Red Hat [Electronic resources] : The Complete Reference Enterprise Linux Fedora Edition؛ The Complete Reference نسخه متنی

Backups

You can back up and restore your system archive tools like

tar , restoring the archives later. For backups,

tar is usually used with a tape device. To automatically schedule backups, you can schedule appropriate

tar commands with the cron utility.

Amanda

To back up hosts connected to a network, you can use the Advanced Maryland Automatic Network Disk Archiver (Amanda) to archive hosts. Amanda uses

tar tools to back up all hosts to a single host operating as a backup server. Backup data is sent by each host to the host operating as the Amanda server, where they are written out to a backup medium such as tape. With an Amanda server, the backup operations for all hosts become centralized in one server, instead of each host having to perform its own backup. Any host that needs to restore data simply requests it from the Amanda server, specifying the file system, date, and file names. Amanda has its own commands corresponding to the common backup tasks, beginning with "am", such as

amdump ,

amrestore , and

amrecover . Configuration files are placed in /etc/amanda and log, and database files in /var/adm/amanda.

Backups with dump and restore

You can back up and restore your system with the dump and restore utilities. dump can back up your entire system or perform incremental backups, saving only those files that have changed since the last backup. dump supports several options for managing the backup operation, such as specifying the size and length of storage media (see Table 27-11).

dump Levels

dump uses dump levels to determine to what degree you want your system backed up. A dump level of 0 will copy file systems in their entirety. The remaining dump levels perform incremental backups, backing up only files and directories that have been created or modified since the last lower-level backup. A dump level of 1 will back up only files that have changed since the last level 0 backup. The dump level 2, in turn, will back up only files that have changed since the last level 1 backup (or 0 if there is no level 1), and so on up to dump level 9. You could run an initial complete backup at dump level 0 to back up your entire system, and then run incremental backups at certain later dates, having to back up only the changes since the full backup.

Using dump levels, you can devise certain strategies for backing up a file system. It is important to keep in mind that an incremental backup is run on changes from the last lower-level backup. For example, if the last backup was 6 and the next backup was 8, then the level 8 would back up everything from the level 6 backup. The sequence of the backups is important. If there were three backups with levels 3, then 6, and then 5, the level 5 backup would take everything from the level 3 backup, not stopping at level 6. Level 3 is the next- lower-level backup for level 5, in this case. This can make for some complex incremental backup strategies. For example, if you want each succeeding incremental backup to include all the changes from the preceding incremental backups, you could run the backups in descending dump level order. Given a backup sequence of 7, 6, and 5, with 0 as the initial full backup, 6 would include all the changes to 7, because its next lower level is 0. Then 5 would include all the changes for 7 and 6, also because its next lower level is 0, making all the changes since the level 0 full backup. A simpler way to implement this is to make the incremental levels all the same. Given an initial level of 0, and then two backups both with level 1, the last level 1 would include all the changes from the backup with level 0, since level 0 is the next lower level—not the previous level 1 backup.

Recording Backups

Backups are recorded in the /etc/dumpdates file. This file will list all the previous backups, specifying the file system they were performed on, the dates they were performed, and the dump level used. You can use this information to restore files from a specified backup. Recall that the /etc/fstab file records the dump level as well as the recommended backup frequency for each file system. With the

-W option, dump will analyze both the /etc/dumpdates and /etc/fstab files to determine which file systems need to be backed up. The

dump command with the

-w option just uses /etc/fstab to report the file systems ready for backup.

dump Operation

The

dump command takes as its arguments the dump level, the device it is storing the backup on, and the device name of the file system that is being backed up. If the storage medium (such as a tape) is too small to accommodate the backup, dump will pause and let you insert another. dump supports backups on multiple volumes. The

u option will record the backup in the /etc/dumpdates file. In the following example, an entire backup (dump level 0) is performed on the file system on the /dev/hda3 hard disk partition. The backup is stored on a tape device, /dev/tape.

dump -0u -f /dev/tape /dev/hda5

The storage device can be another hard disk partition, but it is usually a tape device. When you installed your system, your system most likely detected the device and set up /dev/tape as a link to it (just as it did with your CD-ROMs). If the link was not set up, you have to create it yourself or use the device name directly. Tape devices can have different device names, depending on the model or interface. SCSI tape devices are labeled with the prefix st, with a number attached for the particular device. st0 is the first SCSI tape device. To use it in the

dump command, just specify its name.

dump -0u -f /dev/st0 /dev/hda5

Should you need to back up to a device located on another system on your network, you would have to specify that hostname for the system and the name of its device. The hostname is entered before the device name and delimited with a colon. In the following example, the user backs up file system /dev/hda5 to the SCSI tape device with the name /dev/st1 on the rabbit.mytrek.com system:

dump -0u -f rabbit.mytrek.com:/dev/st0 /dev/hda5

The

dump command works on one file system at a time. If your system has more than one file system, you will need to issue a separate

dump command for each.

Recovering Backups

You use the

restore command to either restore an entire file system or to just retrieve particular files.

restore will extract files or directories from a backup archive and copy them to the current working directory. Make sure you are in the directory you want the files restored to when you run

restore .

restore will also generate any subdirectories as needed.

restore has several options for managing the restore operation (see Table 27-12).

To recover individual files and directories, you run

restore in an interactive mode using the

-i option. This will generate a shell with all the directories and files on the tape, letting you select the ones you want to restore. When you are finished,

restore will then retrieve from a backup only those files you selected. This shell has its own set of commands that you can use to select and extract files and directories (see Table 27-13). The following command will generate an interactive interface listing all the directories and files backed up on the tape in the /dev/tape device:

restore -ivf /dev/tape

This command will generate a shell encompassing the entire directory structure of the backup. You are given a shell prompt and can use the

cd command to move to different directories, and the

ls command to list files and subdirectories. You use the

add command to tag a file or directory for extraction. Should you later decide not to extract it, you can use the

delete command to remove from the tagged list. Once you have selected all the items you want, you enter the

extract command to retrieve them from the backup archive. To quit the restore shell, you enter

quit . The

help command will list the restore shell commands.

If you need to restore an entire file system, you would use restore with the

-r option. You can restore the file system to any blank formatted hard disk partition of adequate size, including the file system's original partition. If may be advisable, if possible, to restore the file system on another partition and check the results.

Restoring an entire file system involves setting up a formatted partition, mounting it to your system, and then changing to its top directory to run the

restore command. First you should use

mkfs to format the partition where you are restoring the file system, and then mount it onto your system. Then you use restore with the

-r option and the

-f option to specify the device holding the file system's backup. In the next example, the user formats the /dev/hda5 partition, and then restores on that partition the file system backup, currently on a tape in the /dev/tape device.

mkfs /dev/hda5
mount /dev/hda5 /mystuff
cd /mystuff
restore -rf /dev/tape

To restore from a backup device located on another system on your network, you would have to specify that hostname for the system and the name of its device. The hostname is entered before the device name and delimited with a colon. In the following example, the user restores a file system from the backup on the tape device with the name /dev/tape on the rabbit.mytrek.com system:

restore -rf rabbit.mytrek.com:/dev/tape