backup2l.8

.TH "BACKUP2L" "8" "August 17, 2010" "backup2l v1.6"

.SH NAME
backup2l \-  low\-maintenance backup/restore tool

.SH SYNOPSIS
.B backup2l
[
.BI "\-c " conffile
] [
.BI "\-t " "backup\-ID"
]
.I command

.SH DESCRIPTION
\fBbackup2l\fP is a tool for generating, maintaining and restoring backups on a mountable file system
(e. g. hard disk). The main design goals are low maintenance effort, efficiency, transparency and robustness.
It features differential backups at multiple hierarchical levels and provides rollback functionality.

All control files are stored together with the archives on the backup device, and their contents are
mostly self\-explaining. Hence, in the case of an emergency, a user does not only have to rely on the
restore functionality of \fBbackup2l\fP, but can \- if necessary \- browse the files and extract
archives manually.

An open driver architecture allows to use virtually any archiving program as a backend.
Built\-in drivers support \fB.tar.gz\fP, \fB.tar.bz2\fP, or \fB.afioz\fP files. Further drivers can be
added by the user. When restoring data, an appropriate driver is selected automatically for each
archive depending on the suffix of the archive file.

The method of hierarchical differential backups is a generalization to the concept of the "daily",
"weekly" and "monthly" backups.
Each backup has a level and a serial number. Maximum\-level backups are comparable with
daily differential backups, level\-0 backups are full backups. For example, let \fBMAX_LEVEL\fP be 4
and \fBMAX_PER_LEVEL\fP be 5. After 5 level\-4 backups (e. g. after 5 days), a new level\-3 backup is made.
After 5 level\-3 backups (and 5*5 at level\-4), a new level\-2 backup is made, and so on.
Each differential backup contains the changes towards the previous backup of the same or a lower level.

This scheme allows to efficiently generate small incremental backups at short intervals.
At the same time, the total number of archives that have to be stored (or processed in the case of a restore)
only increases logarithmically with the number of backups since the last full backup. Time\-consuming
full backups are only sparsely needed.
In the example above, a new full backup is only necessary after 780 (=5^4+5^3+5*5+5) days, while only
at most 20 (=4*5) archives have to be processed.

For \fBbackup2l\fP, each backup archive is identified by its \fBbackup ID (BID)\fP. The number of digits
determines the level. Level\-0 (full) backups have a 1\-digit BID, level\-n backups have a BID of n+1 digits.
The last digit is a serial number, the prefix identifies the lower\-level backups on which a given backup
is based on. For example, the archive 235 contains the differences towards archive 234, and to restore
the file system state of the time it was generated, the full backup 2, the level\-1 backups 21, 22, 23
and the level\-2 backups 231, ..., 235 have to be processed.
All serial numbers are between 1 and 9, a zero in the BID indicates that no archive of the respective level
is contained in the chain. For example, the level\-3 backup 1201 is immediately based on the level\-1
backup 12.

For deciding whether a file is new or modified, \fBbackup2l\fP looks at its name, modification time, size,
ownership and permissions. Unlike other backup tools, the i\-node is not considered in order to avoid problems with
non\-Unix file systems like FAT32.

.SH OPTIONS
.TP
.BI "\-c, \-\-conf " conffile
This argument specifies the configuration file (default: \fBetc/backup2l.conf\fP).
.TP
.BI "\-t, \-\-time " "BID"
If present, this option selects a certain backup for the \fB\-\-locate\fP and \fB\-\-restore\fP commands.
E. g., the latter will restore files and directories exactly as they were on the system at the time when the
specified backup was made. If not present, the latest available backup is selected.
.SH COMMANDS
.TP
.BI "\-h, \-\-help"
Display the usage info.
.TP
.BI "\-e, \-\-estimate " "\fR[\fP level \fR]\fP"
Prints the number of files, estimated amount of data and other information on the backup that would be
generated next. No backup archives are actually created
or removed. If specified, the parameter \fBlevel\fP overrides the \fBMAX_LEVEL\fP setting.
.TP
.BI "\-b, \-\-backup " "\fR[\fP level \fR]\fP"
Creates a new backup and removes old archives based on the given configuration file.
If specified, the parameter \fBlevel\fP overrides the \fBMAX_LEVEL\fP setting. This is useful e. g. shortly
before or after major changes are performed with the file system. In this case, a lower level should be
specified in order to avoid that a large number of files are backed up multiple times again.
.TP
.BI "\-s, \-\-get\-summary"
Shows a table describing each backup (date, size, files, ...) and the file system usage of the backup device.
.TP
.BI "\-a, \-\-get\-available " "\fR[\fP pattern list \fR]\fP"
Shows all files removed and added for all backups. A '+' in the output indicates that the file is new and
thus contained in the archive file. A '\-' indicates that the file has been removed (or replaced).
If one or several patterns are supplied, \fBgrep(1)\fP is used to filter
the list. All status messages go to \fBstderr\fP, so that the generated file list can easily be redirected.

