One of the more common tasks of a system administrator is helping users recover lost or corrupted files. To perform that task effectively, you must set up procedures for backing up files at frequent and regular intervals. This chapter describes how you use resident commands and utilities to back up (archive) and restore files and directories.
Design and implement a disaster recovery plan that describes how you intend to restore your entire operating system and user files to normal operations in the event of a catastrophic failure. This chapter does not describe the disaster recovery process, because it is often very specific to site operations and business requirements. However, backup operations are an important component of such a plan.
The following topics are included in this chapter:
Section 9.1 is an overview of the steps and options involved in creating a backup.
Section 9.2 describes the main tasks involved in creating a backup.
Section 9.3 describes how you set up a backup schedule.
Section 9.4 describes the methods of creating a backup.
Section 9.5 provides information that enables you to prepare for a backup, such as references to other documents that you might need to read, system files created, related utilities, and prerequisite tasks.
Section 9.6
describes the use of the
dump
command to perform a backup.
Section 9.7
describes the use of the
restore
command to recover data from a backup.
Section 9.8 describes the use of commands that enable you to archive individual files and directories, rather than complete file systems.
Section 9.9
describes the use of
dxarchiver, a graphical interface for archiving files and directories.
Section 9.10 describes how you create a bootable tape. This is a bootable backup of the root file system and key system files that might be useful for disaster recovery.
9.1 Understanding Backup Tasks
This chapter describes basic backup operations for a system using the UFS file system. You might also need to use other backup and restore utilities if any of the following conditions apply to your local system:
If you are using the Advanced File System (AdvFS) file system exclusively, or if you are using AdvFS domains on some of the disks attached to your system, refer to the AdvFS Administration guide. Using the AdvFS file system provides you with more backup features, such as the ability to clone domains. One of the disadvantages of the UFS file system is that you must prevent access to a UFS file system during a backup. If a user accesses a file during a backup is in process, the backup might not record changes in the file. To ensure a completely accurate back up of a UFS file system, you might need to take a disk off line or shut the system down to single user mode. If you are unable to schedule system shut downs, consider using the AdvFS file system.
If you are using the Logical Storage Manager (LSM), refer to the Logical Storage Manager guide. Using features of LSM such as mirroring volumes, you might also be able to overcome some of the backup limitations of UFS. For example, you can take an instant, accurate snapshot of a UFS file system by mirroring the file system on a different disk. You can then break the mirror at any time to create an archive, with only a brief pause in system operations. (Using LSM requires spare disk capacity and might be unsuitable for small systems with few disks.)
If you want to back up and restore a root volume to a different system, consider using configuration cloning. This feature is described in the Installation Guide -- Advanced Topics. Configuration cloning enables you to recreate a customized operating system on another processor in the event of a disaster, or to recreate an environment on one or more systems.
This chapter describes only those backup and archiving utilities that are provided in the base operating system when installed.
The Associated Products CD-ROM might include additional backup applications (which might require additional licenses). Refer to the Installation Guide for information. Refer to the documentation that comes with your backup application for information on using third-party products.
The main tasks comprising backup and restore operations are:
Creating your data recovery and disaster recovery plans
Backing up data
Choosing a backup schedule
Creating small archives by using the
pax,
tar,
and
cpio
commands or the associated graphical
user interface,
dxarchiver
Performing a full UFS backup using the
dump
utility
Performing an incremental backup
Performing a remote backup
Using backup tools
Restoring data
Restoring files from small archives
Restoring a file system from a dump
Restoring a dumped file system on a new partition
Restoring files
Restoring files interactively
Performing remote restorations
Restoring standalone systems from bootable tape
9.2 Backing Up Data and System Files
For basic backup, you can use the
dump
and
restore
commands.
See
dump(8)
for full details of all command options
that are supported.
The operating system also provides graphical and command
line tools for archiving and for creating a bootable tape of the standalone
system (SAS).
Prevention of data loss is an important part of any backup and recovery strategy. There are many tools for system monitoring that you can configure to help prevent situations that might result in data loss. For example, some systems support environmental monitoring, and there are tools to test and exercise peripherals. There are also the event and error logging systems that you can configure to monitor the system for priority events such as a backup failure. See Chapter 13 for information on using EVM (event management) to set up the event reporting strategy for your system and site. You can also use EVM to report on the success of your backups, ensuring that you do not miss a scheduled backup event.
It is important that all the files on your system, user files and system files, are protected from loss. Back up your entire system, including the system software. Many system files are static; that is, after you install them they remain unchanged. Therefore, you do not need to back up system files as frequently as data files. Incremental backups are also possible, and you might consider implementing them if your data changes significantly in a short period.
Each file system backup is a single process.
To ease the backup process,
organize your file systems so that dynamic files are on file systems that
you back up regularly and static (system or program) files are on file systems
that you back up occasionally.
You might find that you have dynamic files
on file systems that you back up occasionally.
If this happens and you need
to back them up regularly, just prior to performing a backup, copy the frequently
changing files to systems that you back up regularly.
This allows you to
back up those files without backing up an entire file system.
You can also
write shell scripts to automate these tasks and use the
cron
command to automate the schedule.
See
cron(8)
for more information on scheduling
tasks.
9.3 Choosing a Backup Schedule
To decide how often to back up each file system, consider the balance between the potential loss of user time and data and the time it takes you to perform backups. Ask yourself the question, "How much information can I afford to lose?" The answer to this question helps you determine your minimum backup interval. On most systems the backup interval is daily, but you can choose any other interval.
It is not necessary to back up all the files in a file system at each
backup.
Back up only those files that changed since the previous backup; this
is called an incremental backup.
Using the
dump
and
restore
commands, you can perform up to nine levels of incremental
backups.
For example, while a level 0 dump backs up an entire file system,
a level 1 dump backs up only those files changed since the last level 0 dump,
and a level 7 dump backs up only those files changed since the last lower
level dump.
To integrate incremental backups into your file backup schedule, you need to balance the time and tape space required for backup against the amount of time it could take you to restore the system in the event of a system failure. For example, you could schedule backup levels following the 10-day sequence:
[0 1 2 3 4 5 6 7 8 9]
On the first day you save an entire file system (level 0). On the second day you save changes since the first backup and so on until the eleventh day when you restart the sequence. This makes the amount of time spent and data saved on each backup relatively small each day except the first; however, if a system failure on the tenth day requires that you restore the entire system, you must restore all ten tapes.
Most systems follow some variant of the common Tower of Hanoi backup schedule. Once monthly you make a level 0 dump to tape of all the file systems that you backup regularly. Once weekly, you make a level 1 dump to start a daily sequence of:
[...3 2 5 4 7 6 9 8 9 9 ...]
If you do backups only once a day on the weekdays, you end up with a monthly backup schedule as follows:
[0 1 3 2 5 4 1 3 2 5 4 ...]
This
schedule, although slightly complex, requires that you restore at most four
tapes at any point in the month if a system failure corrupts files.
Of course,
doing a level 0 dump daily requires that you restore at most one tape at any
point, but requires a large amount of time and tape storage for each backup.
On most days in the Tower of Hanoi schedule, you require very little time
and tape storage space for the backup.
9.4 Backup Methods
Depending on your needs and your local system configuration, there are several options for backing up data, as follows:
You can run the following command-line interfaces from a terminal:
dump,
rdump,
restore, and
rrestore
tar,
pax, and
cpio
Use these to create quick file archives or to create scripts that you
run with the
cron
scheduler.
The bootable tape feature,
bttape, is a SysMan
Menu application that you can invoke from the command line, the SysMan
Menu, or from CDE.
Depending on how you invoke it, it either runs the command-line
interface or a graphical interface that is appropriate to your windowing environment.
(See
Chapter 1
for more information.) The commands are
btcreate
and
btextract.
Use the bootable tape feature to create a bootable tape for recovery and to back up critical system data and customized system files. This feature also enables you to use any terminal and a number of windowing environments, and is therefore recommended for remote operations.
