Incremental backup chain restore blocked by missing parent backup in Ceph backend

Context

Cinder incremental backups on the Ceph/RBD backend form a parent-child chain where each child backup depends on its parent for a full restore. The control plane records the relationship, but the backend objects themselves must remain intact.

Problem

A restore request for a child incremental backup fails when the parent backup has been deleted or its underlying Ceph objects are no longer accessible. The failure occurs even though the child backup record shows available status and the destination volume is in available state. The exact error condition surfaced by cinder-backup varies by release and driver version, and the documentation does not specify whether a partial restore or chain reconstruction is ever possible.

Goal

Determine whether a child incremental backup can be restored independently when its parent is missing, and identify the precise failure mode operators should expect in the logs.

Questions

1. What specific error code or log message does cinder-backup emit when a parent backup is absent during a child restore attempt?
2. Is there any supported workflow to reconstruct or bypass the missing parent link without a full-chain backup?
3. Does the behavior differ between the Ceph/RBD driver and other incremental-capable drivers such as NFS or S3?