Netbackup Recovery without Import

The goal of this document is to establish the procedure to allow for quicker recovery of a system or Veritas Netbackup™ domain for backup tapes without importing the individual backup images.

Warning: There are some very important caveats; please carefully read the entire procedure before attempting it.

Note: Implied in this procedure, is that the receiving site has a fully functional master. If this is being done as part of a disaster recovery plan, there are many other steps that are required and which must be addressed. This procedure is only intended to allow tapes to be moved between master servers.

There are many different ways to recover data on a different master than the one used for the original backups. Several of these methods are documented by consulting, and one method -- catalog replication using Veritas Volume Replicator™ -- is being formalized in the near future.

This document will give a specific procedure to follow along with the restrictions and caveats. This should allow sysadmins to accomplish their goals in a manner that will be safe, reliable, and supported by Veritas.

If data is being moved from one active NetBackup domain to another for recovery purposes only or being sent to a disaster recovery site, this procedure can be used. You must follow all the restrictions and caveats.

Restrictions

  1. All tapes being moved must have unique barcodes.
  2. All client names must be unique.
  3. None of the tapes being moved will be used for backups at the new location.
  4. There exists a library at the new location to support the correct tape format.
  5. Both masters must be running the same release of NetBackup.
  6. Both masters must be of similar architecture (both UNIX or both Windows).

Procedure

  1. Perform a catalog backup on both masters. This will allow you to recover from any errors that might occur during this process.
  2. Configure a special volume pool at the new site that will never be used for backups.
  3. Setup barcode rules for the incoming tapes to put them in the special volume pool.
  4. Ensure there are no active backups for the desired clients. One way to ensure this is to stop the NetBackup services/daemons on the master server.
  5. Copy the desired client directories at the originating site. This would be the contents of /usr/openv/netbackup/db/images/<client name> or \Program Files\VERITAS\netbackup\db\images\<client name>
  6. You can now move the client data (tapes) and catalog data to the new site.
  7. Merge the copied images into the catalog at the new location.
  8. Put the tapes in the library at the new location.
  9. To ensure none of the moved tapes can be overwritten, you might want to write-protect them.
  10. Inventory the robot containing the moved tapes.
  11. Put the appropriate FORCE_RESTORE_MEDIA_SERVER entry or entries in bp.conf. When you put this entry in bp.conf, you will need to recycle bprd.
  12. If you moved primary copies of the tapes, then you are ready to restore. If you moved non-primary tapes, then you will need to change the primary copy for all the images being moved.

Caveats

Since this procedure assumes you will not be using the moved tapes for backups, it did not require moving the volume database or the media database. If these tapes are to be returned to the original site, the image directory will not be gracefully cleaned up. When the image or images expire, they will be deleted from the image directory but there will be an error logged in the bptm log that states: "Could not find media ID xxxxxx in database, nothing to delete."

It would be advisable to delete the images from the catalog, preferably by removing the data when the restores are finished and the tapes are removed from the library. When you eject the tapes, you should also consider removing the entries from the volume database. If you want to permanently move the tapes and images, then you are looking at a master merge which is a documented service offered by VERITAS Enterprise Consulting Services.

To make this entire process more easily managed, it would be advisable to ensure that the polices are correctly configured on the originating master.

Some things to consider are:

  • having the target backups not multiplexed to ensure only the desired images are sent offsite.
  • directing the target backups to a specific unique volume pool should be done to make it easier to identify and track the required media.

This will also ensure that only the desired images are taken offsite.

Another way to achieve this would be to duplicate the images that are going to be sent to the new location using either Vault of bpduplicate and directing the duplicate images to a special volume pool.

Careful planning in the configuration at the originating site can make this process work more smoothly.