From the CDE folder Application Manager - System Admin,
open the Storage Management folder and click on the Bootable Tape icon.
This
action invokes the graphical interface to the
bttape
utility.
From CDE, open the Application Manager pop-up menu from the front panel and open the Desktop_Tools folder to use the following utilities:
Archive - For quick archiving of files and folders, such as when archiving projects or user accounts. The related interfaces, Archive List Contents and Archive Unpack, enable you to manage these archives. These are simple graphical interfaces with minimal options.
From the CDE Application Manager - System Admin folder, open the DailyAdmin folder to use the Archiver utility. The Archiver is a graphical interface to the command-line tools that enables you to select archive type and options such as compression. This interface allows you to drag and drop entire file systems or directories (folders) into the backup set.
Some tools provide you with additional options when you run
them as superuser (root).
9.5 Preparing to Perform a Backup
This section contains information that you might need to prepare for a backup. Also included is a list of utilities that can assist you in preparing for a backup, and a list of prerequisite tasks.
Chapter 6
contains information on the UFS file
system.
Chapter 5
contains information on using disk and
tape devices and on determining which disk and tape devices you want to back
up.
Also, refer to the information about the
cron
command
in
Chapter 3
for information on scheduling regular
backups.
The following documentation contains other information that you might
need to perform a backup:
Books
The AdvFS Administration and Logical Storage Manager guides contain information on the AdvFS file system and LSM storage management features.
The owner's manual for any peripherals used (such as tape drives) contain important information. These documents provide you with information on storage volume, media type, compression densities, and general operating instructions for a device.
Reference pages
The
dump(8),
rdump(8),
vdump(8), and the associated
restore(8)
and
rrestore(8)
reference
pages provide information on the basic utilities for dumping file systems
to tape and restoring them back to disk.
The
tar(1),
pax(1), and
cpio(1)
reference pages provide information on basic utilities for
creating and manipulating archive files.
The
btcreate(8),
btextract(8), and
bttape(8)
reference pages provide information on the bootable tape interfaces.
The
cron(8)
and
crontab(1)
reference pages provide information on creating
cron
entries for backup scripts that execute at specific dates and
times.
The
mcutil(1)
reference page describes the media changer manipulation utility.
Both the Archiver and Bootable Tape graphical user interfaces provide online help that describes your options, and defines what data you can enter into the data fields in each window.
Apart from the file system that you specify and the archive files created, the following files are used or created when you create backups:
The
dump
and
restore
commands create or use the following files:
/etc/dumpdates
- Contains a list
of file systems that were backed up, the date that each file system was backed
up, and the backup level
/tmp/rstdir*
- Lists directories
stored on the default tape
/tmp/rstmode*
- Records the owner,
permission mode, and timestamps for stored directories
./restoresymtab
- Holds information
required during incremental restore or
rrestore
operations
The bootable tape feature creates or uses the following files:
/var/adm/btcreate.log
- Provides
a log of the
btcreate
process
/usr/lib/sabt/sbin/custom_install.sh
-
Specifies which files are added to the miniroot
/usr/lib/sabt/etc/addlist
- A data
file that specifies which files and directories are added to the miniroot
file system that is created on the bootable tape
/usr/lib/sabt/etc/fslist
- A data
file that specifies which file systems are backed up
/usr/run/bttape.pid
- A lock file
that prevents multiple instances of the
btcreate
utility
The following utilities are useful when performing backups:
The SysMan Station provides a graphical view of the storage devices available on the system. Use this interface to help you identify disk and tape devices and find their device names.
The CDE Application Manager -- Desktop_Tools folder provides
a Disk Usage tool that runs the
du
command and returns
statistics on disk usage.
Use the Folder Size option to check the size (in
blocks) of any directory, such as
/usr/users.
Command-line
utilities
du
and
df
provide the same
data.
The CDE Application Manager -- DailyAdmin folder provides the System Information interfaces, a graphical view of system resources such as file space usage. You can set this monitor to flash a visual warning when your preset file space limits are exceeded. You can also use the SysMan Station to monitor file systems as described in Chapter 1.
The Event Manager (EVM) provides a way of monitoring file system limits and alerts you of problems or can automatically start backups and cleanup of file systems.
The command-line interfaces
dsfmgr
and
hwmgr
enable you to query the system for information about devices,
such as device names and disk partition size.
You can also get information from the
diskconfig
GUI, which you can invoke from CDE Application Manager - Configuration
folder, or from the SysMan Menu.
This interface provides size information
in megabytes, bytes, and blocks.
(The
disklabel
command
provides a command-line disk configuration interface).
The following prerequisite tasks apply to all the backup methods:
Become familiar with using the interfaces and the sources of information about the commands. Ensure that this information is available to you when the system is down. You must often perform recovery operations in single-user mode, and reference pages might not be available.
Ensure that all the required products or utilities are installed and configured (if necessary). The simplest way to do this is to refer to the reference page for information on invoking the tool, and run a test by invoking command-line interfaces with null input, or by starting up the graphical user interfaces.
Verify that the tape hardware is installed and configured.
If you are unsure, you can use the
/usr/field/tapex
tape
exerciser and refer to the hardware documentation for other test features.
See also the hardware information tools listed in
Section 9.5.2
Check the size of the directories that you want to back up. For example, you can use the following commands:
# df /usr
Filesystem 512-blocks Used Available Capacity Mounted on /devices/disk/dsk0g 1498886 688192 660804 52% /usr
# du -s -x /usr/users 1835 /usr/users
You can also use the graphical or command-line tools listed in Section 9.5.2.
Obtain sufficient quantities of the correct media, ensuring that there is enough storage volume for the files that you intend to back up. This also applies if archiving to disk or any other writeable media, such as WORM drives or magneto-optical floppy drives.
Identify the files or directories that you intend to work with, and choose appropriate names for the archives. You might need some temporary scratch disk space if assembling different directories into a single volume before archiving (although you can do this direct to the archive from the command line or by adding directories to existing archives). Refer to the documentation for the backup utility that you choose to use. Some tools provide default file names and locations. For example, the bootable tape interface prompts you for the following file names. (You can accept the default or provide another file name):
/usr/lib/sabt/etc/fslist
- A data
file that specifies which files and directories are added (appended) to the
miniroot
/usr/lib/sabt/etc/addlist
- A data
file that specifies which file systems are backed up
The Archiver requires the following files:
One or more source files or directories.
In CDE, directories
are identified as folders, and you can drag and drop them into the Archiver
window from File View windows instead of typing long pathnames such as
/usr/lib/sabt/sbin.
A destination file, such as
/usr/backups
for a tar file on disk, or the device name for a tape device, such as
/dev/tape/tape0_d0.
(You do not need to supply an extension or suffix
for the archive file name.
The utilities listed in
Section 9.5.2
can assist you in finding the required device information, particularly if
more than one tape drive is attached to a system.)
If you are restoring (unpacking) an archive, you need to supply
the archive name, such as
/usr/archives/userfiles_990802.Z
or
/dev/tape/tape0_d0, for a tape archive.
The device name for the device or devices that you want to access, and any associated device special file. For example, the following are valid device names and device special files:
| Device name | Device Special File | Description |
dsk0a |
/dev/disk/dsk0a |
Partition a of disk number 0 |
disk1b |
/dev/rdisk/dsk1b |
Partition b of raw disk 1 |
tape0c |
/dev/tape/tape0c |
Default density rewind tape (with compression) |
tape0_d0 |
/dev/ntape/tape0_d0 |
Nonrewind tape device 0. The _d0 suffix specifies the density |
Device names are located in the
/dev
directory under
the
/disk,
/rdisk,
/tape,
or
/ntape
subdirectories.
You can also use the graphical
or command-line tools listed in
Section 9.5.2
to locate devices
and match them with their device names.
Note
Tape devices often support different densities and compression options that enable you to put more information into a single archive. See
tz(7) for information on tape density options, and how you select them by specifying different device names.
