When virtual patching is enough

A virtual patch is enough only while a fix is out of reach, every route to the flaw passes the filter, and the deadline you answer to accepts mitigation. FedRAMP asks for KEV remediation by the due date even when fully mitigated; PCI SSC separates "resolved" from "addressed" (read 19 Nov 2026).

  1. Remediate
  2. Mitigate now, fix laterThis page is about it
  3. Accept with a dated exception

Seven questions that decide it

OWASP defines virtual patching as "A security policy enforcement layer that detects and blocks matching exploitation attempts against a known vulnerability without changing the vulnerable code." It adds that it "reduces exposure only for traffic the enforcement layer inspects and attack variants its rules recognize" (OWASP). Whether that is enough is a question about the flaw and the deadline, not about the rule. How a rule is built and rolled out is on how virtual patching works.

Conditions for relying on a virtual patch. Sources named in the cells and listed below, read 19 Nov 2026.
QuestionEnough whenNot enough when
1. Is a fix available now?No vendor fix exists, or the fix needs a maintenance window that is dated. OWASP: code fixes and virtual patching "are NOT mutually exclusive", so run the rule while the fix is scheduled.The fix is installable today and the rule is being used to avoid the work.
2. Do all routes to the flaw pass the filter?The vulnerable component is reached only through the inspected path.Internal callers, a second hostname, an API gateway or a direct port reach the component without passing the rule.
3. Can a rule describe the flaw?The flaw is an injection or path pattern. OWASP says the approach is "ideally suited for injection-type flaws".The flaw is a logic or authentication error with no request pattern to match, so the rule would block legitimate traffic or nothing.
4. Has the rule been tested in block mode?A dated test shows the rule blocks the plain and the encoded forms. OWASP: "A passing test for one payload does not establish complete protection."The rule is in log-only mode, or the test used one sample.
5. Does the deadline accept mitigation?The rule governs an internal target date or an exposure value (FedRAMP: once deflected, the flaw "should no longer be considered an internet-reachable vulnerability").One of the three deadlines in the next section applies.
6. Was the asset exposed before the rule?The flaw became known and exploited after the rule was in place, or triage shows no compromise.The flaw was exploited before the rule. The rule stops new attempts and removes nothing already planted.
7. Is there an end date?The rule has a removal date set to the fix date. NIST SP 800-40 Rev. 4, Section 3.5.3: "Schedules should be set and enforced for both patch deployment and mitigation removal."The rule is open-ended. Move the finding to a dated exception instead.

Where a deadline rules it out

  • FedRAMP, KEV due dates. "Providers SHOULD remediate Known Exploited Vulnerabilities according to the due dates in the CISA Known Exploited Vulnerabilities Catalog (even if the vulnerability has been fully mitigated)" (VDR-TFR-KEV). The rules were optional from 4 Jul 2026 and are required for Rev5 initial and ongoing certification from 7 Dec 2026, with a grace period to 7 Mar 2027, so in Oct 2026 the KEV clause bound only providers that had adopted early (FedRAMP).
  • Federal civilian agencies, BOD 26-04. CISA's implementation guidance says: "Aside from removing affected assets from the network, the only known technical mitigation to vulnerabilities in the KEV Catalog is to apply the required actions listed in the catalog." (CISA).
  • PCI DSS, critical and high-risk flaws. PCI SSC FAQ 1597 says Requirements 11.3.1 and 11.3.1.1 together specify that critical and high-risk vulnerabilities are "resolved" and lower-risk ones "addressed"; "resolved" is defined as "the entity solves or fixes the vulnerability". Lower-risk flaws are "addressed", meaning the entity decides whether to resolve them or to mitigate "with a compensating control or by disabling a vulnerable service". A WAF rule can address a lower-risk finding; for a critical one it does not by itself meet "resolved", and whether an assessor accepts it as a compensating control for the requirement is the assessor's call (audit guide; PCI SSC).

Worked example: FortiMail CVE-2026-104286, 1 Oct to 19 Nov 2026

Fortinet published advisory FG-IR-26-175 on 1 Oct 2026 for CVE-2026-104286, a path traversal in FortiMail that lets an unauthenticated attacker write files through crafted HTTP or HTTPS requests, reported as exploited in the wild (CVSS 3 score 9.8). The advisory's fixed builds are 8.0.2, 7.6.7 and 7.4.9; on 1 Oct it called all three "upcoming", and the 5 Oct solution update first listed them as available. For the 7.2 line it says "Upgrade to upcoming 7.2.10 or above", and for 7.0 "Migrate to a fixed release". An appliance on 7.2 can therefore move to the 7.4 or 7.6 branch: the 7.4.9 release notes give the upgrade path 6.0.5 > 6.2.4 > 6.4.5 > 7.0.6 > 7.2.2 > 7.4.9 and list the same supported appliance models as the 7.2.9 notes (200F, 400F, 900F, 2000E, 2000F, 3000E, 3200E, 3000F). Fortinet's options for an unfixed appliance are to turn the IBE service off, to block internet access to the webmail interface or limit it to a trusted private network, or, if a WAF sits in front, to block POST requests to /ibe that contain '../'. That third option is the virtual patch.

