← Back to Blog

A 9.8 vCenter Flaw Was Exploited Since September: Patch It and NGINX Now

· 10 min read A 9.8 vCenter Flaw Was Exploited Since September: Patch It and NGINX Now

Two critical, vendor-patched flaws landed in the same 48 hours, and both sit in software that small businesses run without thinking about it. Broadcom patched a VMware vCenter authentication bypass, CVE-2026-123890, rated 9.8 out of 10. Google says a China-linked group had been exploiting it since September, which is roughly ten weeks before the fix. The day before, F5 shipped fixes for five critical NGINX flaws, nicknamed NGINX2Root, led by an unauthenticated HTTP/2 bug that has been in the code since 2015.

If you run your own VMware servers, an NGINX web server or reverse proxy, or you pay a managed IT provider who does, this is a this-week problem. The vCenter flaw matters most because "patched" is not the same as "safe." Attackers had a head start of about two months. Patching closes the door. It does not tell you whether someone already walked through it.

Here's the part that trips people up. You may not know you run either product. NGINX hides inside appliances, hosting plans and container images. vCenter sits in a back closet where an installer put it in 2022. This post is about finding out, patching in the right order, and checking for a prior visit. It is not a panic piece, and no victim has been publicly named.

What happened

The analysis below comes from Apprised.news desk briefs. The tech and cyber desk's November 20 brief covers the vCenter flaw, and its November 19 brief covers NGINX. I'm summarising their reading of public reporting, and the original sources are linked so you can check them.

vCenter. Broadcom patched CVE-2026-123890 after Google said a Chinese-linked group had exploited it since September. Rapid7 describes it as a critical authentication bypass in VMware vCenter. SecurityWeek and BleepingComputer both report it was used as a zero-day, meaning attackers had it before a fix existed. The desk's weighted read is that this is the story that matters most this week and that it is a genuine, serious intrusion campaign.

One caution from the desk. Its Exfiltration analyst reads the access profile as consistent with long-dwell collection, meaning quiet, patient data gathering. That is an inference from how the flaw works, not a report of any victim or any stolen material. Treat it as a reason to check your logs, not as proof anything was taken.

NGINX. F5 shipped fixes for five critical flaws, dubbed NGINX2Root, led by CVE-2026-123007, an unauthenticated HTTP/2 bug present since 2015. The Hacker News reports the flaw is ten years old. Qualys published guidance on finding, mitigating and patching all five, and Tenable published a Q&A that also mentions a separate Linux kernel privilege-escalation flaw that was still unpatched at the time. CSO Online covers how the flaws were found.

Unlike vCenter, there is no report yet of NGINX2Root being exploited. The desk lists "do attackers start using it within days of the patch" as a named watch item. History says the gap between a public patch and working attack code is often short, so this is the window to move.

The wider list. The same desk brief notes CISA added three exploited bugs to its Known Exploited Vulnerabilities (KEV) catalog in a week, and that Apache Roller's CVE-2026-129448 is reported under active attack. Whether the vCenter flaw lands in KEV, and whether the entry flags ransomware use, is another watch item.

What it means for a small business

Three groups are exposed. Anyone running vCenter or the ESXi hosts it manages. Anyone running NGINX directly, as a web server or reverse proxy. And anyone whose IT provider runs either one on their behalf and hasn't said so. The three examples below are hypothetical, with the arithmetic shown.

A 25-person accounting firm with two hosts in a closet

An accounting firm in Ohio runs two VMware hosts managed by one vCenter, set up by an MSP in 2022. The hosts hold the file server and the tax-software database. Exploitation reportedly began in September and the fix arrived in mid-November, so the exposure window is about 10 weeks.

The work: inventory is 2 hosts plus 1 vCenter. Patch time is about 2 hours of work plus a reboot window, call it a Saturday morning. The inventory question that matters most is whether the vCenter management interface was ever reachable from the internet. If it was, treat the window as real exposure and review logs back to September. If it sat on an internal network only, the risk is much lower, though not zero.

