UnixWare bare metal restore from the backup system

This section describes the method for performing a bare metal restore from a backup system with the network support systems. The simplest way to perform a bare metal restore is to select RESTORE Backups to SAME Hard Disk, FULLY AUTOMATED, from the main menu. If files need to be restored, they can be restored using the RESTORE Backups Now option.

Before the restore job begins, make sure that all parameters are set correctly. If not, the settings can be changed from the menu. If View Files Locally is set to yes, the files can be viewed while they are being restored. However, this may cause a reduction in the restore speed.

If the Air-Bag diskettes have been created using a two-filesystem configuration (/dev/root and /dev/u mounted on /usr), and is being converted to a one-filesystem configuration, manually edit the file /etc/AIRBAG.MOUNTS. This file contains the filesystems that the Air-Bag expects to be mounted when it performs the restore. Simply delete the line containing the filesystem that will no longer be in use under the new configuration. Do not make any other modifications to the file or the restore will fail.

When converting from a single-filesystem configuration to a dual-filesystem configuration, add the entry for the second filesystem to the /etc/AIRBAG.MOUNTS file on the CD. The line must have six tokens; the last being the disk number on which the filesystem is found. For the filesystem /dev/u mounted on /usr, use the following command:

/dev/u /usr 54389 K disk1

The header of the file /etc/AIRBAG.MOUNTS should be left untouched. For convenience, copy the line representing the single filesystem (/dev/root):

/dev/root / 41990 K disk1

and change each of the tokens to the appropriate value. The second token represents the name of the directory on which this filesystem is mounted and must be accurate. The Air-Bag uses this to restore data to the filesystem.

The finished output looks like this:

FILE MOUNTED FS SIZE DISK

SYSTEM ON TYPE in K NO

========================================================

/dev/root / vxfs 41990 K disk1

/dev/u /usr ufs 54389 K disk1

The number representing the size in kilobytes does not have to be exact. It only serves as a reminder when using the SliceManager to decide how to allocate space among the filesystems.

Not every combination of hard disks and controllers can be anticipated. There is a good chance that if the CDs were created on a system with a different combination of controllers than when booted, the disk drive mapping can be different.

Let us suppose that there are three SCSI controllers, each with one hard disk (at ID 0) and suppose that the third SCSI controller and the disk drive on it becomes corrupt due to a lightning strike. A replacement hard-drive should be added to the second SCSI controller and make its SCSI ID 0. When booting from the Air-Bag diskettes, it will look for the hard-drive on device /dev/rdsk/c3b0t0d0s0. However, since the controller is no longer present, no device is found there. Instead, that device will be found at /dev/rdsk/c2b0t1d0s0 (SCSI ID 1). The Air-Bag will have recorded all of its database information and mount points as starting with /dev/rdsk/c3b0t0d0XX.

The disk drive name mapping can be changed by using the apply_map utility which checks the file /etc/MAPFILE for mapping information. This file must be manually added to the A2 diskette when it is generated or it must be created on the fly when the system is booted from the Air-Bag. Using the example above, the following entries should be placed in this file:

c3b0t0d0 c2b0t1d0

Execute the apply_map utility from the shell. This informs Air-Bag that the hard-drive at c3b0t0d0 has moved to c2b0t1d0. All previous references to c3b0t0d0 are converted. Thus, all features of the Air-Bag, such as the totally automated restore will work as expected.

To reverse the change, run

apply_map -r

This feature is provided as a convenience. With HOT swappable disks, mirroring, volume managers, and fail-over clustering, there is an almost endless combination of hardware possibilities. Since each system crash is unique, this simple tool provides a flexible way to manage the mapping needed to restore the system. It is fairly easy to create this map. If it does not work, the changes can be removed with the -r option.