1 mailto(rsync-bugs@samba.org)
2 manpage(rsync)(1)(29 Jun 2008)()()
3 manpagename(rsync)(a fast, versatile, remote (and local) file-copying tool)
6 verb(Local: rsync [OPTION...] SRC... [DEST]
8 Access via remote shell:
9 Pull: rsync [OPTION...] [USER@]HOST:SRC... [DEST]
10 Push: rsync [OPTION...] SRC... [USER@]HOST:DEST
12 Access via rsync daemon:
13 Pull: rsync [OPTION...] [USER@]HOST::SRC... [DEST]
14 rsync [OPTION...] rsync://[USER@]HOST[:PORT]/SRC... [DEST]
15 Push: rsync [OPTION...] SRC... [USER@]HOST::DEST
16 rsync [OPTION...] SRC... rsync://[USER@]HOST[:PORT]/DEST)
18 Usages with just one SRC arg and no DEST arg will list the source files
23 Rsync is a fast and extraordinarily versatile file copying tool. It can
24 copy locally, to/from another host over any remote shell, or to/from a
25 remote rsync daemon. It offers a large number of options that control
26 every aspect of its behavior and permit very flexible specification of the
27 set of files to be copied. It is famous for its delta-transfer algorithm,
28 which reduces the amount of data sent over the network by sending only the
29 differences between the source files and the existing files in the
30 destination. Rsync is widely used for backups and mirroring and as an
31 improved copy command for everyday use.
33 Rsync finds files that need to be transferred using a "quick check"
34 algorithm (by default) that looks for files that have changed in size or
35 in last-modified time. Any changes in the other preserved attributes (as
36 requested by options) are made on the destination file directly when the
37 quick check indicates that the file's data does not need to be updated.
39 Some of the additional features of rsync are:
42 it() support for copying links, devices, owners, groups, and permissions
43 it() exclude and exclude-from options similar to GNU tar
44 it() a CVS exclude mode for ignoring the same files that CVS would ignore
45 it() can use any transparent remote shell, including ssh or rsh
46 it() does not require super-user privileges
47 it() pipelining of file transfers to minimize latency costs
48 it() support for anonymous or authenticated rsync daemons (ideal for
52 manpagesection(GENERAL)
54 Rsync copies files either to or from a remote host, or locally on the
55 current host (it does not support copying files between two remote hosts).
57 There are two different ways for rsync to contact a remote system: using a
58 remote-shell program as the transport (such as ssh or rsh) or contacting an
59 rsync daemon directly via TCP. The remote-shell transport is used whenever
60 the source or destination path contains a single colon (:) separator after
61 a host specification. Contacting an rsync daemon directly happens when the
62 source or destination path contains a double colon (::) separator after a
63 host specification, OR when an rsync:// URL is specified (see also the
64 "USING RSYNC-DAEMON FEATURES VIA A REMOTE-SHELL CONNECTION" section for
65 an exception to this latter rule).
67 As a special case, if a single source arg is specified without a
68 destination, the files are listed in an output format similar to "ls -l".
70 As expected, if neither the source or destination path specify a remote
71 host, the copy occurs locally (see also the bf(--list-only) option).
73 Rsync refers to the local side as the "client" and the remote side as the
74 "server". Don't confuse "server" with an rsync daemon -- a daemon is always a
75 server, but a server can be either a daemon or a remote-shell spawned process.
79 See the file README for installation instructions.
81 Once installed, you can use rsync to any machine that you can access via
82 a remote shell (as well as some that you can access using the rsync
83 daemon-mode protocol). For remote transfers, a modern rsync uses ssh
84 for its communications, but it may have been configured to use a
85 different remote shell by default, such as rsh or remsh.
87 You can also specify any remote shell you like, either by using the bf(-e)
88 command line option, or by setting the RSYNC_RSH environment variable.
90 Note that rsync must be installed on both the source and destination
95 You use rsync in the same way you use rcp. You must specify a source
96 and a destination, one of which may be remote.
98 Perhaps the best way to explain the syntax is with some examples:
100 quote(tt(rsync -t *.c foo:src/))
102 This would transfer all files matching the pattern *.c from the
103 current directory to the directory src on the machine foo. If any of
104 the files already exist on the remote system then the rsync
105 remote-update protocol is used to update the file by sending only the
106 differences. See the tech report for details.
