blog

The Apache HTTP/2 flaw is now being used in attacks. Check your web server has the fix

If your website runs on Apache with HTTP/2 switched on, check it has this month's fix. A flaw that can let a stranger run code on the server is now being used in attacks.

The fixes have been out for over a week, and every major Linux distribution has them. So this is mostly a case of making sure the update is actually installed.

What happened

On 11 November the Apache HTTP Server project released version 2.4.71. It fixes three flaws, and the one that matters is CVE-2026-119618, which Apache rates critical. It's a bug in mod_http2, the part of Apache that handles HTTP/2, and Apache's own wording is that it "can lead to RCE". That means remote code execution: someone on the internet getting your server to run their code.

It affects Apache 2.4.55 to 2.4.70. It was reported by Anthropic's open-source scanner and by two researchers at Google, which makes it another one of the AI-found bugs I wrote about with nginx this week.

On 17 November Singapore's Cyber Security Agency issued an alert saying it "is reportedly being actively exploited", and that proof-of-concept attack code is public. Its advice is "Patch immediately". That's the bit that moved this from "do it this month" to "do it this week".

The other two flaws fixed in 2.4.71 are less serious: one crashes Apache processes through mod_md (the module that gets Let's Encrypt certificates), and one is rated low.

Does it affect you?

Shared hosting or a managed platform. Your host runs Apache, so it's theirs to patch. It's fair to ask them whether they have.

Your own server. You're only exposed if HTTP/2 is switched on, and in Apache it isn't by default. The Apache docs give the default as Protocols http/1.1. HTTP/2 only comes on if someone has loaded mod_http2 and added h2 to that line.

On Debian or Ubuntu, check both:

sudo apache2ctl -M | grep http2
sudo grep -rn Protocols /etc/apache2/

On Red Hat, AlmaLinux or Rocky, use httpd -M and look in /etc/httpd/ instead.

If the module isn't loaded, or no Protocols line mentions h2, you're not exposed to this one. I'd still install the update, because the mod_md fix is worth having too.

Apache in Docker. Each container has its own Apache, so your server's updates won't reach it. Pull a fresh image.

Where the fixes are

As with most security updates, your distribution copies the fix into its own build, so the version number will look older than 2.4.71. That's normal. Check the package version instead.

Ubuntu fixed it on 12 November in USN-9090-1 and rates it High. On 24.04 you want apache2 2.4.58-1ubuntu8.19 or later, and on 22.04 2.4.52-1ubuntu4.27 or later. Details are on Ubuntu's CVE page.

Debian has 2.4.71 in its security repository: 2.4.71-1~deb13u1 for Debian 13 and 2.4.71-1~deb12u1 for Debian 12, per the security tracker.

Red Hat says RHEL 8 and earlier aren't affected, and rates it Important on RHEL 9 and 10. Its bulletin says fixes "are now available for most affected products". I couldn't see the exact package numbers from the pages I read, so check with dnf updateinfo on your own server.

What to do this week

1. Install the update. On Debian or Ubuntu that's sudo apt update && sudo apt install --only-upgrade apache2, then apt-cache policy apache2 to check you've got a fixed version. On Red Hat and friends, sudo dnf update httpd mod_http2.

2. Restart Apache. sudo systemctl restart apache2 (or httpd). A restart rather than a reload makes sure nothing is left running the old code.

3. Can't update yet? Switch HTTP/2 off. This is the stopgap Red Hat gives. Change your Protocols line to just:

Protocols http/1.1

Then check the config with sudo apache2ctl configtest and restart. Visitors' browsers quietly fall back to HTTP/1.1, and the worst you'll notice is image-heavy pages feeling a touch slower.

4. Check it took. Red Hat suggests this from the server itself:

curl -kI --http2 https://localhost

If it says HTTP/2 200, HTTP/2 is still on. HTTP/1.1 200 OK means it's off.

5. Put a reminder in to turn it back on. Once you're patched, put h2 back. Otherwise it'll stay off for years and nobody will remember why.

The bigger point

Two of the most common web servers in the world have had critical HTTP/2 flaws in the same fortnight, both found with help from AI. Dev Partners wrote recently about AI turning security patches into working attacks, and here it was six days from the fix to a public warning that attacks had started.

None of this needs heroics. It needs someone to check, install and restart, and to do it promptly rather than whenever.

If Apache sits in front of a bespoke app and you're not sure what switching HTTP/2 off would do to it, the team at Dev Partners look after existing software and can check.

And if you'd rather someone else watched the advisories and did the restarts, that's what my support retainers are for.