Kubernetes, so DNS and cloud-native teams need to check too.

The sixth, CVE-2026-123637, is in Access Policy Manager (APM), the component that checks users and their devices before letting them in. It applies only when APM is provisioned and a virtual server uses a per-request policy that contains both the Client Information agent and Endpoint Security agents. Knowing that APM is installed is not enough. You must look at the policy on each portal.

The SAML bug that can break sign-ins

The seventh flaw, CVE-2026-126822 (CVSS v4.0 8.7), affects APM when an access profile acts as a SAML service provider, meaning BIG-IP accepts sign-in assertions from your identity provider. An integer underflow, a number that wraps around below zero, lets an unauthenticated attacker crash TMM. The published impact is denial of service. Your identity provider can look perfectly healthy while users still cannot sign in, because the BIG-IP in front of the application keeps restarting.

How big is the exposure?

On November 20, Shadowserver counted more than 177,000 BIG-IP instances visible on the internet, and over 14,600 IP addresses showing APM fingerprints. Those numbers show reach, not risk. They cannot tell which devices run an affected build, have a vulnerable iRule on a reachable service or are already patched, and they may include honeypots.

As of November 20, no attacks on the new flaws had been reported. But in September, F5 confirmed attacks on CVE-2026-94127, a separate APM flaw. Health-ISAC also warned that health record portals, telehealth and clinical remote access could be hit hard by even short outages. An unexpected TMM restart is worth looking into, but on its own it does not prove an attack.

How Rohit checks Acme's BIG-IP fleet

Rohit looks after the BIG-IP devices at Acme. Here is the order Rohit works in.

Step 1: Record the exact build on every device

The right hotfix depends on the exact build, not just the branch. On each device, from the BIG-IP command line:

tmsh show sys version
tmsh show sys software status

Note the full version and any engineering hotfix already installed. Two devices on the same branch can need different fixes.

Step 2: List virtual servers with their iRules and profiles

tmsh list ltm virtual rules profiles

Mark every virtual server that is reachable from outside, and note which have iRules and which have an HTTP profile.

Step 3: Read the iRules that are attached

tmsh list ltm rule

Compare each attached iRule with the functions named in F5's advisories: subscriber information, message routing, IP-address functions and DNS processing. Do this together with the iRule's author if possible; the vendor advisory is the only reliable guide to the exact commands affected.

Step 4: Check APM access and per-request policies

For each remote-access portal, open its access profile and per-request policy in the admin interface. Flag any per-request policy that contains both the Client Information agent and Endpoint Security agents (CVE-2026-123637), and any access profile set up as a SAML service provider (CVE-2026-126822).

Step 5: Choose the hotfix for each build

F5's guidance for standard base releases:

Running Install
17.1.x up to base 17.1.3.7 BIGIP-17.1.3.7.0.21.4-ENG
17.5.x up to base 17.5.1.11 BIGIP-17.5.1.11.0.61.12-ENG
21.1.x up to base 21.1.0.4 BIGIP-21.1.0.4.0.19.3-ENG

Exceptions: devices already on certain engineering hotfixes for 17.1.3.6 or 17.5.1.10 need different replacement builds, custom engineering builds need F5 Support, and BIG-IP CNF and BIG-IP Next for Kubernetes have their own packages for the 2.2, 2.3 and 2.4 series. F5 published no workaround for the critical flaws, and it assesses only versions still in technical support, so treat older devices as exposed until proven otherwise.

Step 6: Patch, then confirm real transactions work

Take a configuration backup (a UCS archive) first, patch the standby unit of a high-availability pair before the active one, and plan a maintenance window. Afterwards, check more than "the device is up": resolve a DNS name, load the website, complete a VPN and a SAML sign-in. Also look for recent unexpected TMM restarts:

grep -i tmm /var/log/ltm | tail -n 50
ls -l /var/core

Conclusion

F5's November 2026 hotfixes close six critical BIG-IP flaws that can be reached by anyone who can send traffic to an affected virtual server, plus a SAML bug that can break sign-ins. A private admin console does not help, because the problem sits in the traffic engine. Build an inventory of reachable virtual servers with their iRules, access profiles and exact builds, pick the hotfix that matches each build, and finish by proving that real users can resolve, connect and sign in again.

Merits

  • Fixes exist for all three main TMOS branches and for the cloud-native products.
  • F5 published build-by-build guidance, which reduces guesswork.
  • Exposure is limited to specific configurations, so an inventory can narrow the urgent list.
  • No attacks on the new flaws had been reported at the time of writing, giving a short window to patch.
  • The standard tmsh commands above are enough for a first inventory.

Demerits

  • No workaround exists for the critical flaws, so patching is the only real fix.
  • Engineering hotfixes are build-specific, and a wrong choice wastes a maintenance window.
  • Checking iRules and APM policies one by one takes time on large fleets.
  • Patching BIG-IP interrupts traffic unless high availability is set up well.
  • Devices outside technical support are not assessed by F5.

Caution

This article is educational. Acme, the example person Rohit and the commands are illustrations; check every command and file location against your own BIG-IP version and documentation, and test changes in a maintenance window first. The CVE numbers, scores, affected configurations and hotfix names summarise F5's advisories and public reporting at the time of writing. Read the official F5 security advisories and support guidance for your exact build before acting, and only assess devices you own or are authorised to manage.

Frequently asked questions

  • Which BIG-IP flaws did F5 fix in November 2026? — Six critical flaws (five in iRules and CVE-2026-123637 in APM, all CVSS 9.8) and a high-severity SAML denial-of-service bug, CVE-2026-126822.
  • Does a private management network protect BIG-IP from these flaws? — No, the flaws are in TMM, the traffic engine, which public users reach before any login.
  • How do I know if a virtual server is affected? — It depends on the iRules, HTTP profile and APM policies attached to it, so list them with tmsh and compare them with F5's advisories.
  • Which hotfix should I install on BIG-IP 17.1? — Systems on base 17.1.3.7 or earlier move to BIGIP-17.1.3.7.0.21.4-ENG, unless they already run certain engineering hotfixes.
  • Is there a workaround for the critical BIG-IP flaws? — F5 published no workaround at disclosure, so installing the correct hotfix is the fix.
  • Are the November BIG-IP flaws being exploited? — No attacks had been reported as of November 20, 2026, but a separate APM flaw was exploited in September.
  • Does the SAML flaw allow code execution? — Its published impact is denial of service: TMM can crash and interrupt sign-ins.
  • Are BIG-IP DNS and Kubernetes deployments affected? — CVE-2026-120722 affects BIG-IP DNS, BIG-IP CNF and BIG-IP Next for Kubernetes, which have their own update packages.

Tags

#f5 #bigip #cybersecurity #vulnerability #patchmanagement #networksecurity #irules #apm #remotecodeexecution #devsecops

Free field guide

Kubernetes Security Checklist

Harden cluster access, workload identity, pod security, network boundaries, software supply chain, secrets, and operational monitoring.