108 quote(tt(rsync -avz foo:src/bar /data/tmp))
110 This would recursively transfer all files from the directory src/bar on the
111 machine foo into the /data/tmp/bar directory on the local machine. The
112 files are transferred in "archive" mode, which ensures that symbolic
113 links, devices, attributes, permissions, ownerships, etc. are preserved
114 in the transfer. Additionally, compression will be used to reduce the
115 size of data portions of the transfer.
117 quote(tt(rsync -avz foo:src/bar/ /data/tmp))
119 A trailing slash on the source changes this behavior to avoid creating an
120 additional directory level at the destination. You can think of a trailing
121 / on a source as meaning "copy the contents of this directory" as opposed
122 to "copy the directory by name", but in both cases the attributes of the
123 containing directory are transferred to the containing directory on the
124 destination. In other words, each of the following commands copies the
125 files in the same way, including their setting of the attributes of
129 tt(rsync -av /src/foo /dest)nl()
130 tt(rsync -av /src/foo/ /dest/foo)nl()
133 Note also that host and module references don't require a trailing slash to
134 copy the contents of the default directory. For example, both of these
135 copy the remote directory's contents into "/dest":
138 tt(rsync -av host: /dest)nl()
139 tt(rsync -av host::module /dest)nl()
142 You can also use rsync in local-only mode, where both the source and
143 destination don't have a ':' in the name. In this case it behaves like
144 an improved copy command.
146 Finally, you can list all the (listable) modules available from a
147 particular rsync daemon by leaving off the module name:
149 quote(tt(rsync somehost.mydomain.com::))
151 See the following section for more details.
153 manpagesection(ADVANCED USAGE)
155 The syntax for requesting multiple files from a remote host is done by
156 specifying additional remote-host args in the same style as the first,
157 or with the hostname omitted. For instance, all these work:
159 quote(tt(rsync -av host:file1 :file2 host:file{3,4} /dest/)nl()
160 tt(rsync -av host::modname/file{1,2} host::modname/file3 /dest/)nl()
161 tt(rsync -av host::modname/file1 ::modname/file{3,4}))
163 Older versions of rsync required using quoted spaces in the SRC, like these
166 quote(tt(rsync -av host:'dir1/file1 dir2/file2' /dest)nl()
167 tt(rsync host::'modname/dir1/file1 modname/dir2/file2' /dest))
169 This word-splitting still works (by default) in the latest rsync, but is
170 not as easy to use as the first method.
172 If you need to transfer a filename that contains whitespace, you can either
173 specify the bf(--protect-args) (bf(-s)) option, or you'll need to escape
174 the whitespace in a way that the remote shell will understand. For
177 quote(tt(rsync -av host:'file\ name\ with\ spaces' /dest))
179 manpagesection(CONNECTING TO AN RSYNC DAEMON)
181 It is also possible to use rsync without a remote shell as the transport.
182 In this case you will directly connect to a remote rsync daemon, typically
183 using TCP port 873. (This obviously requires the daemon to be running on
184 the remote system, so refer to the STARTING AN RSYNC DAEMON TO ACCEPT
185 CONNECTIONS section below for information on that.)
187 Using rsync in this way is the same as using it with a remote shell except
191 it() you either use a double colon :: instead of a single colon to
192 separate the hostname from the path, or you use an rsync:// URL.
193 it() the first word of the "path" is actually a module name.
194 it() the remote daemon may print a message of the day when you
196 it() if you specify no path name on the remote daemon then the
197 list of accessible paths on the daemon will be shown.
198 it() if you specify no local destination then a listing of the
199 specified files on the remote daemon is provided.
200 it() you must not specify the bf(--rsh) (bf(-e)) option.
203 An example that copies all the files in a remote module named "src":
205 verb( rsync -av host::src /dest)
207 Some modules on the remote daemon may require authentication. If so,
208 you will receive a password prompt when you connect. You can avoid the
209 password prompt by setting the environment variable RSYNC_PASSWORD to
210 the password you want to use or using the bf(--password-file) option. This
211 may be useful when scripting rsync.
213 WARNING: On some systems environment variables are visible to all
214 users. On those systems using bf(--password-file) is recommended.
216 You may establish the connection via a web proxy by setting the
217 environment variable RSYNC_PROXY to a hostname:port pair pointing to
218 your web proxy. Note that your web proxy's configuration must support
219 proxy connections to port 873.
