Recovering from a failed change to hosted websites and web roots

This article is part of the Spark Rack Website Hosting knowledgebase and covers Websites & Web Roots. This recovery guide is designed for a failed or uncertain change. It prioritizes stopping further damage, preserving evidence, restoring the last known-good state, and validating the whole service rather than only the original screen.

The relevant customer area is Services > Manage or Control Panel > Websites. It is used to manage hosted websites and web roots, including domains, aliases, document roots, index files, logs, application URLs, and website-level settings. Controls can differ by product, lifecycle status, account permission, domain state, payment method, or service configuration. When an action is not shown, confirm eligibility and permission instead of following an unrelated workaround.

Before you begin

• Have an active hosting service and domain
• Know the intended document root
• Create a current file and database backup
• Confirm application compatibility with the target PHP settings
• Schedule production changes during a controlled window
• Identify the exact item associated with hosted websites and web roots and confirm its current status.
• Write down the expected result and the current value before changing anything.
• Plan a rollback or recovery path for any action that can affect access, billing, data, routing, delivery, or availability.

Recovery sequence

1. Stop repeated submissions, automated jobs, deployments, retries, or additional configuration changes that may expand the incident.
2. Preserve the exact error, timestamps, current values, logs available to the customer, and a list of recent changes.
3. Define the affected scope and confirm which functions remain healthy.
4. Protect account access and rotate a credential immediately only when exposure is suspected; otherwise preserve evidence before broad resets.
5. Restore the last known-good value, release, record set, route, or configuration using the documented rollback.
6. Re-test the smallest critical path first, then dependent and customer-facing functions.
7. Check for delayed effects such as queued mail, cached DNS, pending payments, scheduled tasks, or incomplete provisioning.
8. Open a support request when recovery requires provider-side action, hidden logs, data restoration, or authorization changes.
9. Document the root cause, corrective action, and prevention before closing the incident.

Do not make recovery harder

• Do not delete the failing object merely to recreate it unless data retention and dependencies are understood.
• Do not overwrite the only backup with a new copy of the damaged state.
• Do not clear every cache or log before capturing evidence and timestamps.
• Do not disable security controls globally to bypass a specific error.
• Do not assume rollback is complete until customer-side behavior and dependent systems are verified.

Detailed operating procedure

Use this sequence as the baseline workflow for Websites & Web Roots. Some steps may be informational when the selected product does not expose that exact control, but the verification and recordkeeping principles still apply.

1. Open the service’s website or configuration area
2. Confirm the domain, aliases, web root, active status, and current PHP values
3. Place public files only in the intended document root and keep secrets outside it
4. Ensure the expected index file exists with correct case
5. Change one PHP or application setting at a time
6. Wait for Saving Changes to complete and confirm the same tab shows the new value
7. Test public pages, administrative pages, forms, scheduled tasks, and logs
8. Confirm that dependent settings still point to the intended destination and that no older value is overriding the new one.
9. Observe the result long enough to catch delayed processing, caching, queued work, or an intermittent failure.
10. Update internal documentation with the final value, date, owner, result, and any scheduled follow-up.

Verification checklist

• The domain resolves to the intended service
• The expected index page loads over HTTP and HTTPS
• Application URLs and assets use correct paths
• The selected PHP value persists after refresh
• No new fatal error appears in logs
• The selected account item, identifier, domain, invoice, user, or service matches the original request.
• The portal no longer shows a pending or failed action unless delayed processing is expected and documented.
• An independent customer-side test produces the expected result.
• Related billing, security, notification, routing, and renewal settings remain correct.
• The previous value and rollback information are retained until the change is proven stable.

Common failure patterns

Use the symptom to narrow the investigation. Do not apply every possible fix at once.

• Files are uploaded one directory above or below the web root
• An index file has incorrect case
• DNS still points to the old provider
• A plugin or component is incompatible with the selected PHP version
• Several runtime settings are changed at once
• The correct value was saved on the wrong item, environment, domain, mailbox, record, user, or billing account.
• A dependent setting, external provider, cache, client, or application continues to use an older value.
• The item is pending, suspended, expired, unpaid, cancelled, locked, or otherwise not eligible for the requested action.
• The change completed, but verification reused an authenticated session or cached result that hid the actual behavior.
• A temporary error was treated as permanent and followed by multiple conflicting edits.

Security, privacy, and data handling

• Use a unique password stored in a reputable password manager and enable two-factor authentication for every account user who can access it.
• Give each person an individual account user instead of sharing the owner login. Remove access promptly when responsibilities change.
• Do not send passwords, two-factor recovery codes, full payment-card data, private keys, API secrets, or unredacted identity documents in a normal support reply.
• Review unexpected sign-in notices, permission changes, domain changes, payment changes, and credential resets as possible security events.
• After an access-related incident, rotate affected credentials, review delegated users and contacts, verify domains and DNS, and document the recovery actions.
• Limit access to hosted websites and web roots to people who need it and review that access when responsibilities change.
• Before sharing evidence involving domains, aliases, document roots, index files, logs, application URLs, and website-level settings, redact information that is not required to diagnose the issue.

When to contact Spark Rack

Contact Spark Rack when the account shows a failed or inconsistent provider-side action, the required control is missing despite confirmed eligibility and permission, data restoration or protected logs are required, an unauthorized change is suspected, or the problem remains after a controlled rollback and independent verification.

• The account email address and the exact service, domain, invoice, ticket, or other item involved
• The page and section used, including the tab or subtab, without copying session tokens from the address bar
• The expected result and the actual result in separate sentences
• The approximate time of the last successful use and the first failure, including the time zone
• The exact error text, response code, bounce text, or visible status
• The changes made immediately before the issue, including old and new values when known
• The devices, browsers, networks, applications, or external tools used to reproduce the issue
• The troubleshooting steps already completed and the result of each step

Closeout and ongoing care

After Websites & Web Roots is working, record the final state and remove any temporary access, files, forwarding paths, test records, or broad permissions that were created for the work. Review related expiration dates, renewal settings, payment status, contacts, and alerts so a future administrative event does not recreate the problem.

Keep the support history, change record, and safe verification evidence for as long as they are operationally useful. Periodic review is especially important after staff changes, migrations, major application updates, domain renewals, payment-method changes, or security incidents.