Volume replication API

Creates a regional replication relationship and requests its initial copy.

Send exactly one source_volume_id or source_group_id, target region, target_storage_policy_id, and optional interval_seconds. Interval defaults to 300 and accepts integers 60–86400.

Returns 202 with relationship ID, allocated target IDs, and initial-copy operation. A policy or encryption failure rejects admission; transfer failure remains visible on the accepted operation.

Returns enabled state, last captured checkpoint, last completed target checkpoint, backlog, target health, and ETag. Requires volumes:read. Missing observations do not advance the completed target timestamp.

Updates the relationship's enabled state or interval.

Requires If-Match. Changes enabled state or interval. It does not rewrite the source identity or replace a promoted target's new writes.

Promotes a completed recovery checkpoint after source fencing.

Requires recovery permission, the relationship ETag, selected checkpoint, and verified source fencing reference. Accepted work returns 202 with a promotion operation.

An incomplete checkpoint returns 409 recovery_point_incomplete. Unresolved source ownership returns 409 fencing_required. Promotion stops copying into that target and establishes its writable recovery role; it does not preserve writes that never reached the checkpoint.

Removes a replication relationship under an explicit target-retention decision.

Deletion requires resolved transfers and an explicit retained-target decision. A promoted target remains an independent resource. Reversing direction requires a reviewed resynchronization relationship rather than toggling an old source field.

See cross-region replication, consistency groups, and fencing.