Full backups might require that you shut down the system.
You can back up the system while in either multiuser mode or single-user mode.
However, backups performed on file systems actively being modified might
corrupt the backup data.
The
dump
command operates by
checking the inodes of the files you want to back up.
The inodes contain
data such as table entries and other statistics.
If you use the
dump
command to back up files in a file system, an inode is attached
to each file.
If the system or user activity changes a file after the inode
data is recorded, but before the file is backed up, it might corrupt the backup.
To shut down the system, unmount a file system, and check the integrity of a file system:
Shut down the system by using the SysMan Menu General Tasks
option, or with the
/usr/sbin/shutdown
command.
For example,
to shut down the system in 5 minutes and give users periodic warning messages,
enter:
# /usr/sbin/shutdown +5 'System going down to perform backups'
Refer to Chapter 2 for more information on shutting down the system.
Use the
umount
command with the
-a
option to unmount the file systems that you want to back
up:
# /sbin/umount -a
The root file system remains mounted.
Use the
fsck
command to ensure the integrity
of the file system.
For example, use the following command to check a file system on the c partition (the whole disk):
# /sbin/fsck -o /dev/disk/dsk0c
The
dump
command copies all designated file systems
or individual files and directories changed after a specified date to a file,
pipe, magnetic tape, disk, or diskette.
Refer to
AdvFS Administration
for information
on copying AdvFS file systems.
You must have superuser privileges to use the
dump
command.
Note
To produce valid backups on a file system, you must back up a file system while it is inactive. It is recommended that you unmount the file system and check it for consistency. As an added precaution, put the system into single-user mode before starting your backup operations. This is not true for AdvFS.
9.6.1 Performing a Full Backup
Set
up a schedule for performing a full backup of each file system on your entire
system, including all the system software.
A conservative schedule for full
system backups is to do one with each normal level 0 dump (using Tower of
Hanoi, once a month), but you can set any schedule you like within the reliability
of your storage media, which is about two years for magnetic tapes.
To back
up your file system, use the
dump
command.
See
dump(8)
for a description of the command options that you use to specify the characteristics
of the tape device, such as block size, tape storage density, and tape length.
Specify the file system with a full pathname when you use the
dump
command.
The
dump
command can back up only a
single file system at a time, but there might be several
dump
processes simultaneously writing files to different tape devices.
The following list describes the most commonly used options to the
dump
command:
-integerSpecifies the dump level as an integer (0-9).
A dump level
of 0 causes a full dump of the specified file system.
All other dump levels
cause an incremental backup.
That is, only files that have changed since
the last dump of a lower dump level are backed up.
The
/etc/dumpdates
file contains a record of when the
dump
command
was used on each file system at each dump level.
The
-u
option to the
dump
command updates the
dumpdates
file.
-f
dump_fileWrites the dump to the device specified by
dump_file
instead of to the default device,
/dev/tape/tape0_d0.
If you specify the
dump_file
as a dash
( - ), the
dump
command
writes to the standard output.
-uUpdates the
/etc/dumpdates
file with the time of the dump and the dump level
for the file system in the backup.
You use this file during incremental dumps
(by using the dump level option) to determine which files have changed since
a particular dump level.
You can edit the
/etc/dumpdates
file to change any record or fields, if necessary.
See
dump(8), which describes
the format of this file.
To back up your entire file system to the default backup device, use
the
dump
command for each file system on your machine.
The
dump -0u
command option causes a level 0 dump
and updates the
/etc/dumpdates
file with the time and date
of the backup for each file system.
This creates an initial point on which
to base all future incremental backups until the next full or level 0 dump.
Each file system must be backed up individually.
For example, if you want to perform a level 0 dump of the root,
/usr, and
/projects
file system partitions, follow
these steps:
To back up the root file system, load a tape into your tape drive and enter:
# dump -0u /
After completing the backup, remove the tape from your tape drive.
To back up the
/usr
file system, load a
new tape into your tape drive and enter:
# dump -0u /usr
After completing the backup, remove the tape from your tape drive.
To back up the
/projects
file system, load
a new tape into your tape drive and enter:
# dump -0u /projects
You can either back up each file system on an individual tape,
or you can back up multiple file systems on one tape by specifying the no-rewind
device,
/dev/ntape/tape0_d0, as the output device.
The
following examples show the root,
/usr, and
/projects
file systems being backed up on one tape:
# dump -0uf /dev/ntape/tape0_d0 / # dump -0uf /dev/ntape/tape0_d0 /usr # dump -0uf /dev/ntape/tape0_d0 /projects
This example might require additional media management
to cross-reference dump files with tapes, especially when a single dump file
spans media.
Exercise care when labeling this type of backup media.
9.6.2 Performing an Incremental Backup
Set up a routine as part of your backup schedule to make it easier to remember which backup to do each day. Include a mechanism for logging your backups and their dump level and for listing the tapes on which they are made. Because of the chance of system corruption, do not keep this information on the local computer system.
After you establish a system for making incremental backups, the procedure
is simple.
Assume you use the following backup schedule to do a daily backup
of
/usr:
0 1 9 9 9 1 9 9 9 9 ...
On Monday, perform a level 0 dump:
# dump -0u /usr
On Tuesday, perform a level 1 dump:
# dump -1u /usr
The level 1 dump backs up all the files that changed since Monday. On Wednesday through Friday you perform a level 9 dump (which always backs up all the files that have changed since Tuesday's level 1 dump):
# dump -9u /usr
To perform the same level 9 dump to the
tape device named
/dev/tape/tape1_d0
instead of the default
tape device, use the
-f
option as shown in the following
example:
# dump -9uf /dev/tape/tape1_d0 /usr
The argument to the
-f
option specifies a tape device local to the system from
which you are performing the dumps.
9.6.3 Performing a Remote Backup
Some
machines in a networked system environment might lack a local tape drive that
you can use for making backup tapes.
You can use the
rdump
command to make backups on a remotely located tape device.
The
rdump
command is identical to the
dump
command except
that it requires the
-f
option to specify the machine
name and an attached backup device.
See
dump(8)
for a description
of the options to the
rdump
command.
The
rdump
command updates the
/etc/dumpdates
file on the local machine in the same way as does the
dump
command.
The
rdump
command starts a remote
server,
/usr/sbin/rmt, on the remote machine to access
the storage medium.
(This server process is transparent.
See
rmt(8)
for more information.)
To back up the
/projects
file system from bhost1
onto a tape drive on bhost2 with the attached backup device
/dev/rmt0h, enter the following command from bhost1.
The name of bhost1 must
be defined in the
/.rhosts
file of bhost2 to allow access.
# rdump -0uf bhost2:/dev/tape/tape0_d0 /projects
You can automate the backup process by using shell scripts.
The
cron
daemon can execute these shell scripts late in the evening
when there is less chance of the
dump
commands making errors
due to a changing system.
Backup shell scripts often perform the following tasks:
Determine the dump level
Warn the system of the dump
Make a listing of tape contents
Notify the operator upon completion
Some time during the day, load a tape into the tape drive.
At the
specified time, the
cron
daemon runs the backup shell scripts.
After the shell procedures are finished, remove the backup tape and archive
it.
Backup shell scripts are best used when the dump is small enough to
fit on a single tape.
You must specify the no-rewind device and the
-N
option to the
dump
command to inhibit
the tape from automatically going off line when each dump is completed.
After
the
dump
command reaches the end of the tape, it takes
the tape off line and you must replace the tape.
9.7 Restoring Data
Occasionally, you need to retrieve files from your backup tapes, and possibly need to restore entire file systems at some time. If you have set up a good backup procedure, then restoring files or full file systems is a simple task.
If a serious problem occurs, you might have to restore your entire system. Before restoring, determine what caused the problem with the system.
After determining the cause of the problem, reinstall your system from the initial boot tapes. The installation instructions that came with your system explain this procedure.
After your system is up and running, restore the system to the state
it was in just prior to the system crash.
