5 \h'-\w'\fB\\$1\ \fP'u'\fB\\$1\ \fP\c
7 .TH rsync-backup 8 "7 October 2012" rsync-backup
9 rsync-backup \- back up files using rsync
18 script is a backup program of the currently popular
23 ability to create hardlinks from (apparently) similar existing local
24 trees to make incremental dumps efficient, even from remote sources.
25 Restoring files is easy because the backups created are just directories
26 full of files, exactly as they were on the source \(en and this is
31 The script does more than just running
33 It is also responsible for creating and removing snapshots of volumes to
34 be backed up, and expiring old dumps according to a user-specified
39 script should be installed and run on a central backup server with local
40 access to the backup volumes.
42 The script should be run with full (root) privileges, so that it can
43 correctly record file ownership information. The server should also be
46 to the client machines, and run processes there as root. (This is not a
47 security disaster. Remember that the backup server is, in the end,
48 responsible for the integrity of the backup data. A dishonest backup
49 server can easily compromise a client which is being restored from
51 .SS Command-line options
52 Most of the behaviour of
54 is controlled by a configuration file, described starting with the
56 .B Configuration commands
58 But a few features are controlled by command-line options.
61 Show a brief help message for the program, and exit successfully.
66 version number and some choice pieces of build-time configuration, and
72 instead of the default configuration file (shown as
79 Don't actually take a backup, or write proper logs: instead, write a
80 description of what would be done to standard error.
83 Produce verbose progress information on standard output while the backup
84 is running. This keeps one amused while running a backup
85 interactively. In any event,
87 will report failures to standard error, and otherwise run silently, so
88 it doesn't annoy unnecessarily if run by
91 Backing up a filesystem works as follows.
93 Make a snapshot of the filesystem on the client, and ensure that the
94 snapshot is mounted. There are some `trivial' snapshot types which use
95 the existing mounted filesystem, and either prevent processes writing to
96 it during the backup, or just hope for the best. Other snapshot types
97 require the snapshot to be mounted somewhere distinct from the main
98 filesystem, so that the latter can continue being used.
102 to copy the snapshot to the backup volume \(en specifically, to
103 .IB host / fs / new \fR.
104 If this directory already exists, then it's presumed to be debris from a
105 previous attempt to dump this filesystem:
107 will update it appropriately, by adding, deleting or modifying the
108 files. This means that retrying a failed dump \(en after fixing whatever
109 caused it to go wrong, obviously! \(en is usually fairly quick.
113 on the client to generate a `digest' describing the contents of the
114 filesystem, and send this to the server as
115 .IB host / fs / new .fshash \fR.
117 Release the snapshot: we don't need it any more.
121 over the new backup; specifically, to
122 .BI tmp/fshash. host . fs . date \fR.
123 This gives us a digest for what the backup volume actually stored.
127 digests. If they differ then dump the differences to the log file and
128 report a backup failure. (Backups aren't any good if they don't
129 actually back up the right thing. And you stand a better chance of
130 fixing them if you know that they're going wrong.)
132 Commit the backup, by renaming the dump directory to
137 .IB host / fs / date .fshash \fR.
139 The backup is now complete.
140 .SS Configuration commands
141 The configuration file is simply a Bash shell fragment: configuration
142 commands are shell functions.
144 .BI "addhook " hook " " command
145 Arrange that the named
153 .BI "backup " "fs\fR[:\fIfsarg\fR] ..."
154 Back up the named filesystems. The corresponding
156 may be required by the snapshot type.
159 Define a new hook named
165 for more information.
170 commands will back up filesystems on the named
172 To back up filesystems on the backup server itself, use its hostname:
174 will avoid inefficient and pointless messing about
177 This command clears the
181 name, and resets the retention policy to its default (i.e., the to
182 policy defined prior to the first
186 .BI "like " "host\fR ..."
187 Declare that subsequent filesystems are `similar' to like-named
188 filesystems on the named
192 should use those trees as potential sources of hardlinkable files. Be
193 careful when using this option without
196 option: an erroneous hardlink will cause the backup to fail. (The
197 backup won't be left silently incorrect.)
199 .BI "retain " frequency " " duration
200 Define part a backup retention policy: backup trees of the
202 should be kept for the
214 which means the same); the
222 Expiry considers each existing dump against the policy lines in order:
223 the last applicable line determines the dump's fate \(en so you should
224 probably write the lines in decreasing order of duration.
233 commands collectively define a retention policy. Once a policy is
236 operations use the policy. The first
242 command clears the policy and starts defining a new one. The policy
243 defined before the first
247 policy: at the start of each
249 stanza, the policy is reset to the default.
255 snapshot type (see below) doesn't prevent a filesystem from being
256 modified while it's being backed up. If this happens, the
258 pass will detect the difference and fail. If the filesystem in question
259 is relatively quiescent, then maybe retrying the backup will result in a
260 successful consistent copy. Following this command, a backup which
263 mismatch will be retried up to
265 times before being declared a failure.
267 .BI "runhook " hook " " args\fR...
270 The individual commands on the hook are run, in order, as
276 If any command fails (returns nonzero) then no other hooks are run and
278 fails with the same exit code.
281 .BI "snap " type " " \fR[\fIargs\fR...]