221 You may also establish a daemon connection using a program as a proxy by
222 setting the environment variable RSYNC_CONNECT_PROG to the commands you
223 wish to run in place of making a direct socket connection. The string may
224 contain the escape "%H" to represent the hostname specified in the rsync
225 command (so use "%%" if you need a single "%" in your string). For
228 verb( export RSYNC_CONNECT_PROG='ssh proxyhost nc %H 873'
229 rsync -av targethost1::module/src/ /dest/
230 rsync -av rsync:://targethost2/module/src/ /dest/ )
232 The command specified above uses ssh to run nc (netcat) on a proxyhost,
233 which forwards all data to port 873 (the rsync daemon) on the targethost
236 manpagesection(USING RSYNC-DAEMON FEATURES VIA A REMOTE-SHELL CONNECTION)
238 It is sometimes useful to use various features of an rsync daemon (such as
239 named modules) without actually allowing any new socket connections into a
240 system (other than what is already required to allow remote-shell access).
241 Rsync supports connecting to a host using a remote shell and then spawning
242 a single-use "daemon" server that expects to read its config file in the
243 home dir of the remote user. This can be useful if you want to encrypt a
244 daemon-style transfer's data, but since the daemon is started up fresh by
245 the remote user, you may not be able to use features such as chroot or
246 change the uid used by the daemon. (For another way to encrypt a daemon
247 transfer, consider using ssh to tunnel a local port to a remote machine and
248 configure a normal rsync daemon on that remote host to only allow
249 connections from "localhost".)
251 From the user's perspective, a daemon transfer via a remote-shell
252 connection uses nearly the same command-line syntax as a normal
253 rsync-daemon transfer, with the only exception being that you must
254 explicitly set the remote shell program on the command-line with the
255 bf(--rsh=COMMAND) option. (Setting the RSYNC_RSH in the environment
256 will not turn on this functionality.) For example:
258 verb( rsync -av --rsh=ssh host::module /dest)
260 If you need to specify a different remote-shell user, keep in mind that the
261 user@ prefix in front of the host is specifying the rsync-user value (for a
262 module that requires user-based authentication). This means that you must
263 give the '-l user' option to ssh when specifying the remote-shell, as in
264 this example that uses the short version of the bf(--rsh) option:
266 verb( rsync -av -e "ssh -l ssh-user" rsync-user@host::module /dest)
268 The "ssh-user" will be used at the ssh level; the "rsync-user" will be
269 used to log-in to the "module".
271 manpagesection(STARTING AN RSYNC DAEMON TO ACCEPT CONNECTIONS)
273 In order to connect to an rsync daemon, the remote system needs to have a
274 daemon already running (or it needs to have configured something like inetd
275 to spawn an rsync daemon for incoming connections on a particular port).
276 For full information on how to start a daemon that will handling incoming
277 socket connections, see the bf(rsyncd.conf)(5) man page -- that is the config
278 file for the daemon, and it contains the full details for how to run the
279 daemon (including stand-alone and inetd configurations).
281 If you're using one of the remote-shell transports for the transfer, there is
282 no need to manually start an rsync daemon.
284 manpagesection(EXAMPLES)
286 Here are some examples of how I use rsync.
288 To backup my wife's home directory, which consists of large MS Word
289 files and mail folders, I use a cron job that runs
291 quote(tt(rsync -Cavz . arvidsjaur:backup))
293 each night over a PPP connection to a duplicate directory on my machine
296 To synchronize my samba source trees I use the following Makefile
300 rsync -avuzb --exclude '*~' samba:samba/ .
302 rsync -Cavuzb . samba:samba/
305 this allows me to sync with a CVS directory at the other end of the
306 connection. I then do CVS operations on the remote machine, which saves a
307 lot of time as the remote CVS protocol isn't very efficient.