If you are using AdvFS, use the
vrestore
command.
Refer to
AdvFS Administration
for information on restoring
the AdvFS file system.
If you used the
vdump
command
to back up a UFS file system, you can also use the
vrestore
command to recover it.
However, if you used the
dump
command
you must use the
restore
command to recover files.
Because
the
dump
command saves only a single file system at a time,
you must execute the
restore
command for each file system
you want to restore.
See
restore(8)
for information on the command
syntax.
9.7.1 Restoring a File System
This section describes a general procedure for restoring a file system, such as after a disk failure or other loss of data. To restore individual files, go to Section 9.7.2.
To restore a file system, create a new file system and restore the files from the dump files by using the following commands:
newfs
- Creates a new UFS file system.
See
newfs(8).
mount
- Mounts the file system, making
it available for general use.
See
mount(8).
cd
- Changes your current working
directory.
See
cd(1).
restore
- Restores archived files
from a backup media to a disk.
See
restore(8).
Refer to the AdvFS Administration guide for information on restoring an AdvFS file system.
If the disk does not have a label, write the label
by using the
disklabel
command before you create the new
file system.
See
disklabel(8).
Writing a label with customized partition table settings might affect the entire disk. Use the following command to write the default disk partition table:
# /sbin/disklabel -rw dsk1
Invoke the editing option of the
disklabel
command
to use the customized partition table settings.
Refer to
Chapter 6
for more information.
You can also use the
diskconfig
Disk
Configuration interface.
See
diskconfig(8).
The following example shows the commands you use to restore a file system
called
/usr/projects
from a tape:
# disklabel -rw dsk1 # newfs /dev/rdisk/dsk1c # mount /dev/rdisk/dsk1c /usr/projects # cd /usr/projects # restore -Yrf /dev/tape/tape0_d0
9.7.2 Restoring Files Manually
If users lose data files, they ask their system administrator to restore those files. Users might also ask you to restore an earlier version of a file. Whatever the reason for a file restoration, you must determine which tape contains the correct version of the file. Enquire when the file was lost and when it was last modified, you can use your backup log to determine which tape contains the most recent version of the wanted file.
Use the
-t
option with the
restore
command to determine whether a file is on the selected tape.
The
-t
option creates a list of files and directories on the tape.
For example, to list the contents of the
working
subdirectory
of the
/usr
file system on a particular backup tape, load
the tape and enter:
# restore -t ./working
To create a list of the entire contents of a backup tape, load the backup tape and enter:
# restore -t
Make a listing of each backup tape after you create it. This verifies a successful backup and gives you a place to look up what files are on the tape.
After determining the location of the file, create a new directory for the file. If you restore the file into an existing directory and the file already exists, the restored file overwrites the existing file.
For example, to restore the
working/old.file
file
from a
/usr
file system backup tape into your current directory,
load the backup tape and enter:
# restore -x ./working/old.file
To restore the entire contents of the working subdirectory from the same tape, enter:
# restore -x ./working
If your dump media contains multiple dump images, you need to know the sequence of the dump images in order to restore a file from one of the images. To examine the contents of the first dump image on the media, load the tape and enter:
# restore -ts 1
The
-s
option
followed by the number
1
specifies the first dump image.
For example, to restore the
working/old.file
file
from a
/usr
file system, which is the third dump image
on the backup tape into your current directory, load the backup tape and enter:
# restore -xs 3 ./working/old.file
9.7.3 Restoring Files Interactively
To ease the task of restoring multiple files, use the
-i
option to the
restore
command.
This option
starts an interactive
restore
session.
The interactive
mode has commands similar to shell commands.
To begin an interactive
restore
session, enter:
# restore -i
The system responds with the following prompt:
restore >
The following command-line options are available in the interactive restore mode:
ls
[
directory
]Lists files in the current or specified directory.
Directory
entries end with a slash (/).
Entries that are marked for
reading begin with an asterisk (*).
cd
[
directory
]Changes the current directory to the directory specified by the directory argument.
pwdLists the pathname of the current directory.
add
[
files
]Adds the files in the current directory or the files specified by the
files
argument to the list of files recovered from the tape.
Files are marked with an asterisk (*) if they are identified
as
"to be read"
by the
add
command.
You see
this asterisk when you use the
ls
command to list files.
delete
[
files
]Deletes all the files in the current directory or the files specified by the files argument from the list of files recovered from the tape.
extractRestores from the tape
the files that are marked
"to be read"
into the current working
directory.
The
extract
command prompts you for the logical
volume that you want to mount (usually 1), and whether the access modes of
the dot (.) current directory are affected; answer
yes
when you are restoring the entire
root
directory.
setmodesSets owner, access
modes, and file creation times for all directories added to the files-to-read
list; no files are recovered from the tape.
Use this command to clean up
files after a
restore
command is prematurely aborted.
verboseToggles verbose mode.
In verbose mode, each file name is printed to the standard output.
By default,
verbose mode is set to off.
This is the same as the
-v
command line option to the
restore
command.
helpLists a summary of the interactive commands.
?Lists a summary of the interactive commands.
whatLists the tape header information.
quitQuits the interactive restore session.
xitExits from the interactive
restore session.
The
xit
command is the same as the
quit
command.
To interactively restore the
./working/file1
and
./working/file2
files from a backup tape, load the
tape and enter:
# restore -i
After you switch to interactive mode, follow these steps to add the files to the list of files that you want to extract:
Change to the
working
directory:
restore > cd working
At the prompt, enter the file name as follows:
restore > add file1
Enter the name of the second file as follows:
restore > add file2
Extract the files as follows:
restore > extract
You are prompted for the logical volume you want to mount;
usually you respond to this prompt with
1
as shown in the
following example:
You have not read any tapes yet. Unless you know which volume your file(s) are on you can start with the last volume and work towards the first. Specify next volume #: 1
You are then asked whether the extract affects the access modes of the dot
(.) current directory.
For this example, reply with
n.
set owner/mode for '.'? [yn] n
After the files are extracted, quit the interactive session as follows:
restore > quit
The
file1
and
file2
files
are now in the current directory.
You can automate this procedure in a command file that is read by the
-F
option to the
restore
command.
For
example, the following command file, named
restore_file,
performs the restore operation shown in the previous example:
cd working add file1 add file2 extract 1 n quit
To read and execute this shell script, enter the following command:
# restore -iF restore_file
The result of the procedure
in this script is identical to that of the previous interactive restore session.
9.7.4 Restoring Files Remotely
You use the
rrestore
command to restore files to
local directories from a remote tape device.
The
rrestore
command requires the
-f
option to specify the machine
name and its backup device.
See
rmt(8)
for more information and
Section 9.7
for a description of the options to the
rrestore
command.
You must specify the name of the remote system where the backup device is attached, and the name of the backup device on that remote system in the format system:device
To restore the
./working/file1
file onto the local
directory on system1 from a backup tape mounted on system2 where the backup
device
/dev/rmt0h
is attached, enter the following command
from system1.
The name system1 must be in the
/.rhosts
file of system2 to allow access from system1 to system2.
# rrestore -xf system2:/dev/tape/tape0_d0 ./working/file1
The
rrestore
command starts a remote
server,
/usr/sbin/rmt, on the remote system to access the
storage medium.
9.7.5 Restoring or Duplicating a System (Root) Disk
In previous versions of the operating system, device names were assigned based on the physical location of the drive, according to the SCSI bus target. In Version 5.0 and higher, device names are assigned logically and stored in a database. They have no relationship to the bus address of the device. The device database must be recovered and possibly updated to successfully restore the root file system or if you want to move the root disk to a disk with larger capacity. You might also need to install devices (such as a tape device) to the device database when you restore the device from tape backup media.
After you reboot the system during the restoration, you might see the following message:
Unable to save existing hardware configuration. New configuration will be used
This message indicates that the device database is not recoverable and you must restore it.
The following procedure is a generic method for recovering or duplicating (cloning) a root disk. It covers the following possible scenarios:
The disk and the root partition are not damaged but you want to replace it with a different disk, possibly a different model with larger capacity.