284 for subsequent backups. Some snapshot types require additional
285 arguments, which may be supplied here. This command clears the
290 Specify the user name on the remote host. Without this, calls to
294 won't specify any user name, so the default (probably from the
297 .SS Configuration variables
298 The following shell variables may be overridden by the configuration
302 The hash function to use for verifying archive integrity. This is
307 so it must name one of the hash functions supported by your Python's
314 The name of a SQLite database initialized by
315 .BR update-bkp-index (8)
316 in which an index is maintained of which dumps are on which backup
317 volumes. If the file doesn't exist, then no index is maintained. The
319 .IB localstatedir /lib/bkp/index.db
322 is the state directory configured at build time.
325 The number of log files to be kept for each filesystem. Old logfiles
326 are deleted to keep the total number below this bound. The default
330 The metadata directory for the currently mounted backup volume.
335 is the backup mount directory configured at build time.
338 Command-line options to pass to
340 in addition to the basic set:
347 .B \-\-one-file-system
349 .BR "\-\-filter=""dir-merge .rsync-backup""" .
356 snapshots are mounted on subdirectories below the
358 .IR "on backup clients" .
363 is the backup mount directory configured at build time.
366 The volume size option to pass to
368 when creating a snapshot. The default is
370 which seems to work fairly well.
373 Where the actual backup trees should be stored. See the section on
380 is the backup mount directory configured at build time.
383 The name of the current volume. If this is left unset, the volume name
384 is read from the file
386 once at the start of the backup run.
388 The configuration file can modify the behaviour of the backup in two
389 main ways: by adding commands to hooks (see the
391 command); and by redefining shell functions.
393 The following hooks are defined.
395 .BI "commit " host " " fs " " date
396 Called during the commit procedure. The backup tree and manifest have
397 been renamed into their proper places. Typically one would use this
398 hook to rename files created in a corresponding
403 The backup has completed;
405 will exit with status
408 .BI "precommit " host " " fs " " date
409 Called after a backup has been verified complete and about to be
410 committed. The backup tree is in
412 in the current directory, and the
416 A typical action would be to create a digital signature on the
419 .BI "setup " host " " fs " " date
420 Called when a backup of a particular filesystem is about to start. It
421 can return with code 99 to skip the backup.
424 Invoked before performing any actual dumps (the first time
428 The following shell functions can be redefined by users.
430 .BI "backup_commit_hook " host " " fs " " date
433 hook for compatibility.
435 .BI "backup_precommit_hook " host " " fs " " date
438 hook for compatibility.
440 .BR "whine " [ \-n ] " " \fItext\fR...
441 Called to report `interesting' events when the
443 option is in force. The default action is to echo the
445 to (what was initially) standard output, followed by a newline unless
449 The following snapshot types are available.
452 A trivial snapshot type: attempts to back up a live filesystem. How
453 well this works depends on how active the filesystem is. If files
454 change while the dump is in progress then the
456 verification will likely fail. Backups using this snapshot type must
457 specify the filesystem mount point as the
461 A slightly less trivial snapshot type: make the filesystem read-only
462 while the dump is in progress. Backups using this snapshot type must
463 specify the filesystem mount point as the
467 Create snapshots using LVM. The snapshot argument is interpreted as the
468 relevant volume group. The filesystem name is interpreted as the origin
469 volume name; the snapshot will be called
472 .IB SNAPDIR / fs \fR;
473 space will be allocated to it according to the
477 .BI "rfreezefs " client " " vg
478 This gets complicated. Suppose that a server has an LVM volume group,
479 and exports (somehow) a logical volume to a client. Examples are a host
480 providing a virtual disk to a guest, or a server providing
481 network-attached storage to a client. The server can create a snapshot
482 of the volume using LVM, but must synchronize with the client to ensure
483 that the filesystem image captured in the snapshot is clean. The
485 program should be installed on the client to perform this rather
486 delicate synchronization. Declare the server using the
488 command as usual; pass the client's name as the
491 server's volume group name as the
493 snapshot arguments. Finally, backups using this snapshot type must
494 specify the filesystem mount point (or, actually, any file in the
495 filesystem) on the client, as the
498 Additional snapshot types can be defined in the configuration file. A
499 snapshot type requires two shell functions.
501 .BI snap_ type " " snapargs " " fs " " fsarg
502 Create the snapshot, and write the mountpoint (on the client host) to
503 standard output, in a form suitable as an argument to
506 .BI unsnap_ type " " snapargs " " fs " " fsarg
509 There are a number of utility functions which can be used by snapshot
510 type handlers: please see the script for details. Please send the
511 author interesting snapshot handlers for inclusion in the main
513 .SS Archive structure
514 Backup trees are stored in a fairly straightforward directory tree.
516 At the top level is one directory for each client host. There are also
517 some special entries:
519 .B \&.rsync-backup-store
520 This file must be present in order to indicate that a backup volume is
521 present (and not just an empty mount point).
524 The cache database used for improving performance of local file
525 hashing. There may be other
527 files used by SQLite for its own purposes.
530 Part of the filesystem used on the backup volume. You don't want to
534 Used to store temporary files during the backup process. (Some of them
535 want to be on the same filesystem as the rest of the backup.) When
536 things go wrong, files are left behind in the hope that they might help
537 someone debug the mess. It's always safe to delete the files in here
538 when no backup is running.
540 So don't use those names for your hosts.
542 The next layer down contains a directory for each filesystem on the
545 The bottom layer contains a directory for each dump of that filesystem,
546 named with the date at which the dump was started (in ISO8601
547 .IB yyyy \(en mm \(en dd
548 format), together with associated files named
550 There is also a symbolic link
552 referring to the most recent backup of the filesystem.
554 .BR check-bkp-status (8),
560 .BR update-bkp-index (8).
562 Mark Wooding, <mdw@distorted.org.uk>