Move an offsite backup repository without losing the restore path

Write the cutover contract

Identify whether the source is an ordinary file collection or a repository managed by a backup tool. A repository can contain indexes, locks, configuration, and encrypted data objects whose names are not meaningful files. Copying a visible folder alone may omit the objects that make snapshots readable.

Record the source endpoint, bucket, prefix, expected object count, total bytes, repository identifier, tool version, credential owner, and available migration window. Identify ongoing writers and maintenance jobs. Preserve the original repository while the destination is being qualified.

• Repository scope
• Active writers
• Maintenance schedule
• Transfer budget
• Credential owner
• Rollback destination

Make the first transfer reversible

Use an initially empty destination prefix and a transfer mode whose deletion behavior you understand. rclone's copy command does not delete destination files, whereas sync can delete files to make the destination match the source. The official command references describe that difference. A copy still needs review because it can replace objects with matching names.

NordenVault's cloud-migration page describes an import workflow from S3-compatible and file-sharing providers. Treat that as a provider-specific starting point. Check whether the import preserves the complete repository object set, versions, and required metadata before selecting it for a backup repository.

Reconcile the final state

Plan a final period in which repository writers and destructive maintenance are stopped or coordinated through a supported migration procedure. Transfer the remaining objects, record failures, and compare the final inventory. A count or byte total alone cannot establish that every required object is readable and complete.

Use NordenVault's documentation for the destination endpoint and client setup, then use the backup tool's repository checks and restore procedure. Do not run two independent maintenance jobs against copied repositories merely because both endpoints appear reachable. Keep the cutover and maintenance ownership explicit.

Switch only after a destination restore

Restore a chosen snapshot into a new directory through the destination credentials. Open representative files and compare them with the expected recovery point. Record elapsed time and any missing permissions, metadata, or decryption dependencies. If the result fails, keep writes on the source and investigate the destination copy.

After a successful cutover, confirm the next scheduled backup reaches the new repository and can also be restored. Retain the original according to the rollback plan, with a named review date and owner for eventual retirement.