Blog / WordPress /

A restore rehearsal for a WordPress site

The backup plugin says “successful”. Green tick, nice date. Has anyone ever restored the site from it?

A backup you haven’t restored is a hope, not a backup. A restore rehearsal tells you whether the copy is complete, reachable and usable before an incident forces you to find out the hard way. Plan it with the site owner and run it somewhere safe.

Define what must be restored

List the site files, database, uploads, configuration and external services that matter. Confirm where backups are stored, who can access them and how often they run. A database restore may not recover uploaded media or configuration; clarify the scope. Protect backup credentials and do not place them in a shared report.

Choose a staging or isolated environment and avoid overwriting production data. Before I start a rehearsal, I write down what success means: the site boots, key pages load, the administrator can sign in, forms behave safely and the data matches the chosen restore point.

Run and record the rehearsal

Follow the documented procedure and time each major step without treating a single rehearsal as a guaranteed recovery time. Check representative pages, media and important workflows. Confirm that email, payment and analytics integrations are disabled or safely pointed to test services. Record missing steps, permissions or dependencies.

Improve the recovery plan

Update the instructions and assign an owner for gaps. Schedule a new rehearsal after major infrastructure changes.

Protect the rehearsal from real-world side effects

A restore may bring back old settings that reconnect to production services. Before rehearsal, identify email, payment, analytics, webhooks and scheduled tasks. Disable or redirect them safely so test activity cannot charge a customer, send a real notification or alter production data. Ask the site owner which integrations can be used in staging.

Treat restored data as sensitive. Limit access, use an isolated environment and delete the copy under the approved retention process afterwards. Keep database dumps out of chat and tickets. If the backup includes user information, apply the organisation’s privacy and security controls.

Test recovery details, not just the homepage

Check a page with media, a recent post, navigation, forms and administrator access. Confirm that permalinks and required configuration work after restoration. If a plugin license or external credential is needed, record the dependency securely rather than placing secrets in the rehearsal report. Note which steps require a person with privileged access.

Compare the restored result with the expected recovery point and document any missing content. A rehearsal should end with a cleanup and access review.

Verify the recovery point

A restore rehearsal should test the backup the organisation would actually depend on, not a convenient local copy. Confirm the backup date, contents and integrity before starting. If the site changes frequently, note the potential gap between the backup and the latest content. Decide who would assess whether that gap is acceptable during an incident.

Test a restoration from the documented storage location and verify any encryption or access step. A backup that only one person can retrieve is a dependency worth addressing. Keep recovery credentials in the approved secure system and make sure more than one authorised person knows how to access the process without sharing passwords informally.

Include the human recovery steps

Some recovery needs DNS changes, hosting access, licence keys or vendor help. Document who owns each dependency and how to reach them, without putting private contact details in a public note. If a provider is needed, check the account is active and the support route is known.

After the rehearsal, return the test environment to its safe state and document the cleanup. Review the process with the site owner and assign fixes. A rehearsal that uncovers gaps has done its job, as long as each gap gets an owner and a next step. I’d rather find them on a quiet afternoon than during an outage, and I don’t trust a recovery plan that lives in one person’s head.

Test that the restored site can be maintained

A restored site should be usable by the people responsible for operating it. Confirm administrators can sign in, update content and access the tools needed to monitor the site. Check that scheduled tasks, permalinks, media paths and required plugins work in the isolated environment. If a license or external service is unavailable, record the dependency and its owner.

Do not assume that a successful homepage load proves the restore is complete. Test a recent article, a representative media file, a key form in safe mode and any custom post type the business relies on. Verify that content and settings correspond to the chosen recovery point.

Know when a rehearsal should stop

A restore exercise should have a safe stop condition. If it unexpectedly connects to a production payment service, sends real email or modifies live data, stop the test and involve the administrator. Do not try to fix a risky connection by changing credentials without the service owner. Document what happened and update the isolation checklist before the next rehearsal.

If the backup is corrupt or incomplete, do not keep trying unverified copies on production. Find the approved recovery source and ask the hosting or backup owner to confirm integrity. A rehearsal can reveal that the process needs outside support; record the contact route and dependency.

After the test, delete temporary credentials and return services to their intended state. Confirm production was unaffected. So when would you stop your rehearsal? Write that rule down before you start.