  1. 1 Oct: Fortinet publishes FG-IR-26-175; CISA adds CVE-2026-104286 to KEV with a due date of 4 Oct, 3 days later.
  2. 5 Oct: Fortinet revises the advisory (revision 2, solution update), which first lists 8.0.2, 7.6.7 and 7.4.9 as fixed builds, and remediates FortiMail Cloud by updating all firmware to 7.6.7 or 8.0.2, so cloud customers take no action.
  3. 7 Oct: revision 3 clarifies the FortiMail Cloud fix. This is the last revision.
  4. 19 Nov: the advisory, still at revision 3, reads "Upgrade to upcoming 7.2.10 or above" and gives no release date. Fortinet's document library lists FortiMail 7.2 documents up to 7.2.9 and none for 7.2.10, so no fixed 7.2 build is published 49 days after the KEV add date. The fixed 7.4.9 and 7.6.7 builds are.

Take an appliance on 7.2.9 with a WAF in front. The example's own dates: the WAF rule goes live on 2 Oct 2026 after a block-mode test that morning. Working the seven questions:

Seven-question test applied to FortiMail 7.2.9. Facts from Fortinet FG-IR-26-175 and its CVRF revision history, the FortiMail 7.4.9 and 7.2.9 release notes, and CISA KEV (catalog 2026.11.18), read 19 Nov 2026.
QuestionAnswer for the 7.2.9 appliance
1. Fix available?Yes, in another branch. 7.2.10 is still only announced in the advisory on 19 Nov 2026, with no date, but 7.4.9 and 7.6.7 are listed as fixed and the 7.4.9 release notes give an upgrade path from 7.2.2 to 7.4.9 on the same models. The cost is a branch upgrade: a configuration backup, a reboot and a mail-flow interruption, plus the release-note caveats of the new branch. The rule is a bridge for the time that change needs, not for a missing fix.
2. All routes pass the filter?Only if no one reaches the appliance's web interface except through the WAF. Check the appliance's allowed access settings, not only the WAF policy.
3. Can a rule describe it?Yes, it is a path traversal on one path, which suits a pattern rule. Fortinet gives the rule's shape.
4. Tested in block mode?Required before it counts. Record the test date.
5. Deadline accepts mitigation?Not under a KEV clause. The 4 Oct due date came 3 days after the add date. FedRAMP's VDR-TFR-KEV is a SHOULD, bound on 1 Oct only providers that had adopted early, and it counts remediation, not mitigation, so a provider on 7.2.9 could not follow it by the 4 Oct date: 7.4.9 and 7.6.7 were first listed as fixed on 5 Oct, a day later. The provider reports the open finding and its reductions, and mitigation does not satisfy the clause; the steps are on FedRAMP deviation requests. An organization with no such clause can treat the rule as enough for the window to its scheduled upgrade.
6. Exposed before the rule?Possibly. The advisory already reported exploitation on 1 Oct, so the 1 to 2 Oct window before the rule went live is searched in the logs, and the configuration is backed up before changes, as CISA's forensic triage note asks. Fortinet's indicators are listed below the table.
7. End date?Set to the date of the scheduled 7.4.9 or 7.6.7 upgrade, or to 7.2.10 if it ships first. The 7.0 appliances have no fix inside their branch, so their end date is the migration date, and the same release notes list a path from 7.0.6 through 7.2.2 to 7.4.9. Until then the rule is the control. That is an exception, not a patch state.

Fortinet's published indicators of compromise for FG-IR-26-175, as the advisory prints them (Fortinet writes addresses with a bracketed dot):

  • IP addresses: 79[.]141.169.187 and 45[.]129.0.192.
  • System event log: a cron entry beginning type=event subtype=system pri=debug user=system ui=cron msg="(root) CMD (/bin/sh -c 'O=/migadmin ... (Fortinet's text is cut off there).
  • Admin session log: type=kevent subtype=admin pri=information user=admin ui=(null) action=logout status=success reason=unknown msg="User admin logged out from (null)."
  • Config log: type=kevent subtype=config ... msg="Added 'archive234' to 'archive account' : rotation-size[50]rotation-time[1] rotation-hour[14]destination[remote]remote-ip[79.141.169.187]remote-username[archive234]remote-password[***]remote-directory[/uploads] (user: admin, from: cli)"
  • Encryption log: Caught BufferException(2), BufferImpl.cpp:973, 'Invalid Base64 Encoding at pos 0. Character=0x2a' and Internal user *@domain.tld failed to log in.

The simplest of Fortinet's three options, switching IBE off, removes the path with no rule at all and costs the IBE secure mail feature. The WAF rule keeps the feature running and carries the testing burden in question 4. Which to choose is a cost question; the evidence needed for either is on verifying a mitigation. Products that ship WAF and IPS rules are on tools.

What to write in the record when you rely on it

Rule ID and version, block or log-only mode, the routes covered, the dated test, the approver and the removal date. The control entry for a WAF rule lists the same evidence, and the row for a control already in front shows when the finding stays open. NIST SP 800-40 Rev. 4 says organizations "should plan to replace emergency mitigations with permanent fixes".

Sources

Last reviewed