Thorn Tech Marketing Ad
Skip to main content
Version: 1.3.1

nginx HTTP/2 vulnerabilities (CVE-2026-123633, CVE-2026-123007): impact on StorageLink

TLDR - Quick Summary

What: On November 18, 2026, two vulnerabilities were published in nginx, the web server in front of the StorageLink web interface: CVE-2026-123633 and CVE-2026-123007. Both are in nginx's HTTP/2 support, which is turned on in our images. The flaws are in nginx itself, not in StorageLink code.

Affected: StorageLink 1.0.0 through 1.4.0, on every cloud and in containers.

Action: Turn off HTTP/2 in nginx and restart nginx (Step 1). This protects a server today with no product upgrade, and the web interface keeps working. Then install the fixed nginx package when your operating system releases it (Step 2). As of November 18, neither Amazon Linux nor Ubuntu has. No exploitation has been reported.

Last updated: 2026-11-18. We will update this article when Amazon Linux and Ubuntu release fixed nginx packages.

Summary​

No exploitation of these issues has been reported. As of November 18, 2026, there are no public reports of them being used in real attacks. A proof-of-concept for CVE-2026-123007 was published the same day; its authors describe it as suitable only for a lab demonstration. Both vulnerabilities carry a CVSS 4.0 score of 9.2 (Critical); Ubuntu and Amazon Linux rate them 8.1 (High) under CVSS 3.1. Because the affected code can be reached without signing in, we recommend acting now.

Affected versions​

StorageLink 1.0.0 through 1.4.0. The web interface of these versions answers over HTTP/2.

Which servers are exposed​

Any server whose web interface answers over HTTP/2. This does not depend on the processor or instance type. To check a server, run:

curl -sk --http2 -o /dev/null -w '%{http_version}\n' https://localhost/

If the command prints 2, HTTP/2 is on and the server is exposed. If it prints 1.1, HTTP/2 is off.

What the risk is​

The affected nginx code can be reached through the StorageLink web interface over HTTPS (port 443) before sign-in. In the worst case described by the researchers who found the flaws, an attacker could run code on the server. An attack that only crashes nginx would interrupt the web portal, its admin pages and any transfers in progress, because users connect through the web interface.

If the web interface does not need to be reachable from the whole internet, restricting it to trusted IP addresses reduces your exposure right away.

Step 1: Turn off HTTP/2 in nginx​

These steps are written for a single server and for containers. If you run high availability or a custom build, adapt them to your environment; the change and the check are the same.

On a VM​

  1. Run the command for your platform. Copy and paste it rather than typing it. It is safe to run more than once.

    AWS (Amazon Linux):

    sudo sed -i 's/ ssl http2;/ ssl;/' /etc/nginx/conf.d/website.conf

    Azure and Google Cloud (Ubuntu):

    sudo sed -i 's/ ssl http2;/ ssl;/' /etc/nginx/sites-available/website.conf
  2. Check the configuration and restart nginx. This interrupts the web interface, including any uploads or downloads in progress, for less than a second, so choose a quiet time.

    sudo nginx -t && sudo systemctl restart nginx
  3. Confirm HTTP/2 is off. This should now print 1.1:

    curl -sk --http2 -o /dev/null -w '%{http_version}\n' https://localhost/

The change stays in place after a reboot and after operating system updates.

After an in-place product upgrade: run the check again. If it prints 2 and nginx has not been updated yet (Step 2), repeat Step 1.

New servers: a server launched from a current image has HTTP/2 turned on. Repeat Step 1 on it until nginx is updated.

On a container (Docker, Kubernetes, Helm)​

When the web interface container serves HTTPS itself, which is the default, it has HTTP/2 turned on and no setting that turns HTTP/2 off. Restrict access to its HTTPS port to trusted IP addresses.

  • Helm: list the trusted addresses under ui.service.loadBalancerSourceRanges in your values file, then run helm upgrade as you normally do.

    ui:
    service:
    loadBalancerSourceRanges:
    - "203.0.113.50/32"
  • Docker, Docker Compose and other Kubernetes deployments: restrict the published HTTPS port with your firewall, security group or network policy.

If an ingress or load balancer terminates TLS in front of the container, and the container runs with EXTERNAL_TLS_TERMINATION=true (on Helm, ui.externalTlsTermination: true), the container serves plain HTTP/1.1 only and no change is needed.

To check a container, run this from a machine that can reach its HTTPS port. 2 means exposed, 1.1 means HTTP/2 is off:

curl -sk --http2 -o /dev/null -w '%{http_version}\n' https://<host>:<https-port>/

Step 2: Update nginx when a fixed package is available​

As of November 18, 2026, neither Amazon Linux nor Ubuntu has released a fixed nginx package. Amazon lists both vulnerabilities as "Pending Fix" and Ubuntu lists them as "Vulnerable". Step 1 protects the server in the meantime. nginx updates available today fix earlier issues but not these two, so keep Step 1 in place.

When Amazon releases the fix, install it and restart nginx:

sudo dnf --releasever=latest update nginx
sudo systemctl restart nginx

The --releasever=latest option is needed because Amazon Linux 2023 installs packages only from the release your server was built on, which does not include the fix. It applies to this command only and does not change the release your server follows.

sudo yum update nginx
sudo systemctl restart nginx

To see which Amazon Linux your server runs, use grep PRETTY_NAME /etc/os-release.

To check the installed version on Amazon Linux, run rpm -q nginx. Upstream nginx fixed both issues in version 1.30.7. We will list Amazon's fixed package versions here once they are released.

Ubuntu images (Azure, Google Cloud)​

Our Ubuntu images install security updates automatically once a day, so most servers will install the fixed package on their own after Ubuntu releases it and restart nginx. Step 1 stays in place through that update. To install it right away:

sudo apt-get update
sudo apt-get install --only-upgrade nginx
sudo systemctl restart nginx

To check whether the installed package has the fix, run this command. It prints 0 until the fixed package is installed:

zgrep -c CVE-2026-123007 /usr/share/doc/nginx/changelog.Debian.gz

Container images​

nginx is part of the container image, so it cannot be updated inside a running container. Restricting access (Step 1) protects the container.

Once nginx is updated, HTTP/2 can be turned back on by reversing Step 1. It is also safe to leave it off.

Other nginx CVEs published the same day​

The nginx release that fixes these two issues also fixes several others, and security scanners may list them against the nginx package: CVE-2026-118899, CVE-2026-123634, CVE-2026-121810, CVE-2026-113425, CVE-2026-125283, CVE-2026-120967 and CVE-2026-123640. On container images, which use a newer nginx line, scanners may also list CVE-2026-120993. Most of these need features that the StorageLink web configuration does not use (the resolver, HTTP/3, the mail proxy, the stream module, the PROXY protocol, or alias). CVE-2026-118899 needs TLS 1.3: the web interface on our VM images does not accept TLS 1.3, but the container does, so restricting access to a container's HTTPS port (Step 1) also covers it. They are fixed by the same nginx update as the two issues above.

References​