\fBNote:\fP The search pattern is not just applied to the file names, but to the whole entry in the \fB.list.gz\fP
file. This allows you to not only search for file names but also for other attributes like
ownership, modification time etc. . In order to apply a search pattern to file names only, precede it by
"<space>/.*".
.TP
.BI "\-l, \-\-locate " "\fR[\fP pattern list \fR]\fP"
Shows most recent backup location for active files. If one or several patterns are supplied, \fBgrep(1)\fP
is used to filter the list in the same way as for \fB\-\-get\-available\fP (see above) . All status messages
go to \fBstderr\fP, so that the generated file list can easily be redirected.

Active files are files that have been on the system at the time of the selected backup, which is either
the latest backup or the one specified by \fB\-\-time\fP (see above). Files that were removed at that time
but are still stored in some later archive will not be shown.

Altogether, this command tells you, which files have to be extracted from which archive in order to
restore the state of the system at the time of the selected backup.
.TP
.BI "\-r, \-\-restore " "\fR[\fP pattern list \fR]\fP"
Performs the same steps like \fB\-\-locate\fP and then restores the respective files. All files are restored
relative to the current directory. They can be restored to their original location by cd'ing into / before,
but this is not recommended.
.TP
.BI "\-p, \-\-purge " "BID list"
Removes the specified backup archive(s) and all depending backups.
.TP
.BI "\-m, \-\-make\-check " "\fR[\fP BID list \fR]\fP"
Creates (a) check file(s) for the specified archive(s) using \fBmd5sum(1)\fP.
If no BID is specified, check files are created wherever missing.
.TP
.BI "\-v, \-\-verify " "\fR[\fP BID list \fR]\fP"
Verifies the specified backup archive(s). If no BID is specified, all existing archives are checked.
If a check file exists, this allows a comprehensive test including e. g. media failures. If the check file
is missing, only the existence of all files and the immediate base archive are verified.
.TP
.BI "\-x, \-\-extract " "capacity max\-free BID\-list"
Split and collect files to be stored on removable media (e. g. CDs). \fBcapacity\fP is the medium
capacity in MB. \fBmax\-free\fP is the maximum amount of empty space on each medium (except for the last
one, of course). \fBBID\-list\fP specifies the archives and may contain wildcards, e. g.: 1 '2*'.
The operation generates enumbered subdirectories representing the media contents.
Some more files are generated that may be useful, e. g. to print labels.

While guaranteeing a minimum waste of \fBmax\-free\fP MB per medium, the collection procedure preserves
the ordering of files and keeps all control files of an archive always together on the same medium.
Large archive files are split into multiple files with serial numbers appended to their names.

The operation is interactive. Just run it and look what it is about to do.
If that is not what you want, you can stop it.

.SH CONFIGURATION
In the configuration file (\fB/etc/backup2l.conf\fP by default), the following variables have to be set,
following the \fBbash(1)\fP syntax:
.TP
.BI "FOR_VERSION=" "version"
Defines the \fBbackup2l\fP version for which the configuration file is written. This way,
future versions can automatically print a warning if the syntax has changed.
.TP
.BI "SRCLIST=(" " source list " ")"
This is a blank\-separated list of all top\-level directories to make backups of.
Directory names with spaces have to be quoted, e. g.: SRCLIST=("/my dir" /another/dir).
The last elements of the list may be options for \fBfind(1)\fP, for example \fB\-xdev\fP in
order to skip subdirs on other file systems like /dev or /proc.
.TP
.BI "SKIPCOND=(" " find condition " ")"
Files for which this condition is 'true' are not considered for backup.
See \fBfind(1)\fP for information on how to formulate possible conditions.
Special characters ("(", ")", "!", ...) must be quoted by a leading backslash
("\\(", "\\)", " \\!", ...). An empty condition (i. e. if you do not want
any files to be skipped) must be specified as "( \-false )".
.TP
.BI "[ BACKUP_DEV=""" "mount_point" """ ]"
If defined, \fBbackup2l\fP mounts the backup device before any operation.
Afterwards, it is unmounted unless it was already mounted before.
.TP
.BI "BACKUP_DIR=""" "backup dir" """"
Destination directory for backup files. This must be different from \fBMOUNT_POINT\fP, i. e.
a subdirectory on the device.
.TP
.BI "VOLNAME=""" "volname" """"
This is a common prefix for all backup and control files. Multiple backup volumes are possible
if for each volume a separate configuration file is written.
.TP
.BI "MAX_LEVEL=" "max_level"
Maximum backup level. Possible values are 1..9.
.TP
.BI "MAX_PER_LEVEL=" "max_per_level"
Number of differential backups per level. Possible values are 1..9.
.TP
.BI "MAX_FULL=" "max_full"
Number of full backups kept. Possible values are 1..8.
.TP
.BI "GENERATIONS=" "generations"
Number of backup generations to keep for each non\-zero level. Old backups are automatically removed
as long as at least \fBGENERATIONS * MAX_PER_LEVEL\fP backups for the respective level remain.

