Backing Up and Restoring Files

Selecting a Backup Strategy and Media

Your Linux system's hard disk contains everything needed to keep the system running, as well as other files, such as documents and databases you need to keep your business running. You need to back up these files so that you can recover quickly and bring the system back to normal in case the hard disk crashes. Typically, you have to follow a strict regimen of regular backups because you can never tell when the hard disk might fail or the file system might get corrupted. To implement such a regimen, you need to decide which files you want to back up, how often, and what backup storage media to use. This is what I mean by selecting a backup strategy and backup media.

Your choice of backup strategy and backup media depends on your assessment of the risk of business disruption due to hard disk failure. Depending on how you use your Linux system, a disk failure may or may not have much impact on you.

For example, if you use your Linux system as a learning tool (to learn about Linux or programming), all you may need are backup copies of some system files required to configure Linux. In this case, your backup strategy can be to save important system- configuration files on one or more floppies every time you change any system configuration.

On the other hand, if you use your Linux system as an office server that provides shared file storage for many users, the risk of business disruption due to disk failure is much higher. In this case, you have to back up all the files every week and back up any new or changed files every day. You should perform these backups in an automated manner (this is where you can use the job-scheduling features from earlier in this chapter). Also, you probably need a backup storage medium that can store large amounts (multiple gigabytes) of data on a single tape. In other words, for high-risk situations, your backup strategy is more elaborate and requires additional equipment (such as a tape drive).

Your choice of backup media depends on the amount of data you have to back up. For a small amount of data, such as system-configuration files, you can use floppy disks as the backup media. If your PC has a Zip drive, you can use Zip disks as backup media; these are good for backing up a single-user directory. To back up servers, you should use a tape drive, typically a 4-mm or 8-mm tape drive that connects to a SCSI controller. Such tape drives can store several gigabytes of data per tape, and you can use them to back up an entire file system on a single tape.

When backing up files to these backup media, you have to refer to the backup device by name. Table 20-6 lists device names for some common backup devices.

Taking Stock of Commercial Backup Utilities for Linux

This chapter describes how to back up and restore files using the tape archiver (tar) program that comes with Red Hat Linux. Although you can manage backups with tar, a number of commercial backup utilities come with GUIs and other features to simplify backups. Here are some well-known backup utilities for Red Hat Linux:

• BRU-Backup and Restore Utility from The TOLIS Group, Inc. (http://www.tolisgroup.com/)

• LONE-TAR-Tape-backup software package from Lone Star Software Corporation (http://www.cactus.com/)

• Arkeia-Backup and recovery software for heterogeneous networks from Knox Software (http://www.knox-software.com/)

• CTAR-Backup and recovery software for UNIX systems from UniTrends Software Corporation (http://www.unitrends.com/products.html)

• BrightStor ARCserve Backup for Linux-Data protection technology for Linux systems from Computer Associates (http://www3.ca.com/Solutions/Product. asp?ID=3370)

Using the Tape Archiver-tar

You can use the tar command to archive files to a device such as a floppy disk or tape. The tar program creates an archive file that can contain other directories and files and (optionally) compress the archive for efficient storage. The archive is then written to a specified device or another file. In fact, many software packages are distributed in the form of a compressed tar file.

The command syntax of the tar program is as follows:

tar options destination source

Here, options are usually specified by a sequence of single letters, with each letter specifying what tar should do. destination is the device name of the backup device. And source is a list of file or directory names denoting the files to back up. Optionally, you can add a hyphen prefix for the options.

tar zcvf /dev/fd0 /etc/X11

tar ztf /dev/fd0

tar zxvf /dev/fd0 /

Backing Up and Restoring a Multivolume Archive

Sometimes the capacity of a single storage medium is less than the total storage space needed to store the archive. In this case, you can use the M option for a multivolume archive-meaning the archive can span multiple tapes or floppies. Note, however, that you cannot create a compressed, multivolume archive. You have to drop the z option. To see how multivolume archives work, log in as root, place one disk in the floppy drive, and type the following tar command:

tar cvfM /dev/fd0 /usr/share/doc/ghostscript*

Note the M option in the option letters; it tells tar to create a multivolume archive. The tar command prompts you for a second floppy when the first one is filled. Take out the first floppy, and insert another floppy when you see the following prompt:

Prepare volume #2 for `/dev/fd0' and hit return:

When you press Enter, the tar program continues with the second floppy. In this example, you need only two floppies to store the archive; for larger archives, the tar program continues to prompt you for floppies if more floppies are needed.