The disk drive on which the root partition is located is damaged and you must:
Install a new disk drive, possibly of a different type and capacity
Choose an alternate disk drive that is already installed and available for use
The root (/), and possibly the
/usr
or
/var
file systems are corrupt, but the
disk drive on which they are located is fully functional.
Note
This procedure does not specifically address recovery methods from network backups and it does not address recovery of an AdvFS file system. See AdvFS Administration.
Depending on your knowledge of your system, you might not need to read all the following sections:
Section 9.7.5.1 - Describes how you prepare for a recovery
Section 9.7.5.2 - Describes the requirements for recovery
Section 9.7.5.3 - Describes the recovery (or duplication) procedure
Section 9.7.5.3 - Describes alternate procedures
9.7.5.1 Preparing for Recovery or Duplication
Depending on how your system is set up, and your level of system knowledge, you might need the following:
A replacement root disk drive. This procedure assumes that if the original root disk is unusable, you have either installed a new replacement disk, or you have decided to use an alternate disk that is already installed in your system. Install the drive as described in the owner's manual for the drive. The operating system automatically detects the drive.
There are steps in the procedure that assist you in identifying the new or alternate drive.
Verify that your system's firmware is current. You can obtain information and download kits from the Firmware Updates web page at the following URL:
http://ftp.digital.com/pub/DEC/Alpha/firmware
(You can also go to the Compaq Web site at http://www.compaq.com and select the Support option to search for information.)
Information about console commands.
You use Alpha System
Reference Manual (SRM) console commands at the system console prompt (>>>) to perform some tasks.
These commands are documented in the
owner's manual for your AlphaServer system.
If you cannot find the printed document, it is usually shipped as a printable file on a CD-ROM supplied with the system. If the CD-ROM is unavailable, you can often find the documentation at the Alpha Systems Technology web page at the following URL:
http://www.compaq.com/alphaserver/technology/index.html
The procedure instructions are typical for newer processors. If your system is older, you must consult the owner's manual and Installation Guide for your version of the operating system to obtain the actual commands and procedures.
The status of your system must be as described in
Table 9-1:
Table 9-1: Recovery Preparation
| Requirement | Description |
| A Full and Recent Backup | You need a full backup of all operating system
file sets that are on the root volume.
This might include root (/),
/usr, and
/var. |
| System Configuration | This procedure applies to all configurations
where there is a single disk drive used for the root partition which might
also contain the
|
| Logical Storage Manager | If you are using the Logical Storage manager (LSM), refer to the Logical Storage Manager guide for information on recovering the root volume. |
| User Interface | This procedure requires a console login. |
| Impact on System Availability | Except on clustered systems, loss of the root disk invariably involves one or more shutdowns and reboots of the system. This procedure is intended to help you restore full operation as quickly as possible. The time required for duplicating or recovering a disk depends on the disk size. |
| Privileges | You must be a root user with physical access to the system's storage array and backup devices |
9.7.5.2 Determining the Restoration Requirements
You might need the following resources to complete the restoration of your root disk. If are very familiar with your system's configuration, or if you have a recovery plan which records all the information you need to perform a recovery, you do not need to read this section. You might need the following items:
Distribution Media for the Operating System.
You use the installation shell to restore the root disk.
The installation
shell is a compact version of the operating system from which you can execute
commands, such as
mount.
The shell is packaged with the
operating system software as part of the distribution kit.
Your local site might also provide a Remote Installation Service (RIS) server from which you can boot your system across the network. If RIS services are available in place of CD-ROM media, follow your site-specific procedures and consult the Installation Guide.
CD-ROM Drive Name or Network Device Name.
To restore the root volume, you boot your system from a CD-ROM drive or a network device. You might need to refer to the owner's manual for your system to find the correct commands. Typically, you determine the CD-ROM device name at the console prompt as follows:
>>> show device | grep -E 'RR|CD' dka400.4.0.5.0 DKA400 RRD47 1206
You typically determine the network device name as follows:
>>> show config | more
After you
enter the preceding command, the complete system configuration is displayed
one page at a time.
Scroll down to the section headed
Slot Option
and locate the network device.
Network devices are typically named
ew*
or
ei*, where
*
is a letter.
For example:
11 DE500-BA Network Con ewa0.0.0.0.11.0 08-00-99-1Z-67-BB
For information on booting your system from a RIS server, refer to the Installation Guide -- Advanced Topics guide. For a full discussion of the System Reference Manual (SRM) console device naming conventions, see the owner's information for your system.
Boot Device Name
Determine the default boot device name as follows:
>>> show bootdef_dev bootdef_dev dka0.0.0.5.0
In
this example the default boot device is
dka0.
Note
If you are using Fibre Channel the name of the boot device is as you defined it during configuration of the storage devices.
If the current root device is usable and you are restoring to the same device, you use the device name later in the restore procedure. If you intend to install a new disk or use an alternate, you must specify the name of the disk. You determine the alternate by translating its b/t/l into the boot device name during the restore procedure.
Backup Media and the Restore Device
Depending on what file systems were on your original root disk, you
might need full and current backup tapes for the root (/),
/usr
and
/var
file systems.
In cases where you are duplicating (cloning) a disk, such as to increase the disk space available by using a disk with larger capacity, you can opt to back up directly from the source disk to the target disk.
Disk Label for the Target Disk
If the original drive is usable, you can choose to restore the root file system to the same drive. If the drive is damaged, you must select an alternate drive or install a new drive. The alternate or new drive must have enough storage capacity to hold the restored file systems and it must be partitioned to hold the restored file systems.
If the original root drive held custom partitions, restore the custom disk label or at least plan to select partitions that can adequately contain the restored file systems (and possibly allow for future expansion of those file systems). Depending on what data is stored on the original disk, you might need to plan for the following partitions:
The 256MB
a
partition to hold the root
(/) file system.
If the
/usr
file system is on the root
drive and you also need to restore it, you need a partition at least large
enough to contain the restored file system.
(Consider expansion requirements,
if appropriate.)
If the
/var
file system is on the root
drive and you also need to restore it, you need a partition at least large
enough to contain the restored file system unless it is included in
/usr
on the original root drive.
If primary or tertiary swap partitions were on the original root drive, you must recreate these partitions on the replacement drive.
You can restore other file systems as required, or restore them to different
devices and then remount them by updating the
/etc/fstab
file after the restoration is complete.
See
disklabel(8)
for more information on creating a disk label with custom
partitions.
9.7.5.3 Applying the Procedure
Some steps in the procedure are dependent on your system's original configuration. Ignore these steps if they do not apply to your configuration. The optional steps are marked [Configuration Dependent].
In the procedure, you always proceed to the next step unless redirected.
Boot the system from the operating system distribution media by using one of the following methods:
Insert the distribution CD-ROM that contains the operating system version that you wish to restore and boot the operating system specifying the CD-ROM reader device name that you determined previously. For example:
>>> boot dka400
Boot from your local RIS server. For example:
>>> boot ewa0
[Configuration Dependent] If you are already using the character-cell installation procedure, go to step 3, otherwise complete the following task.
If your system has a graphics console, the installation defaults to graphical mode. Wait until the installation procedure displays a dialog box titled Installation Welcome.
Pull down the File menu and select the Exit option to invoke character-cell mode.
Verify the status of the backup device and the target disk (the restored root disk) by using the following command:
# hwmgr view devices
The
hwmgr
command
displays a list of all devices currently recognized by the system as shown
in the following example:
HWID: Device Name Mfg Model Location
------------------------------------------------------------------------
4: /dev/kevm
28: /dev/disk/floppy0c 3.5in floppy fdi0-unit-0
31: /dev/disk/dsk0c DEC RZ26L (C) DEC bus-0-targ-0-lun-0
32: /dev/disk/dsk1c DEC RZ26 (C) DEC bus-0-targ-1-lun-0
33: /dev/disk/dsk2c COMPAQ HB00931B93 bus-0-targ-3-lun-0
34: /dev/disk/cdrom0c DEC RRD45 (C) DEC bus-0-targ-4-lun-0
35: /dev/disk/dsk3c COMPAQ HB00931B93 bus-0-targ-5-lun-0
37: DEC TLZ06 (C)DEC bus-0-targ-6-lun-0
Locate and write down the following data:
The device name for the target disk.
