Quick recovery is a method that directly uses backup data stored in the third-party backup app as the VM's data source, allowing the VM to power on and become operational immediately. Depending on the specific use case, quick recovery supports the following two recovery modes:

1. Quick recovery without data restoration: Create a VM and adjust the VM disk settings to reference the backup disk file stored on the NFS datastore. It is suitable for scenarios where only the availability of the backup data needs to be verified (e.g. VM1 in Figure 3‑1 Principles of VM Quick Recovery).
2. Quick recovery with data restoration: Based on the first method, this approach also merges the backup data from the NFS datastore into the VM's current disk. It is suitable for general backup recovery scenarios (e.g. VM2 in Figure 3‑1 Principles of VM Quick Recovery).

The principles of VM quick recovery are as follows:

Figure 3‑1 Principles of VM Quick Recovery

1. Prerequisites

The VM must be powered off.

2. Execution

(1)The third-party backup app packages the backup data into virtual disk files in raw or qcow2 format (with a cluster size of at least 64 KB), and places these files in the NFS datastore.

(2)Call the OpenAPI (SCP: POST /janus/20210725/storages/attachnfs, HCI: POST /hci/vt/api/v1/datastores/actions/attachnfs) to mount the NFS datastore to the HCI VM.

(3)Call the OpenAPI (SCP: POST /janus/20180725/servers/{server_id}/rebase-backing-file, HCI: POST /hci/vt/api/v1/vms/{id}/actions/rebasebackingfile) to change the base image of each VM disk to the virtual disk file stored on the third-party NFS datastore.

(4)At this point, the VM has completed quick recovery and is marked as a "quickly recovered VM". It can now be powered on normally. To recover the backup data to production storage, continue with the steps below:

(5)Call the OpenAPI (SCP: /janus/20180725/servers/{server_id}/merge-backup-data, HCI: POST /hci/vt/api/v1/vms/{id}/actions/startmergebackupdata) to start merging the backup data from the NFS datastore into the VM's disk.

(6)Query the VM details and wait until the VM transitions from a "quickly recovered VM" to a "non-quickly recovered VM". This indicates that the backup data has been fully merged into the production storage, and the VM no longer depends on the NFS datastore.

(7)Unmount the NFS datastore from the VM.

Notes:

1. When the VM is classified as a "quickly recovered VM", it relies on the backup data on the NFS datastore. Attempting to unmount the NFS datastore from the VM in this state will trigger an error indicating that the storage resource is currently in use.
2. Forcibly unmounting the NFS datastore from the quickly recovered VM may cause the VM to become inaccessible due to loss of access to its disk data. To safely unmount the NFS datastore, follow this procedure: (1) Power off the VM. (2) Delete the VM. (3) Unmount the NFS datastore from the VM.