Compare the cost. A Saturday morning of an MSP's time, perhaps $300 to $600 at typical rates, against a tax-season outage or a client-data incident. At 25 staff billing even $60 an hour of productive time, one lost day is 25 x 8 x $60 = $12,000 before you count client notification. The patch is cheap by comparison.

A 12-person online retailer on a $40 VPS

An online retailer runs NGINX as a reverse proxy in front of its store on a $40-a-month VPS. It last updated the package in the spring. The owner runs nginx -v to see the version, applies the distribution's security update, and confirms HTTP/2 is only enabled on the sites that need it.

The arithmetic: a 30-minute maintenance window at an off-peak hour. If the store does $2,400 a day, that's about $100 an hour on average, and less overnight, so a half-hour window costs well under $50 in lost sales. A compromised proxy that takes the store offline for a day costs the full $2,400 plus cleanup. Even if you never get attacked, the window is the cheaper path.

A 40-person manufacturer that has never seen an asset list

A manufacturer outsources IT to a managed provider and has never been shown a list of what the provider runs. The owner can't answer "do we use vCenter?" and shouldn't have to guess. So she sends five written questions: Do we run vCenter, and which version? Was it patched? Was the management interface ever internet-reachable? Can you show logs back to September? Do we run NGINX anywhere, including inside appliances?

She sets a deadline of 3 business days. If the provider answers in an hour with specifics, good. If the answer is vague or late, that tells her something too. Asking costs one email. Not asking leaves the risk invisible.

What to do this week

  1. Inventory in one sitting. List every VMware vCenter, every NGINX server or reverse proxy, and every appliance or hosting product that quietly bundles NGINX. Ask your IT provider for the same list in writing, then compare the two.
  2. Patch vCenter first, then NGINX. vCenter was exploited for about two months before the fix, so it goes first. Test on a snapshot or staging host if you have one, and schedule a reboot window.
  3. Take vCenter's management interface off the internet. Restrict it to a VPN or a short list of admin addresses. Patching fixes this flaw. Closing the exposure helps with the next one.
  4. Look for a prior visit. Because exploitation began in September, review vCenter and ESXi logs, new accounts, scheduled tasks and unexpected outbound connections back to September. Or have your provider do it and confirm the result in writing.
  5. Update NGINX the boring way. Use your OS package manager or vendor, then confirm the running version afterward (a patched package on disk doesn't help if the old process is still running). If you can't patch immediately, read the Qualys mitigation guidance linked above.
  6. Watch the CISA KEV catalog weekly. If a product you run appears, treat CISA's due date as your own deadline.
  7. If you can't patch or inspect within a week, shrink the exposure. Ask your provider about taking the system off the internet or putting it behind a VPN until you can.
  8. Write down ownership. For each system, name who owns patching and the maximum days you'll allow for a critical fix. Put the date on the calendar.

Two free jwatte.com tools help with adjacent checks. The email breach check tells you whether staff addresses have turned up in known breaches, which is worth knowing if an intruder was in your network. And the DNS and email audit covers the mail-spoofing side of the house while you're in a security mindset. Neither replaces patching.

Copy-paste prompts

Each prompt works best if you paste in your own documents or lists. Tell the AI to say when it is unsure rather than guess.

Use this first, to work out whether you run either product at all:

I run a small business with these systems: [paste your server list, hypervisors, web servers, hosting plan, and any appliance model numbers]. Which of them might use VMware vCenter or NGINX, including inside appliances, containers or managed hosting? Give me a checklist of how to confirm each one without needing to be an engineer. If you are not sure whether a product bundles NGINX or vCenter, say so plainly and tell me what to look up rather than guessing.

Use this to put the questions to your IT provider in writing:

Write a plain-English email from a business owner to our managed IT provider asking five specific questions about VMware vCenter CVE-2026-123890 and the NGINX CVE-2026-123007 flaws: whether we run them, whether they are patched, whether management interfaces were ever exposed to the internet, whether logs go back to September, and a response deadline of three business days. Keep it polite and firm. Here is our current service agreement for context: [paste relevant sections]. If the agreement says nothing about patch responsibility, point that out.

