Continuing Backups and New Base Backups After Master Server Recovery

There are many circumstances where you are required to run a base backup or where base backups are strongly recommended.

Any server recovered with BMR requires a new base backup to continue protection; this includes the master server if it was recovered with BMR.

Any server that is virtualized with either IV, FV, or RRP needs a new base backup to continue protection.

Agentless recovery, either instant or full, has no CBT tracking enabled and triggers a new base backup. Note that if the backing store for your virtual machines is NFS, there are known VMware conditions outside of DPX control where the base backup transfers data equal to the total size of the provisioned VMDK, regardless of thin provisioning or actual used VMDK space.

In a block-level environment, circumstances may warrant performing new base backups, which can cause the backup storage to fill up. In these cases, protecting the existing data with tape and starting new backups to empty storage should be considered.

For NetApp storage, perform an NDMP SMTAPE backup of volumes. Once the backup is completed, these volumes can be removed and new volumes created for new jobs. Removing these volumes eventually causes condense failures as the condense looks for these snapshots for removal, however, these failures are generally benign. If necessary, the SMTAPE data can be recovered and various operations such as management console restore, BMR, and manual IA map procedure can be employed to recover data.

For open storage, you can protect an entire storage volume with image backup. Once the backup is completed these volumes can be deleted. Removing these volumes eventually causes condense failures as the condense looks for these snapshots for removal, however, these failures are generally benign. If necessary, the image backup can be recovered to a new volume or alternate open storage server. That data can then be used for management console restore via alternate secondary if that catalog still contains references to those snapshots. BMR can also be used to recover data.

NDMP backups are strongly recommended for running a base backup, regardless of storage vendor or data format. Some data formats are naturally full-base only. Others that support incremental data formats may do so in a vendor specific way which prevents the data protection software from capturing an accurate set of data. This accounts for the gap experienced from reverting the master server Catalog.

File-level backups can continue to run as incrementals. However, for consistency of Catalog and ease of understanding from the management console, base backups are strongly recommended.

In a real disaster where both master server and block-level storage is affected, consider the time and space needed to execute restore and then make the necessary space available for new backups. Recover your disk-based storage one volume at a time, use that data to recover clients and data, either with BMR or the alternate secondary feature of IA and virtualization, and then set aside new volumes to run new base backups for your servers. Do not try to recover disk-based storage with the intention of continuing incremental backups; in many cases the software prevents this and forces a base backup.