For example, with \fBMAX_LEVEL=3, MAX_PER_LEVEL=5, GENERATIONS=2\fP it is always possible to access
the last 10 level\-3 (e. g. daily) backups, the last 10 level\-2 backups (e. g. 5, 10, 15, ..., 50 days old),
and so on.
.TP
.BI "PRE_BACKUP () { " "do something" " }"
This function is called before writing the backup. It can be used to dump some important system information,
e. g. the HD's partition table, to a file which is then backed up.
.TP
.BI "POST_BACKUP () { " "do something" " }"
This function is called after writing the backup. Together with \fBPRE_BACKUP\fP it can be used to stop and restart
e. g. database or mail services which may frequently alter some files that have to be backed up.
.TP
.BI "[ AUTORUN=1 ]"
If set to 1, \fBbackup2l\fP performs the \fB\-\-backup\fP operation when invoked without arguments.
Otherwise, the usage information is shown.
.TP
.BI "[ SIZE_UNITS= " "B | K | M | G" " ]"
Sets the units for archive sizes in summary listings to bytes, KB, MB, or GB.
If unset, a user\-readable format is chosen automatically. If set, the units are the same for the whole table,
which may be even more user\-friendly.
.TP
.BI "[ CREATE_DRIVER=""" "archive driver" """ ]"
Selects an archive driver for creating backups. An archive driver is responsible for managing backup
files. If unset, the default driver "DRIVER_TAR_GZ" is used. The \fB\-\-help\fP operation lists all
available drivers. More drivers can be defined in the configuration file (see below).
.TP
.BI "[ USER_DRIVER_LIST=""" "user\-defined drivers" """ ]"
Declares additional, user\-defined archive drivers which are implemented in the configuration file.
The sample configuration file contains a commented example. Read it in order to learn how to
implement your own driver.

.SH FILES
.TP
.B /etc/backup2l.conf
Configuration file.
.TP
.IB "VOLNAME" "." "BID" ".tar.gz, " "VOLNAME" "." "BID" ".afioz, ..."
Archive files.
.TP
.IB "VOLNAME" "." "BID" ".list.gz"
List of all active files when the backup was made. Each file is preceded with its size,
modification time, and other information.
.TP
.IB "VOLNAME" "." "BID" ".new.gz"
List of all new ore modified files when the backup was made (pathnames only). Unless an error
occured, this list reflects the contents of the archive.
.TP
.IB "VOLNAME" "." "BID" ".obsolete.gz"
List of all obsolete files when the backup was made (pathnames only).
.TP
.IB "VOLNAME" "." "BID" ".skipped.gz"
Complete list of all files that were skipped according to \fBSKIPMASK\fP.
.TP
.IB "VOLNAME" "." "BID" ".error.gz"
This file is generated by comparing the \fB.new.gz\fP file with the actual archive
contents using \fBdiff(1)\fP. If the error file is non\-empty, something may have gone wrong.
.TP
.IB "VOLNAME" "." "BID" ".check"
MD5 check sums of all files of the present archive and the \fB.list.gz\fP file of the base archive.
This file is optional and may be used by the \fB\-\-verify\fP operation.
.SH INVOCATION BY CRON
\fBbackup2l\fP is designed to be run autonomously as a cron job.
If the variable \fBAUTORUN\fP is set, it generates a backup if invoked without any parameters, and
you can simply create a symlink, e. g. by:

.ce
ln \-s `which backup2l` /etc/cron.daily/zz\-backup2l

The "zz\-" prefix causes the backup job to be the last one executed, so that other jobs are not delayed
if the backup takes somewhat longer. The status output is e\-mailed to root by the cron daemon.

.SH MANIPULATING FILES AND CONFIGURATIONS
\fBbackup2l\fP has been designed to be robust with respect to errors and configuration changes.
If the backup process is interrupted, e. g. because of a shutdown while it is running,
no serious data corruption can occur. Some temporary files may remain which are cleaned up during the
next run. If file is changed during the backup generation, it may not be contained in the current
backup. However, it is guaranteed that it is considered modified during the next backup.

In order to save disk space, e. g. after some archives have been copied to external media,
archive files (.tar.gz or .afioz, for example) can safely be removed from the backup directory.
As long as all control files are kept, \fBbackup2l\fP retains full functionality as far as possible.
The \fB\-\-restore\fP command prompts for eventually missing archive files for the respective request
(and only those). The \fB\-\-extract\fP command completely ignores all backups with missing archive files.

The configuration, especially the settings for \fBMAX_LEVEL\fP,  \fBMAX_PER_LEVEL\fP,
\fBMAX_FULL\fP and the specification of source files,
can be arbitrarily changed without having to expect data corruption. \fBbackup2l\fP will
gracefully adapt the new settings during the next run.

.SH BUGS
After a restore operation, the modification time of directories is equal
to the restoration time while for files it is equal to the original modification
time.

.SH AUTHOR
\fBbackup2l\fP was written by Gundolf Kiefer <gundolfk at web.de>.

This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published
by the Free Software Foundation, either version 2 of the License,
or (at your option) any later version.

This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.

.SH SEE ALSO
.BR "tar(1), afio(1), find(1), grep(1), md5sum(1)"