This is important if you installed a new replacement disk.
The device
name is the entry under the
Device Name
column, such as
/dev/disk/dsk2.
Ignore the partition suffix (c).
If there is no entry for a newly installed target disk you cannot proceed. You must shut down the system, verify the disk's physical installation, and restart the recovery procedure.
The device name for the backup device.
The device name appears in the
Device Name
column.
For example, if you are restoring the root disk from the default tape device
TLZ06
the
Device Name
column must contain a device
special file name such as
/dev/ntape/tape0.
There might be no device special file name in the
Device Name
column for the backup device, as shown in the preceding example.
In this case, go to
4
to install the tape device.
Install the backup device by using the following command:
# dn_setup -install_tape
To
verify the installation and determine the device name (such as
tape0_d0), repeat the
hwmgr
command in step
3.
[Configuration Dependent] If the original file system format is unknown, you can now ascertain it and verify that you have a readable backup tape as follows:
Load the backup (dump) media into the device.
Invoke the interactive mode of the
restore
command, specifying the backup device name that you determined in step
4.
For example:
# restore -i -f /dev/ntape/tape0_d0
If the backup is good, a prompt for interactive restoration
is displayed.
Enter the
what
command to display the header
and record the information.
Create and apply a disk label by using the following information:
The partition plan that you created during recovery planning, including any swap space requirements. (See Section 9.7.5.2.)
The new root device name determined in step 3.
Specify the
a
partition and label the drive as a
bootable device.
For example:
# disklabel -wr /dev/disk/dsk2a
Create your UFS target file systems as follows:
You must create file systems on the new root drive for each file system
that you need to restore.
For example, to create the new root and
/usr
file systems on partitions
a
and
g, use commands similar to the following:
# newfs /dev/disk/dsk2a # newfs /dev/disk/dsk2g
Mount the
replacement disk on the temporary mount point
/mnt
according
to the type of file system.
For example:
# mount /dev/disk/dsk2a /mnt
Use the
vrestore
or
restore
command to restore files.
For example:
# cd /mnt # vrestore -x device
Shut down and halt the system by using the following command:
# shutdown -h now
Boot the system to single-user mode, specifying the restored root drive as the boot device. For example:
>>> boot dka2 -flags s
If you are using an alternate drive, or if you installed a new
drive, you might need to translate the system device name to the appropriate
boot device name.
In step
3, you used the
hwmgr
command to determine the device database entry for the new
device.
For example:
33: /dev/disk/dsk2c COMPAQ HB00931B93 bus-0-targ-3-lun-0
Use the following command to display the devices:
>>> show device
Map
the value of the
b/t/l
(in this case 0.3.0) to
the alternate or new device and identify its boot device name, such as
dka300.
If the boot is successful, run the following script to update the device database:
# /sbin/mountroot
While the
dsfmgr
command
attempts to update the device database, some error or warning messages might
be displayed.
You can ignore the messages.
[Configuration Dependent] If you installed a new drive for root, or you specified an alternate device, you need to rename devices. Using the device name information that you determined in step 3, rename the devices as follows:
If you remove the old root disk and replace
it with a new device, use the
dsfmgr
command with the
-m
option to move the device names.
For example:
# dsfmgr -m dsk20 0
If the old root disk is still connected to the system, use
the
dsfmgr
command with the
-e
option to
exchange the device names.
For example:
# dsfmgr -e dsk20 0
Using the interactive mode of the
vrestore
command, load the backup media into the restore device and restore the device
directories.
This step ensures that all appropriate devices, including any
custom device drivers are recreated:
Delete the existing directories as follows:
# rm -rf /cluster/members/member0/dev*
If you used the
dump
command, restore the directories as follows:
# mkdir -p /usr/tmp /usr/var/tmp # restore -i -f /dev/ntape/tape0_d0 restore > add /cluster/members/member0/dev restore > add /cluster/members/member0/devices restore > extract
Use the
dsfmgr
command to verify the device
databases and device special file names as follows:
# dsfmgr -v
This is the end of the procedure.
Assuming that you have restored
all the required file systems, including
/usr
and (if necessary)
/var, you can now shut down the system, redefine the boot device,
and reboot the system to multiuser mode as follows:
# shutdown -h now >>> set bootdef_dev dka300 >>> boot
You can verify success by checking the boot process for any
error messages relating to devices.
If you determine that the procedure is
not successful, your only option is to reinstall the operating system from
the distribution media and recreate your customized environment.
9.7.5.4 Using Alternative root Disk Duplication Methods
Other methods of restoring or duplicating the root disk depend on whether or not you configured your system in anticipation of such problems. The following methods are available:
Methods of duplicating the root disk, such as mirroring, which are available when LSM and AdvFS are in use. Refer to the AdvFS Administration and Logical Storage Manager guides for more information.
Refer to the Logical Storage Manager guide for information on setting up a recoverable root volume.
Bootable tape provides a method of restoring customized systems that is faster than a conventional backup tape. This feature is available only on certain configurations. Refer to Chapter 9 for more information.
Backing up your customized system files can assist recovery
if you have to reinstall the operating system.
If you used the update installation
feature (updateinstall) for the previous installation,
you can create an archive of customized system files.
Refer to the
Installation Guide
for more information.
9.7.6 Restoring the /usr and /var File System
You might need to restore
the root file system as described in
Section 9.7.5
before
you can restore the
/usr
file system.
If the
/var
directory is on a file system other than
/usr,
repeat the steps in this section for restoring the
/var
file system.
The procedure in this section requires that you have access to the most
recent dump files of your
/usr
file systems.
The following
steps show how you restore from a level 0 dump of files, by using the text-based
(or character cell) interface to perform the task:
Label the disk if required by using the
disklabel
command as follows:
# disklabel -rw /dev/disk/dsk0
Note
The options used with the
disklabelcommand in this procedure will write a default disk partition table to the disk. If the disk you are restoring has a customized partition table, invoke the editing option of thedisklabelcommand or restore the partition table from your protofile label. See Chapter 6 anddisklabel(8) for more information.
Create a new file system by using the
newfs
command.
For example:
# newfs /dev/rdisk/dsk1c
If necessary, create mount points by using the
mkdir
command.
For example:
# mkdir /usr
Mount the file system by using the mount command, For example, to mount the file system created in the previous step, enter:
# mount /dev/disk/dsk1c /usr
Restore the file system:
If you are restoring dump files from a local file system,
change to the restore directory, insert the medium containing the dump file,
and enter the
restore
command.
For a tape, you might enter
the following commands:
# cd /mnt # restore -Yrf /dev/tape/tape0_d0
If you are restoring dump files from a remote system, change
to the restore directory and use the
rsh
command.
You might
need to specify the following command options:
remote_hostname - The host name of the remote system that contains the dump file
dumpfile - The full pathname of the dump file on the remote system.
blocksize - The block size is required to read data from a tape.
Read the dump file using the same block size that you specified when you wrote the tape. The default dump record size is 10 KB.
For example, to restore a dump file on a TLZ06 from the remote system
remotesystem
that you wrote by using the default block size, enter
the following:
# cd /mnt # rsh remotesystem "dd if=/dev/tape/tape0_d0 bs=10k" \ | restore -Yrf -
9.8 Using the Command-Line Utilities, tar, pax, and cpio
The
tar,
pax, and
cpio
command-line utilities provide a method of quickly creating an
archive from the command line or for writing scripts to back up files.
The
disadvantage is that you might have to type long command strings, and backing
up or restoring large volumes of files and directories is not easy when using
these interfaces.
