Keep a remote copy

Keep a copy of every object of a bucket on a storage cluster in another location, read from while the cluster is unavailable.

Last updated 18 November 2026View as Markdown

A bucket on a Deutsche Cloud storage cluster (see Storage locations) can also keep a copy of every object in the platform store of the cluster's remote zone; the answer's zone names it (for example de1b, Nuremberg). It is on by default for new buckets on a storage cluster that has a remote zone. Replace example-uploads with your bucket's name.

Prerequisites

  • Permission for s3:PutBucketRemoteCopy on the bucket (s3:GetBucketRemoteCopy to read it).
  • The bucket is on a storage cluster that has a remote zone (available: true in dcloud s3 bucket-remote-copy get).

Turn the remote copy on or off

Not available in Console. The remote copy is set with the CLI and the API.

Result

  • Each changed object is queued for copying when it is written, encrypted with the bucket's own key; a full comparison once an hour finds what was missed. The copy mirrors each key's current state, not its version history. pendingKeys and failedKeys say how many still wait or failed.
  • If the whole storage cluster is unavailable, objects are read from the copy. An object not copied yet, listings, versions and every write answer 503 ServiceUnavailable until the cluster is back, never 404.
  • Turning it off deletes the copy after 7 days. Without it, an outage of the cluster makes the bucket answer 503 until the cluster is back.
  • Turning it on copies every existing object too.

If it is refused

409 This bucket’s storage has no remote location: a remote copy is offered for buckets on a storage cluster. available: false in the answer means the bucket's storage keeps no copy elsewhere (buckets in a zone, own hosts).

Did this page help you?