Apache Rated an HTTP/2 Bug Critical and Shipped 2.4.71 on November 11. If Your VPS Runs httpd 2.4.55 Through 2.4.70, Here Is the Ten-Minute Check.
Apache released httpd 2.4.71 on November 11 and listed three fixes, one of them rated critical. That one is CVE-2026-119618, a race condition in assign_headers() inside mod_http2 that Apache's own advisory says "can lead to RCE." It affects httpd 2.4.55 through 2.4.70, which covers most Apache installs that got a normal update in the last couple of years. If you run a single VPS with Apache in front of your app, this is the check I'd do before Monday.
What the bug is, in plain terms
HTTP/2 lets one connection carry many requests at once. mod_http2 is the Apache module that handles that, and the bug is in how it assigns headers when several streams touch shared state at the same time. Apache's description calls it "Concurrent Execution using Shared Resource with Improper Synchronization," which is the formal name for a race condition. Wiz, which published an analysis the same day, says the heap corruption can let a remote, unauthenticated attacker run code with the privileges of the web server process.
Two groups found it independently. Apache credits Anthropic's OSS Scanner and Anthony Weems and Simon Scannell of Google Cloud Vulnerability Research. The advisory's timeline shows Anthropic's report on October 26 and Google's on November 3, with the fix committed and 2.4.71 released on November 11.
The standalone mod_h2 module on GitHub is affected too, versions 2.0.8 through 2.0.42. If you built that yourself to get newer HTTP/2 features, you need a newer module build as well as a newer httpd.
How big is the exposure
Wiz says 64% of cloud environments it can see have resources vulnerable to this CVE, and 51% have publicly exposed instances. Wiz adds a caveat I think matters: those numbers don't take configuration into account. A vulnerable version with HTTP/2 switched off isn't the same risk as one serving h2 to the internet.
As of Wiz's post on November 11, there was no public exploit and no reported exploitation in the wild. I wouldn't lean on that for long. A race condition with a published fix diff is the kind of thing people start working on immediately, and this month we've already had Cloudflare report a sharp jump in automated AI pentest traffic against its customers.
The ten-minute check
Here's what I'd run on each box. First, the version:
apachectl -v # Debian/Ubuntu also accept: apache2 -v httpd -v # RHEL, Alma, Rocky, Amazon Linux
Second, whether mod_http2 is loaded at all:
apachectl -M 2>/dev/null | grep -i http2
Third, whether any virtual host actually offers HTTP/2. In stock Apache the Protocols directive defaults to http/1.1, so h2 is only served where a config line adds it. Control panels, Certbot-era tutorials and performance guides add it all the time, so look:
grep -RniE '^\s*Protocols' /etc/apache2 /etc/httpd 2>/dev/null
If you see h2 or h2c in that output and your version is in the 2.4.55 to 2.4.70 range, that server is the one to fix first.
One trap: distribution packages often backport security fixes without changing the upstream version string. An Ubuntu or Debian package might still report an older 2.4.x number after the patch is applied. So for distro-packaged Apache, the version number alone doesn't answer the question. Check your distribution's security tracker for CVE-2026-119618 and compare against the package version from dpkg -l apache2 or rpm -q httpd. If the tracker says the fix isn't released for your release yet, treat the box as vulnerable.
Patch, or turn HTTP/2 off until you can
The fix is 2.4.71 or a distro package that includes the patch. On a normal Debian or Ubuntu box that's an apt update && apt upgrade followed by a restart of Apache, once the package lands.
If no package is available yet, or you can't restart tonight, Wiz's stopgap is to drop HTTP/2 and serve HTTP/1.1 only, by removing the h2 and h2c references:
# Restrict protocols to HTTP/1.1 until patched Protocols http/1.1
Then apachectl configtest and a graceful restart. The cost is small for most solo sites. You lose HTTP/2 multiplexing to browsers, which matters a little for asset-heavy pages and not at all for an API behind a CDN. If Cloudflare or another CDN sits in front, visitors still get HTTP/2 or HTTP/3 to the edge, and only the edge-to-origin hop drops to HTTP/1.1.
I'd also check the boxes you forgot about. The old client WordPress site on a $5 droplet, the staging server, the Raspberry Pi that runs your status page. These are the ones that sit on 2.4.6x for a year because nothing on them ever breaks.
Where I could be wrong
I'm treating this as urgent based on Apache's critical rating and Wiz's exposure numbers. Neither Apache nor Wiz published a CVSS score in what I read, and Wiz's own numbers ignore configuration, so real exposure is lower than 51%. A commenter on Hacker News pointed out that many PHP setups on the prefork MPM may not be affected in practice. I haven't been able to confirm that from a primary source, so I wouldn't use it as a reason to skip the patch. If it turns out exploitation needs an unusual config, this becomes a routine update. It's still a routine update you should do this week.
What I'd actually do
Today: run the three commands above on every server you own, list the ones with h2 enabled on 2.4.55 to 2.4.70, and either upgrade or set Protocols http/1.1 on those. That's ten minutes for one box and maybe forty for a small fleet.
This week: once your distro ships the patched package, upgrade everything, then put Protocols h2 http/1.1 back where you want HTTP/2. Add a calendar reminder to check the package version in seven days, because "I'll patch when the package lands" is exactly how a box stays vulnerable for months.
Author
Lukas
@lukcombinator