309 I mirror a directory between my "old" and "new" ftp sites with the
312 tt(rsync -az -e ssh --delete ~ftp/pub/samba nimbus:"~ftp/pub/tridge")
314 This is launched from cron every few hours.
316 manpagesection(OPTIONS SUMMARY)
318 Here is a short summary of the options available in rsync. Please refer
319 to the detailed description below for a complete description. verb(
320 -v, --verbose increase verbosity
321 --info=FLAGS fine-grained informational verbosity
322 --debug=FLAGS fine-grained debug verbosity
323 -q, --quiet suppress non-error messages
324 --no-motd suppress daemon-mode MOTD (see caveat)
325 -c, --checksum skip based on checksum, not mod-time & size
326 -a, --archive archive mode; equals -rlptgoD (no -H,-A,-X)
327 --no-OPTION turn off an implied OPTION (e.g. --no-D)
328 -r, --recursive recurse into directories
329 -R, --relative use relative path names
330 --no-implied-dirs don't send implied dirs with --relative
331 -b, --backup make backups (see --suffix & --backup-dir)
332 --backup-dir=DIR make backups into hierarchy based in DIR
333 --suffix=SUFFIX backup suffix (default ~ w/o --backup-dir)
334 -u, --update skip files that are newer on the receiver
335 --inplace update destination files in-place
336 --append append data onto shorter files
337 --append-verify --append w/old data in file checksum
338 -d, --dirs transfer directories without recursing
339 -l, --links copy symlinks as symlinks
340 -L, --copy-links transform symlink into referent file/dir
341 --copy-unsafe-links only "unsafe" symlinks are transformed
342 --safe-links ignore symlinks that point outside the tree
343 --munge-links munge symlinks to make them safer
344 -k, --copy-dirlinks transform symlink to dir into referent dir
345 -K, --keep-dirlinks treat symlinked dir on receiver as dir
346 -H, --hard-links preserve hard links
347 -p, --perms preserve permissions
348 -E, --executability preserve executability
349 --chmod=CHMOD affect file and/or directory permissions
350 -A, --acls preserve ACLs (implies -p)
351 -X, --xattrs preserve extended attributes
352 -o, --owner preserve owner (super-user only)
353 -g, --group preserve group
354 --devices preserve device files (super-user only)
355 --specials preserve special files
356 -D same as --devices --specials
357 -t, --times preserve modification times
358 -O, --omit-dir-times omit directories from --times
359 --super receiver attempts super-user activities
360 --fake-super store/recover privileged attrs using xattrs
361 -S, --sparse handle sparse files efficiently
362 -n, --dry-run perform a trial run with no changes made
363 -W, --whole-file copy files whole (w/o delta-xfer algorithm)
364 -x, --one-file-system don't cross filesystem boundaries
365 -B, --block-size=SIZE force a fixed checksum block-size
366 -e, --rsh=COMMAND specify the remote shell to use
367 --rsync-path=PROGRAM specify the rsync to run on remote machine
368 --existing skip creating new files on receiver
369 --ignore-existing skip updating files that exist on receiver
370 --remove-source-files sender removes synchronized files (non-dir)
371 --del an alias for --delete-during
372 --delete delete extraneous files from dest dirs
373 --delete-before receiver deletes before xfer, not during
374 --delete-during receiver deletes during transfer (default)
375 --delete-delay find deletions during, delete after
376 --delete-after receiver deletes after transfer, not before
377 --delete-excluded also delete excluded files from dest dirs
378 --ignore-missing-args ignore missing source args without error
379 --delete-missing-args delete missing source args from destination
380 --ignore-errors delete even if there are I/O errors
381 --force force deletion of dirs even if not empty
382 --max-delete=NUM don't delete more than NUM files
383 --max-size=SIZE don't transfer any file larger than SIZE
384 --min-size=SIZE don't transfer any file smaller than SIZE
385 --partial keep partially transferred files
386 --partial-dir=DIR put a partially transferred file into DIR
387 --delay-updates put all updated files into place at end
388 -m, --prune-empty-dirs prune empty directory chains from file-list
389 --numeric-ids don't map uid/gid values by user/group name
390 --usermap=STRING custom username mapping
391 --groupmap=STRING custom groupname mapping
392 --chown=USER:GROUP simple username/groupname mapping
393 --timeout=SECONDS set I/O timeout in seconds
394 --contimeout=SECONDS set daemon connection timeout in seconds
395 -I, --ignore-times don't skip files that match size and time
396 --size-only skip files that match in size
397 --modify-window=NUM compare mod-times with reduced accuracy
398 -T, --temp-dir=DIR create temporary files in directory DIR
399 -y, --fuzzy find similar file for basis if no dest file
