Backup Process

A backup for OpenStack images or instances includes the following stages:

  1. For each instance, create a Nova snapshot.

    A Cinder snapshot is created for any volumes attached to the instance. Snapshots are crash consistent.

  2. Create volumes from Cinder volume snapshots.
  3. For an image backup, image reader is used to read the data directly and back it up to media.

    A full backup is always performed for images, regardless of the backup type that is specified for the backup job.

  4. For volume backups, each volume is attached on the proxy, and then data is read and backed up to media. The following limits apply:
    • For a VirtIO controller, up to 25 data volumes can be attached to the VSA proxy, occupying device slots vdb - vdz.
    • For a SCSI controller, up to 4 data volumes can be attached to the VSA proxy, occupying device slots sdb - sdf. This limit is a libvirt version 1.2.17 limitation with RedHat 7.2.

    These limits include any volumes already attached to the proxy. For example, if the VirtIO controller for a proxy already has vdb and vdc slots occupied, then only 23 data volumes can be attached.

  5. Back up the volumes, using CRC for incremental changes and directing data through the MediaAgent for the storage policy associated with the backup.
  6. Unmount, detach, and delete the volumes attached on the proxy created during the backup.
  7. Delete the snapshots created during the backup.

Last modified: 1/18/2019 3:52:41 PM