If a backup or import fails, first check which operation ran and whether it saved anything. Keep the exact error and the time it occurred before trying again.
A browser error does not always mean the server stopped. Repeating an operation without checking can create duplicate records, replace data, or interrupt recovery.
Identify the operation and its result
Use the page where you started to find the latest result:
- Tools → Backups: Creates or uploads Cedros backup bundles. A restore from this page replaces selected live data with a recovery point.
- Tools → Import: Applies a Cedros site-setup JSON package, such as collection definitions, navigation, and settings.
- Tools → Migrate: Imports WordPress content or supported database records into a selected destination.
These files are not interchangeable. Use a Cedros backup ZIP in Backups, a Cedros setup JSON package in Import, and a supported export in Migrate.
- Confirm you are on the intended site and reopen the relevant page to read its current state.
- Record the operation, error message, time and time zone, and backup ID or source filename. For an import, also record the destination or collection.
- Check whether it is still running, completed, or failed. For an import with an uncertain result, inspect the destination for records or settings already saved.
- Resolve the reported cause before retrying. If the state is unclear, ask your site operator to check the operation first.
For logs and connection checks, see Checking site health and investigating errors. Do not include passwords, connection strings, encryption phrases, or private file contents in a support message.
When a backup does not start or finish
Backup now is missing. Finish the backup encryption setup first. Store the recovery phrase securely before completing setup. See Setting up automatic backups.
The controls are disabled or read-only. Wait for loading to finish and check whether a backup is already running. Backup settings and backup operations use different permissions; ask your administrator to check your access if the required action remains unavailable.
Backup in progress remains visible. Creating a large backup can take time. Check the reported start time and current operation state. Avoid starting another run while it is active. If progress appears stuck, give your operator the start time and any error or log details so they can check the running job, database, and storage.
The request was accepted. Starting a job is not the same as completing a backup. Wait for Backup completed and confirm the expected backup appears in the retained list. Use Creating and downloading a backup for the full completion checks.
Backup failed. Read the specific error. A database dump, encryption, local storage, optional Vault or mail capture, or offsite upload can fail independently. Ask your operator to check the named dependency, available disk space, access, and configuration. Keep existing recovery points while investigating.
The offsite upload failed. A recoverable local copy may remain even though the operation reports failure. After the storage problem is fixed, a subsequent backup run can finish the pending upload before creating a new recovery point. Check the backup ID and timestamp afterward; a successful retry may have completed the earlier backup rather than captured the latest data. Do not remove pending backup files manually.
Backups are Local only. This means the listed backup is on the server; it does not establish an offsite copy. Check the remote-storage status and any error before relying on offsite recovery.
A scheduled backup is overdue. Open Schedule and check Automatic daily backups, the saved time, time zone, and any save error. The scheduler must be running for an overdue job to start. If schedule saving failed, resolve the error and use Try again; do not assume the visible draft settings were saved. Persistent overdue status needs an operator check.
When a backup download or upload fails
A download appeared successful, but the file is missing. Check your browser's downloads list and destination folder. The page can hand the bundle to the browser without proving that the browser finished saving it. Check for a blocked, cancelled, or incomplete download before trying again.
The download reports an error. Record the backup ID and whether it is local, remote, or both. Remote-only downloads depend on the configured storage being reachable. Ask your operator to investigate an inaccessible or incomplete recovery point.
The upload rejects the file. Use the original Cedros backup ZIP. A file ending in .zip still needs a valid backup manifest and the expected artifacts. Do not rename a database dump, setup JSON, or WordPress export to make it look like a backup.
The manifest or artifact size does not match. Obtain a fresh copy of the original bundle and compare its downloaded size with the expected file. Do not edit the manifest to bypass validation. If the original bundle is also rejected, preserve it and ask your operator to investigate.
The backup ID already exists. Check the retained list for the earlier upload. A retry may be unnecessary. Do not delete a usable recovery point or change the bundle's internal ID simply to clear this error; have your operator resolve any conflict.
An uploaded bundle is registered for recovery. Uploading it does not restore the site or prove that its encrypted contents can be restored with the available phrase.
When a site-setup import fails
For Tools → Import, follow Exporting and importing a site's setup.
Invalid import JSON. Re-copy the entire Cedros setup package into Setup contents. Check that it is JSON from the setup export, rather than a backup bundle or a data export. Valid JSON syntax alone does not establish that it is a supported setup package.
The package format is unsupported or newer than the site supports. Ask your operator to check compatibility between the source and destination. Do not change the format number to bypass the check.
A custom schema or collection rule conflicts. Have the reported incompatibility reviewed. The option to replace collection rules only controls those rules; it does not prevent changes to included menus, redirects, site settings, or templates.
The result says Preview — nothing was changed. That was a preview, not an applied import. Confirm the package and intended replacement choices before turning off Preview the import without making any changes and selecting Import Setup.
Run Preview Import again after changing the JSON or options. A summary or older preview still on screen is not confirmation that the latest input was checked. If an applied import disconnects or returns an uncertain result, inspect the destination before submitting it again.
When a content or database import fails
For Tools → Migrate, use Migrating from WordPress or Importing supported database data for the matching source.
Review or Import stays disabled. Supply your own source data, select the correct destination, and run Preview. Import requires a nonempty preview with no row issues. Changes to the source, destination, or mappings clear the previous preview.
No matching items or a missing field. Check the export type and field mappings. A WordPress pages-only export produces no Blog items. Enter source field names in mappings, not example values, and correct any misspelled or nested field path.
The batch is too large. Split it into smaller batches. Each run supports at most 500 records, including file-based imports. Keep the batches distinct so a later run does not repeat earlier records.
A content path already exists. Inspect that entry first. Exclude content already imported, or choose a different path for different content. Content import does not overwrite an existing article at that path.
Custom data replaced another record. Check the destination collection and Unique ID mapping. Matching destination keys replace the existing payload. Distinct source IDs can normalize to the same key, and fallback keys such as row-1 can repeat between batches. Correct the keys and recover overwritten information from the source or backup before continuing.
The connection or media download failed. Check source access and the reported URL or query with your operator. For unavailable database connections, use a supported export file. Media must be reachable by Cedros; a successful preview does not prove its download will succeed.
Some records appeared before the error. Migrate can save earlier items before a later item fails. Compare the source batch with the actual destination, then prepare a batch containing only what remains. Start over clears the form; it does not undo imports. A warning that history could not be saved also does not, by itself, mean the import failed.
When restore or recovery needs an operator
Use Restoring a site from a backup for restore preparation and the deployment-specific recovery method. Restore is not a troubleshooting shortcut for an import error; it can replace changes made since the selected backup.
If a bundle cannot be decrypted, confirm that the correct recovery phrase is available through your secure recovery process. A new phrase does not decrypt an older bundle. Do not send the phrase in a support ticket.
If validation rejects a bundle as incomplete or inconsistent, keep the original artifact and have it reviewed. If Restore is disabled on the deployment or a selected recovery target is unavailable, ask the operator to use the appropriate recovery procedure.
If a restore fails or is interrupted, read the reported stage and maintenance requirement. When the page says to keep the deployment in maintenance mode, have the operator verify or complete recovery before serving traffic again. Do not clear that state merely to make the site appear online.
After any fix, verify the intended outcome: a completed, identifiable recovery point; the expected imported settings or records; or a recovered site whose important workflows work. Save the result alongside the original error so the incident has a clear resolution.