You might use these utilities make a small archive of files
for distribution to other users, such as a program, its sources, and associated
documentation.
The following examples demonstrate how you can create or restore typical
archive files by using the command-line utilities.
Using tar to Create an Archive
The
tar
command saves and restores multiple files
on a single device such as a disk or tape.
To use the
tar
to create an archive on tape drive
/dev/tape/tape12_d0, enter a command such as the following
# tar cvfb /dev/tape12 -e ./.netscape -C /usr/glenn
The resulting archive contains all files and directories
in the
/usr/glenn
directory, except for file
./.netscape.
See
tar(1)
for more information.
Using pax to Create an Archive
The
pax
command extracts, writes, and lists members
of archive files.
It also copies files and directory hierarchies.
To use the
pax command to create a archive
of the current directory to device
/dev/tape/tape0_d0,
enter:
# pax -w -f /dev/tape/tape0
The following command reads the archive
a.pax,
extracting all files rooted in the
/usr
directory, relative
to the current directory:
# pax -r -s ',^//*usr//*,,' -f a.pax
See
pax(1)
for more information.
Using cpio to Create an Archive
The
cpio
command copies files between archive storage
and the file system.
It is used to save and restore data from traditional
format
cpio
archives.
To use the
cpio
command to create an archive to
tape device
/dev/tape/tape12_d0, enter:
# cpio -ov < file-list -O/dev/tape12_d0
See
cpio(1)
for more information.
9.9 Using dxarchiver
The Archiver,
dxarchiver, is a graphical user interface
for the command-line utilities described in
Section 9.8.
Use
this interface to:
Copy and store multiple files to a single, named archive file or output device such as a tape or floppy disk
Uncompress incoming archive files and compress newly created files
Retrieve stored files from an archive file or device such as a tape or floppy disk
Because the
dxarchiver
GUI is a CDE application,
you can drag and drop files and directories (folders) to assemble an archive
set, without having to type long commands.
It is assumed that you gathered the information Section 9.5.3, and you have loaded or unloaded a tape or other media into the target device as described in the owner's manual. To create an archive, proceed as follows:
Invoke the
/usr/bin/X11/dxarchiver
GUI
from a terminal command line, or open the CDE Application Group: System_Admin.
Then open System Admin Subgroup: Daily Admin and click on the Archiver icon.
Select the Archive Type:
tar,
cpio, or
pax.
Not all command-line options might
be available from the graphical interface.
Select any Archive Options.
You can only append to an existing
archive, and you cannot further compress an existing archive that was compressed
on creation.
Specify either an absolute or a relative pathname as the method
of storing the directories.
(An absolute pathname is the full path, beginning
at the root directory such as
/usr/users.
A relative pathname
begins at the current directory, for example dot (.) or
users/chan.)
During future recovery of these files, you can write them to a temporary location only if you specified a relative path during the original archiving process. Otherwise, files are restored to their original locations. (Potentially overwriting the existing version of the file unless you rename it.)
Specify the source, the files, and directories to archive. You can type pathnames or you can open a File Manager view and drag files and directories (CDE folders) to the Source Container box within the Archiver window. If you type pathnames, use the OK button to add them to the container.
After all required files are specified, choose the Archive... option and the Archiver: Archive window is displayed.
Enter a destination path, such as:
/dev/tape/tape0_d0
for the default tape
device.
/usr/backup/myback_991803
for a disk archive.
You do not need to enter a file name extension; the Archiver adds an identifier
such as
.Z.
After you choose OK, the destination is displayed under the Destination Container box.
Press Create Archive. A window titled Archiver working is displayed, flashing a green button to indicate that the archive is being written. The files being archived are displayed in the Destination Container.
After the archive is complete, you can optionally print a copy of the files list to keep as a record with the tape.
Choose Cancel to return to the Archiver main window. You can optionally enter the name of the archive file and use the Show Contents... option to verify that the archive was written correctly. The tape or archive file is read and the contents displayed in the Show Contents Window.
To extract an archive, you must specify a destination on a target device such as a disk. If you are not recovering a damaged file system on a complete disk partition, you might consider using a temporary location rather than overwriting existing directories. You can then restore individual files and directories as needed. You can also restore selected files from the archive as follows:
Invoke the
/usr/bin/X11/dxarchiver
GUI
from a terminal command line, or open the CDE Application Group: System_Admin.
Then open System Admin Subgroup: Daily Admin and click on the Archiver icon.
Choose Show Contents... to select individual files and directories. The tape or archive file is read and the contents displayed in the Archiver Show Contents window. Select individual files or directories as follows:
In the Archiver Show Contents window, click on a file or directory to highlight it.
Move to another file or directory, hold down the Ctrl key and click to select it.
After you select all the files that you want, choose OK in the Archiver Show Contents window. The files are displayed in the Source Container box in the Archiver main window. You can use the Edit menu to make additional changes to selections. For example, highlight an entry in the source container and choose Edit: Clear Selected Source to delete it.
Choose the Extract... option to display the Archiver Extract window.
Enter a destination directory.
This directory can be the same
as the archive, assuming that files can be overwritten.
Alternatively, give
the path to a temporary location.
This path must be to an existing directory,
or you must open a terminal and create it with the
mkdir
command.
Alternatively, create a folder by using the New Folder option in
CDE File Manager.
The destination is displayed under the Destination Container
box.
Choose Extract Contents to begin the extraction. A window titled Archiver Working is displayed, flashing a green button to indicate that the archive is being extracted. The files being recovered are displayed in the Destination Container.
After the archive is complete, you can optionally print a copy of the files list, to keep as a record.
Choose Cancel to return to the Archiver main window. Before exiting, use the File Manager or a terminal window to ensure that the files were recovered as expected and that the file contents are not corrupted. This step is strongly recommended before you proceed to remove any archives from tape or other media.
You can now remove the tape or other media as described in the
owner's manual for the device, and store the media in a safe location (or
in accordance with your site backup policy and procedures).
9.10 Creating a Standalone System Kernel on Tape
You can create a bootable standalone system (SAS) kernel on tape. The SAS kernel has a built-in memory file system (mfs), which contains the minimum commands, files, and directories needed to restore the system image. This is referred to as the miniroot file system. You can also add required file systems to the tape for data or programs that you might need on the recovered system.
To create the SAS kernel, you must use the SysMan Menu Create a
Bootable Tape option or the
btcreate
command-line utility.
After you create the kernel, you can restore the customized image by using
the
btextract
utility.
You can invoke only a single instance
of the
btcreate
utility.
The
/usr/run/bttape.pid
locking file prevents multiple instances of the utility.
The following sections provide an overview of the
bttape
interfaces, SysMan Menu task, and the
btcreate
and
btextract
command-line utilities.
9.10.1 Tape Device Requirements
If you use QIC tape drives to create bootable tapes, you must use only high-density tapes of 320 or more megabytes. The QIC-24, QIC-120, and QIC-150 format tapes of fixed-512 blocks do not work. Tapes with a variable block size, such as the QIC-320 and QIC-525, work with bootable tape. Using an improperly configured QIC tape drive to create a bootable tape results in an I/O error, a write error, or permission denied error. Therefore, you must take one of the following actions:
Configure the drive at installation time
Rebuild the kernel if the drive was attached to the system after the installation
A QIC tape created with the
btcreate
command
might fail with the following error message when booted:
failed to send Read to mka... Be sure that the tape is write protected before booting.
If you are creating a bootable tape with a file system that extends
to multiple tapes, the
/sbin/dump
command displays a message
indicating that you must change the tape.
If you do not change the tape promptly,
warning messages repeat periodically until you change the tape.
The behavior of the open call to a tape device has changed. You can no longer use write mode to open a write-protected tape. An attempt to open the tape fails, returning the following message:
EACCES (permission denied)
If an application is written so that it attempts to open the tape
device with
O_RDWR
permission when the intention is only
to read the tape, the open attempt fails.
Change any applications or scripts
to open the device with
O_RDONLY
permission.