Use this if you administer a single Linux box yourself:

I administer a single Ubuntu server running NGINX. Here is the output of nginx -v and my OS release: [paste]. Give me step-by-step commands to check the installed NGINX version, apply the vendor security update, restart safely, and verify the new version is actually running, with a rollback plan if the site breaks. Tell me where my output suggests something unusual, and say when you are unsure what a command will do on my setup.

Use this after patching vCenter, when you need to look for a prior visit:

A critical vCenter flaw was reportedly exploited from September and patched in mid-November. Here are exports from my vCenter and ESXi logs, my account list and my scheduled tasks: [paste or summarise]. List the kinds of entries, accounts, tasks and outbound connections I should treat as suspicious, flag anything in my data that looks odd, and tell me what to do if I find something. If the data I gave you is not enough to judge, say what else you would need.

Use this to turn the lesson into a standing rule:

Draft a one-page patch policy for a [number]-person company. Cover who owns patching, the maximum number of days for critical (CVSS 9 and above) and exploited-in-the-wild flaws, how we use the CISA Known Exploited Vulnerabilities list, and how we record exceptions. Here is our current list of systems and owners: [paste]. Where my list has gaps, say so instead of inventing owners.

Download the kit

The kit is a printable markdown file with a system inventory worksheet, a patch-order and maintenance-window checklist with rollback notes, a post-patch compromise review checklist, an email template for your IT provider, a one-page patch policy, a hardening checklist for management interfaces, and the prompts above. Get it here: /downloads/vcenter-nginx-critical-patches-small-business-kit.md

Fact-check notes and sources

  • Apprised.news tech and cyber desk, November 20: the vCenter patch, the Google attribution, the desk's weighted read and the long-dwell inference. apprised.news/desk/tech/2026-11-20
  • Apprised.news tech and cyber desk, November 19: the five NGINX flaws, CISA's three added bugs and the Apache Roller item. apprised.news/desk/tech/2026-11-19
  • Rapid7: describes CVE-2026-123890 as a critical authentication bypass in vCenter. Rapid7 post
  • SecurityWeek: reports the vCenter flaw was exploited as a zero-day. SecurityWeek
  • BleepingComputer: reports exploitation since September. BleepingComputer
  • Qualys: guidance on finding, mitigating and patching the five NGINX flaws. Qualys
  • Tenable: Q&A on NGINX2Root and the separate Linux kernel flaw. Tenable
  • The Hacker News: reports the HTTP/2 flaw is ten years old. The Hacker News
  • CSO Online: coverage of how the NGINX flaws were found. CSO Online
  • Security Affairs: Apache Roller CVE-2026-129448 reported under active attack. Security Affairs
  • Worked examples: the three businesses, their prices and the costs are hypothetical illustrations, not real companies.

Related reading

This post is informational, not legal, financial or security-consulting advice. It draws on AI-written Apprised.news analysis of public reporting; check the linked sources before acting. Mentions of third parties are nominative fair use. No affiliation is implied.

← Back to Blog

Accessibility Options

Text Size
High Contrast
Reduce Motion
Reading Guide
Link Highlighting
Accessibility Statement

J.A. Watte is committed to ensuring digital accessibility for people with disabilities. This site conforms to WCAG 2.1 and 2.2 Level AA guidelines.

Measures Taken

  • Semantic HTML with proper heading hierarchy
  • ARIA labels and roles for interactive components
  • Color contrast ratios meeting WCAG AA (4.5:1)
  • Full keyboard navigation support
  • Skip navigation link
  • Visible focus indicators (3:1 contrast)
  • 44px minimum touch/click targets
  • Dark/light theme with system preference detection
  • Responsive design for all devices
  • Reduced motion support (CSS + toggle)
  • Text size customization (14px–20px)
  • Print stylesheet

Feedback

Contact: jwatte.com/contact

Full Accessibility Statement • Privacy Policy

Last updated: April 2026