Security updates
The cloud automatically checks supported services it manages on your servers. You do not need to start checks manually or select individual services. Manually installed software outside cloud management is outside this check's scope.
When an alert appears
See also PHP: October 2026 security fixes and Copy Fail (CVE-2026-31431) for specific procedures. The kernel fix has a separate procedure and requires a server reboot.
Only critical vulnerabilities with a known fixed version affecting an installed service are shown. You can find them in the global security alerts overview; the affected service also has a security update icon.
The vulnerability overview is hidden when there are no matching findings. An empty overview does not confirm that all software on the server is free of vulnerabilities: lower severity issues, issues without a known fix and software outside cloud management are not displayed here.
A known fixed version does not yet mean that all required packages are available for your server. Before installation, the cloud verifies actual fix availability and its source. If it cannot apply the fix safely, it displays the reason and an option to contact support.
Start an update
- Open the server's services and click the security update icon next to the affected service.
- Review the installed and fixed versions. The vulnerability details link to the related security advisories.
- For PHP, optionally select a restart of managed Nginx or Apache on the same server. PHP-FPM restarts automatically.
- Confirm the security update. The cloud prepares the fix, checks the current state and applies the update without another confirmation.
- Follow operation status and in-app notifications. The cloud automatically verifies the result and refreshes the service and security alerts overview after completion.
Starting an update requires access to the server's developer tools. Automatic repair availability depends on the service; if it is not offered, use the support contact in the details.
What gets updated and restarted
Updates stay within the current supported branch. For example, a PHP 8.4 fix uses a patched PHP 8.4 release rather than upgrading to PHP 8.5. The cloud verifies packages for your server and changes only the managed service's scope.
For PHP, this includes already installed, supported components and extensions of that branch. Before installation, the cloud verifies the trusted repository and restores the managed PHP repository configuration when necessary. An update requiring changes outside the supported service scope is blocked.
Restarting PHP-FPM may briefly interrupt requests to all websites using that branch. A selected web server restart may briefly affect other websites it serves. This update restarts selected services; it does not reboot the server.
Progress and outcome
| State | Meaning and next step |
|---|---|
| Queued or preparing | The cloud is waiting for processing or checking the current state and fix availability. Wait for the next status. |
| Installing | The managed service is being updated, followed by the related restarts. |
| Verifying | The cloud checks the installed version and, for PHP, that PHP-FPM and selected web servers are running. No manual verification is needed. |
| Outcome cannot yet be confirmed | The cloud automatically retries verification. It does not repeat installation; further changes remain blocked until the outcome is resolved. Contact support if this state persists. |
| Successfully completed | Installation and automatic verification succeeded. Resolved findings disappear from the overview. |
| Failed or blocked | Read the specific reason. A preparation failure means installation did not start; a failure after installation started may need support to resolve it. |
In-app notifications report progress, success and failure. You do not need an email or Slack rule for these notifications. After successful verification, the overviews in the open application refresh automatically.
How support can help
Support uses the server, service, time and exact message to identify the related operation and escalate technical issues to an authorized operations engineer. The engineer can investigate why an update was blocked, whether the fixed package is available or where an update stopped.
If only confirmation of the result was lost, verification can be repeated without reinstalling. An operation confirmed as not started can be safely cancelled. A partially applied update requires individual technical recovery; sending a support request does not automatically repair it.
Support cannot make an unpublished vendor fix available or bypass the management scope of your server. Changes beyond the originally confirmed scope require a separately agreed procedure.
See troubleshooting and support for guidance on submitting a request.