Use the following
command to obtain the previous behavior of the open call for applications
that cannot be changed:
# sysconfig -r cam_tape open_behaviour=0
9.10.2 Using the btcreate Utility
To build a bootable SAS kernel on UFS or AdvFS file systems only, you
must use the
btcreate
utility.
The following sections
provide an overview of the information you must have to create the SAS kernel
on tape.
The
btcreate
command provides both a noninteractive
and interactive user interface.
Both require that you have superuser (root)
privileges.
9.10.2.1 Gathering Information
To prepare to use the
btcreate
command, you must
have the following information:
Name of the kernel configuration file in the
/usr/sys/conf
directory.
The default is the same as the system (HOST) name in
capital letters.
Name of the disk partition (for example,
dsk2e)
where the miniroot file system is to reside.
Minimum size needed on the disk
is 38000 blocks (512 bytes per block).
This disk partition must not be in
a mounted state when
btcreate
is executed.
Name of the tape device, for example
/dev/tape/tape0_d0, where the SAS kernel and file systems are to reside.
Device name, mount point, and type of each file system (UFS or AdvFS) that you want to back up to the tape device. The following examples show valid UFS and AdvFS entries:
UFS:
/dev/dsk1a / ufs /dev/dsk1g /usr ufs /dev/vol/rootdg/rootvol /ufs
AdvFS:
root_domain#root / advfs usr_domain#usr /usr advfs
Note
Do not select swap partitions for file system backups.
For UFS file systems that are stored on LSM volumes, the
vdump
and
vrestore
utilities are used during
bootable tape creation.
An addlist_file file, which lists the files or directories you want to include on the miniroot file system.
An fslist_file file, which specifies the file systems to back up.
A
/usr/lib/sabt/sbin/custom_install.sh
script, if you want to customize the restored system image.
Write the script
using the Bourne shell language (sh1) because it is the
only shell provided on the miniroot file system.
The
btcreate
command copies the
custom_install.sh
file onto tape and
places it in the
sbin
directory on the miniroot file system.
The
btextract
command invokes the
custom_install.sh
script before exiting.
The following additional features might be useful in planning your bootable tape layout:
Use the
-d
option to specify the location
where
btcreate
creates its temporary files.
If you do
not specify a location, 156000 blocks (512 bytes per block) of disk space
in the
/usr
file system is required.
You can label disks using your own custom
disklabel
script, which must meet the following requirements:
It must be located in the/usr/lib/sabt/etc
directory.
It must be named
custom_disklabel_file.
If a custom
disklabel
script is not present, the
btextract
command labels the disks in the usual manner.
See
disklabel(8)
for more information.
9.10.2.2 Creating the SAS Kernel
To create the SAS kernel, the
btcreate
command copies
the
/usr/sys/conf/YOUR_SYSTEM_NAME
configuration file to
/usr/sys/conf/YOUR_SYSTEM_NAME.BOOTABLE
and modifies it as follows:
config vmunix root on md pseudo-device memd 38000
These modifications configure a memory file system of 38000 blocks. The memory file system and the disk partition where the miniroot file system reside are equivalent in size.
After modifying the configuration file, the
btcreate
command executes the
doconfig
command and moves the bootable
kernel to the
/usr/sys/bin
directory.
For information on
the command syntax and options, see
btcreate(8).
9.10.3 Using the btextract Utility
The
btextract
command is a shell script that restores
file systems from tapes that contain a SAS kernel created by using the
btcreate
utility.
You can perform a default restoration or an
advanced restoration of the system.
During a default restoration, you can choose to duplicate the customized
system on more than one system of the same hardware platform type.
You cannot
specify which disk partitions to use for the restore operation.
Instead, the
btextract
command restores file systems using the disk partition
information gathered during the
btcreate
session and all
existing information is overwritten.
If you are performing an advanced restoration,
you can optionally specify which disk partition to use.
However you can duplicate
the customized system only on a system of the same hardware platform type.
To use the
btextract
command, place the system in
a halt state, initialize the system, then boot from the tape as follows:
>>> init >>> show dev >>> boot -fl "nc" MKA500
In this example, the
show
dev
command provides the device name under BOOTDEV and
MKA500
is the BOOTDEV.
After the initial boot completes, the shell invokes the
btextract
utility.
If you created a
/usr/lib/sabt/sbin/custom_install.sh
script during the
btcreate
session, the
btextract
command invokes the
custom_install.sh
script before exiting.
You can also create a
custom_prerestore
answer file to automate the recovery procedure.
See
btcreate(8)
for more
information.
After the
btextract
command completes its task, you
must shut down and reboot the system from the restored disk as follows:
# shutdown -h now >>> boot DKA100
In this example,
DKA100
is the BOOTDEV.
For more information and examples, see
btextract(8).
9.10.4 Using the SysMan Menu boot_tape Option
The following steps describe the basic process for creating a bootable tape. It assumes that you have gathered the necessary device data as described in Section 9.10.2.1, and the tape device is ready to save.
Invoke the Create a Bootable Tape task from the SysMan Menu, or enter the following command at the prompt:
# sysman boot_tape
A window titled Bootable Tape Creation on hostname is displayed. Complete the fields or choose options as follows:
In the Kernel Name field, the default kernel name for the host is displayed. This is usually the same as the local host name. However, you can enter any name for the saved kernel.
The Miniroot File System field provides the following options:
The option to create the miniroot as a memory file system (mfs) or a Disk Partition. Click on the button for the option of your choice.
The option to specify a disk partition name such as
dsk0b
with the Specify Disk Partition/mfs...
button.
(This button
displays a dialog box in which you enter the disk partition name.)
The Tape Device field contains the name of the default tape
device, usually
tape0_d1.
This is the name of the device
on which the SAS kernel is saved, but you can specify any other supported
device.
The Customizing the Miniroot File System field displays the
default file location for the
addlist
file.
This is a data
file that contains a list of additional files that you want to include, such
as commands or utilities.
You cannot exceed 360 KB of data in the mfs.
This
list is stored in the
/usr/lib/sabt/etc/addlist
file by
default but you can choose to create your own location.
To create a new append file, or modify an existing append file:
Click on the Create/Modify Miniroot Append File button to display the Create/Modify window.
Click on the Add button to display the Add/Modify window.
Specify the location of the file that you want on the local host.
For example,
to add the
kill
command, enter
/sbin/kill.
Then specify the location on the miniroot file system where the file is located,
such as
/sbin.
Click on the OK button to return to the
Create/Modify window.
The Contents of file: box contains a list of the files to be appended. Click on the OK button to return to the main window for Bootable Tape Creation.
The Selecting File Systems option enables you to back up file
systems, such as
/usr
or an AdvFS domain such as
root_domain#root.
The list of files to be backed up is stored in
/usr/lib/sabt/etc/fslist, but you can specify any name that you
want.
Add file systems as follows:
Click on the Create/Modify File Systems Backup File... button to display the Create/Modify window.
Click on the Add button to display the Add/Modify window.
Specify the disk partition mounted on the local host, such as
/dev/disk/dsk0g, then specify the mount point, such as
/usr.
Click on the OK button to return to the Create/Modify window.
The file systems to be backed up are listed in the Contents of file: box. Click on the OK button to return to the main window for Bootable Tape Creation.
After completing the required fields, you are ready to create the tape. In the main window for Bootable Tape Creation, click on the OK button to proceed. A message window is displayed to indicate that the task has started. The creation of the tape can take twenty or more minutes, depending on the speed of the devices used.
If the task cannot be completed, a message is displayed informing you
that the error log is located in
/var/adm/btcreate.log.
After the tape is successfully written, a message is displayed
confirming the success and the location of the log file,
/var/adm/btcreate.log.
Print
btextract(8)
and store it with the tape for future reference.
Use the instructions in
Section 9.10.3
and
btextract(8)
to restore the bootable SAS kernel.
Consider running a test
recovery to ensure that any future recovery works as intended.