To restore from this multivolume archive, type cd /tmp to change the directory to /tmp. Then type:

tar xvfM /dev/fd0

The tar program prompts you to feed the floppies as necessary.

du -s /etc
14532   /etc

The resulting output shows that the /etc directory requires at least 14,532K of storage space to back up. If you plan to back up on multiple high-density floppies, you need about 14,532/1,440 = 11 floppies.

Backing Up on Tapes

Although backing up on tapes is as simple as using the right device name in the tar command, you do need to know some nuances of the tape device to use it well. When you use tar to back up to the device named /dev/st0 (the first SCSI tape drive), the tape device automatically rewinds the tape after the tar program finishes copying the archive to the tape. The /dev/st0 device is called a rewinding tape device because it rewinds tapes by default.

Secret

If your tape can hold several gigabytes of data, you may want to write several tar archives-one after another-to the same tape (otherwise, much of the tape may be empty). To do this, you do not want the tape device to rewind the tape after the tar program finishes. To help you with this, there are several Linux tape devices that are nonrewinding. The nonrewinding SCSI tape device is called /dev/nst0. Use this device name if you want to write one archive after another on a tape.

After each archive, the nonrewinding tape device writes an end-of-file (EOF) marker to separate one archive from the next. You can use the mt command to control the tape-you can move from one marker to the next or rewind the tape. For example, after you finish writing several archives to a tape using the /dev/nst0 device name, you can rewind the tape with the following command:

mt -f /dev/nst0 rewind

After rewinding the tape, you can use the following command to extract files from the first archive to the current disk directory:

tar xvf /dev/nst0

After that, you must move past the EOF marker to the next archive. To do this, use the following mt command:

mt -f /dev/nst0 fsf 1

This positions the tape at the beginning of the next archive. You can now use the tar xvf command again to read this archive.

Performing Incremental Backups

Suppose that you back up your system's hard disk on a tape by using tar. Because such a full backup can take quite some time, you do not want to repeat this task every night. (Besides, only a small number of files may have changed during the day.) You can use the find command to list those files that have changed in the past 24 hours:

find / -mtime -1 -type f -print

This command prints a list of files that have changed within the last day. The -mtime -1 option means that you want the files that were last modified less than one day ago. You can now combine this find command with the tar command to back up only those files that have changed within the last day:

tar cvf /dev/st0 `find / -mtime -1 -type f -print`

When you place a command between single back quotes, the shell executes that command and places the output at that point in the command line. The net result is that the tar program saves only the changed files in the archive. Thus, you get an incremental backup that includes files that have changed since the previous day.

Performing Automated Backups

Earlier in this chapter, you learn to use crontab to set up recurring jobs (called cron jobs). The Linux system performs these tasks at regular intervals. Backing up your system is a good use of the crontab facility. Suppose your backup strategy is as follows:

• Every Sunday at 1:15 a.m., your system backs up the entire disk on the tape.

• Monday through Saturday, your system performs an incremental backup at 3:10 a.m. by saving only those files that have changed during the past 24 hours.

  To set up this automated backup schedule, log in as root and type the following lines in a file named backups (this example assumes that you use a SCSI tape drive):

  15 1 * * 0 tar zcvf /dev/st0 /
  10 3 * * 1-6 tar zcvf /dev/st0 `find / -mtime -1 -type f -print`

  Next, submit this job schedule by using the following crontab command:

  crontab backups

  Now, you should be set for an automated backup. All you need to do is place a new tape in the tape drive everyday. You should also give each tape an appropriate label.