· via dev.to (home feed)
Turning vCenter exposure counts into closures after CVE-2026-123890
A dev.to walkthrough uses VMware's CVE-2026-123890 authentication bypass to show how teams can turn internet exposure counts into owned, verified and documented security closures.

The vulnerability behind the exercise
On November 19, 2026, Broadcom disclosed CVE-2026-123890, an authentication bypass in the VMware Directory Service carrying a 9.8 severity rating and no published workaround. Google's Threat Intelligence Group and Mandiant reported that exploitation had already occurred by early September, describing an intrusion chain that involved exfiltrating the LMDB database, abusing the SAML security token service, creating a rogue SSO administrator, enabling SSH through the VAMI interface, and moving on to ESXi hosts.
A dev.to post published on November 21, 2026 uses that incident as a case study in how to convert raw exposure measurements into accountable security decisions.
What the exposure queries found
According to the post, an all-assets query for the fingerprint app="VMware vCenter" returned 27,430 public observations. Intersecting the same fingerprint with specific management ports returned 13,570 hits on port 443 and 23 on port 9080, while the port 2012 and port 5480 intersections each returned zero. A query for changed assets also returned zero in the same review.
The post's core argument is that these numbers are inputs rather than conclusions. The measurable question worth asking, it says, is which vCenter management planes are reachable from untrusted networks and which of those run a verified fixed build — not simply how many vCenter assets exist or how many public services resemble one.
Zeros get the same scrutiny as positive counts. A zero means the query ran successfully and found nothing matching that scope at that time; it does not prove the service is absent from internal networks or that the flaw cannot be exploited.
Evidence and ownership
The post recommends storing the exact query string, subtype, timestamp, count, status and selection reason for each run, arguing that a screenshot or a bare number without a collection time is not reproducible evidence.
Every open item should map to a ticket or control record per asset group, not one ticket covering the entire public count. The assigned owner must be able to identify the deployment, its management path, running build, patch state and business service. Findings that turn out to be false positives — a shared certificate, a load balancer or a monitoring page masquerading as vCenter — need documented evidence for the dismissal. Legitimate but restricted systems need the firewall rule, allowed source and controlling control recorded so the record survives staff turnover.
A six-rung "closure ladder"
The post lays out a sequential process: confirm identity and ownership, confirm the product version against the vendor's fixed build, confirm the network path and limiting control, review for signs of exploitation where a system was both vulnerable and reachable, verify the fix took effect by re-running the agreed query or an internal test, and retain all evidence in one review record.
A key warning: do not treat a closed patch ticket as the finish line. Because the Google report describes post-exploitation behavior that can persist after the vulnerable service is updated, the response may also require removing accounts, rotating keys and reviewing host access.
Rescan rhythm and context
Rescans should follow the operational decision rather than run ad hoc: short intervals during active response, monthly or quarterly in steady state. A shifting public count without corresponding asset records is noise, and teams should investigate whether a new deployment, rule change, fingerprint change or observation-set change explains it.
The post also flags vCenter-specific facts from the incident: port 2012 sits on the SSO control path, 5480 is the VAMI interface that can enable SSH, and 9080 was among the ports the actor scanned. Vendor port guidance and the incident report should be combined, it argues, but neither replaces knowledge of an organization's own appliance configuration.
Why it matters
Headline exposure counts are easy to produce and easy to misread. The dev.to post reframes success not as a shrinking number but as a set of vCenter control planes with a known owner, a verified fixed build, an explicit management path, a tested restriction and a recorded exploitation review where needed — plus defined reopen triggers such as a new public observation, a changed SSO account, a VAMI state change, an SSH session or an unexplained ESXi login. The same closure process generalizes to any cloud or infrastructure exposure program where a fingerprinted count must become a verified, owned decision.
- #vmware
- #vcenter
- #security
- #cloud
- #vulnerability-management