For a PHP or WordPress service on Ubuntu, separate three decisions: which vendor package applies, which service or deployment unit owns HTTP/2, and what the operator must demonstrate before maintenance is accepted. This article concerns CVE-2026-119618 in Apache's mod_http2; it does not establish that a particular website is affected.
Keep upstream and Ubuntu guidance separate
Apache's security page records an upstream repair in HTTP Server 2.4.71, released November 11, 2026. Its affected httpd range is 2.4.55 through 2.4.70; the separately distributed GitHub mod_h2 range is 2.0.8 through 2.0.42. Apache HTTP Server security advisory.
Ubuntu's USN-9090-1, published November 12, supplies distribution-specific corrections. Its CVE tracker explains that newer module versions were backported into stable Ubuntu releases. Consequently, an Ubuntu Apache version prefix below 2.4.55 does not by itself rule this issue out, and a prefix below 2.4.71 does not by itself prove that a corrected Ubuntu package remains affected. Ubuntu CVE-2026-119618 status, updated November 13.
For Ubuntu-provided packages, use the complete distribution revision and matching release lane. At this review, the vendor lists:
| Ubuntu release | Corrected apache2 revision for this CVE |
|---|---|
| 26.10 | 2.4.68-1ubuntu6.2 |
| 26.04 LTS | 2.4.66-2ubuntu2.8 |
| 24.04 LTS | 2.4.58-1ubuntu8.19 |
| 22.04 LTS | 2.4.52-1ubuntu4.27 |
| 20.04 LTS | 2.4.41-4ubuntu3.23+esm11, with the stated Ubuntu Pro access condition |
The CVE tracker marks 18.04, 16.04 and 14.04 not affected by this specific issue. USN-9090-1 also contains other CVEs and older-release package rows; those rows must not be used to expand this CVE's applicability. Consult the current vendor record before choosing a maintenance target. Ubuntu USN-9090-1.
Find the operator responsible for the request path
Ask the hosting or infrastructure operator which component terminates HTTP/2 for the site. It may be a front-end Apache service, a proxy, a managed hosting layer or a different deployment unit from the PHP application. Identify every relevant instance rather than assuming that an application-server update covers the edge.
For each Apache instance in scope, request a redacted inventory row containing the Ubuntu release, repository or package origin, full package revision, relevant loaded module, and service unit or container/image identifier. A control-panel build or source installation needs its own vendor instructions; the Ubuntu package table is not a repair recipe for those channels.
This inventory is a scoped request to the authorized operator. It does not require the website owner to send credentials, raw request logs or customer data through a public contact form.
Agree activation and restoration before changing the service
The operator should select the supported update and describe how it will become active in the identified serving environment. Do not assume that changing files on a host proves that every process or container serving requests uses them.
Record the maintenance owner, planned window, configuration validation, activation method and restoration route. Apache documents configuration testing and different restart behaviors; the operator must select a procedure appropriate to the actual service and package integration. A generic command copied into the wrong environment is not an acceptance plan. Apache control-interface documentation.
If there are multiple instances, account for all of them. A healthy response from one node cannot close an inventory row for another. Likewise, a successful application build does not demonstrate that its upstream proxy has been maintained.
Use a maintenance acceptance record
The following worksheet connects technical evidence to a business decision. Its checks are proposed review steps, not tests performed for a customer during this research.
| Acceptance item | Evidence supplied by the responsible operator | Who accepts it |
|---|---|---|
| Correct package lane | Exact Ubuntu release, origin and full revision | Infrastructure maintainer |
| Relevant serving instance | Module and service/container identity | Hosting or platform owner |
| Activation completed | Deployment or activation record for that instance | Release owner |
| Intended site behavior | Synthetic login, content and checkout journey where applicable | Application owner |
| Recovery route available | Named owner, restoration conditions and retained configuration/release reference | Maintenance coordinator |
Keep security applicability and business-function validation separate. A working checkout does not prove that a vulnerable module is corrected. Conversely, the right package revision does not show that login or payment-provider callbacks still behave as intended.
If package origin or service ownership is unknown, the next deliverable is an inventory and handoff. If the appropriate correction is available, agree implementation scope and acceptance before maintenance. If behavior fails afterward, the operator should follow the agreed restoration decision while preserving the need to address the applicable security issue; an older package is not automatically a safe long-term destination.
For a client-facing agency, this makes the work order concrete: identify the responsible hosting lane, obtain the evidence, coordinate the maintenance and confirm the agreed user journeys. A Recovery Review can clarify that scope. The agency route preserves the agency's role in the client relationship.
No affected customer, successful exploitation, runtime inspection or completed repair is asserted here. This is an independent maintenance decision framework based on the cited vendor records.