400 --compare-dest=DIR also compare received files relative to DIR
401 --copy-dest=DIR ... and include copies of unchanged files
402 --link-dest=DIR hardlink to files in DIR when unchanged
403 -z, --compress compress file data during the transfer
404 --compress-level=NUM explicitly set compression level
405 --skip-compress=LIST skip compressing files with suffix in LIST
406 -C, --cvs-exclude auto-ignore files in the same way CVS does
407 -f, --filter=RULE add a file-filtering RULE
408 -F same as --filter='dir-merge /.rsync-filter'
409 repeated: --filter='- .rsync-filter'
410 --exclude=PATTERN exclude files matching PATTERN
411 --exclude-from=FILE read exclude patterns from FILE
412 --include=PATTERN don't exclude files matching PATTERN
413 --include-from=FILE read include patterns from FILE
414 --files-from=FILE read list of source-file names from FILE
415 -0, --from0 all *from/filter files are delimited by 0s
416 -s, --protect-args no space-splitting; wildcard chars only
417 --address=ADDRESS bind address for outgoing socket to daemon
418 --port=PORT specify double-colon alternate port number
419 --sockopts=OPTIONS specify custom TCP options
420 --blocking-io use blocking I/O for the remote shell
421 --stats give some file-transfer stats
422 -8, --8-bit-output leave high-bit chars unescaped in output
423 -h, --human-readable output numbers in a human-readable format
424 --progress show progress during transfer
425 -P same as --partial --progress
426 -i, --itemize-changes output a change-summary for all updates
427 -M, --remote-option=OPTION send OPTION to the remote side only
428 --out-format=FORMAT output updates using the specified FORMAT
429 --log-file=FILE log what we're doing to the specified FILE
430 --log-file-format=FMT log updates using the specified FMT
431 --password-file=FILE read daemon-access password from FILE
432 --list-only list the files instead of copying them
433 --bwlimit=KBPS limit I/O bandwidth; KBytes per second
434 --write-batch=FILE write a batched update to FILE
435 --only-write-batch=FILE like --write-batch but w/o updating dest
436 --read-batch=FILE read a batched update from FILE
437 --protocol=NUM force an older protocol version to be used
438 --iconv=CONVERT_SPEC request charset conversion of filenames
439 --checksum-seed=NUM set block/file checksum seed (advanced)
440 -4, --ipv4 prefer IPv4
441 -6, --ipv6 prefer IPv6
442 --version print version number
443 (-h) --help show this help (see below for -h comment))
445 Rsync can also be run as a daemon, in which case the following options are
447 --daemon run as an rsync daemon
448 --address=ADDRESS bind to the specified address
449 --bwlimit=KBPS limit I/O bandwidth; KBytes per second
450 --config=FILE specify alternate rsyncd.conf file
451 -M, --dparam=OVERRIDE override global daemon config parameter
452 --no-detach do not detach from the parent
453 --port=PORT listen on alternate port number
454 --log-file=FILE override the "log file" setting
455 --log-file-format=FMT override the "log format" setting
456 --sockopts=OPTIONS specify custom TCP options
457 -v, --verbose increase verbosity
458 -4, --ipv4 prefer IPv4
459 -6, --ipv6 prefer IPv6
460 -h, --help show this help (if used after --daemon))
464 rsync uses the GNU long options package. Many of the command line
465 options have two variants, one short and one long. These are shown
466 below, separated by commas. Some options only have a long variant.
467 The '=' for options that take a parameter is optional; whitespace
471 dit(bf(--help)) Print a short help page describing the options
472 available in rsync and exit. For backward-compatibility with older
473 versions of rsync, the help will also be output if you use the bf(-h)
474 option without any other args.
476 dit(bf(--version)) print the rsync version number and exit.
478 dit(bf(-v, --verbose)) This option increases the amount of information you
479 are given during the transfer. By default, rsync works silently. A
480 single bf(-v) will give you information about what files are being
481 transferred and a brief summary at the end. Two bf(-v) options will give you
482 information on what files are being skipped and slightly more
483 information at the end. More than two bf(-v) options should only be used if
484 you are debugging rsync.
486 In a modern rsync, the bf(-v) option is equivalent to the setting of groups
487 of bf(--info) and bf(--debug) options. You can choose to use these newer
488 options in addition to, or in place of using bf(--verbose), as any
489 fine-grained settings override the implied settings of bf(-v). Both
490 bf(--info) and bf(--debug) have a way to ask for help that tells you
491 exactly what flags are set for each increase in verbosity.
493 dit(bf(--info=FLAGS))
494 This option lets you have fine-grained control over the
496 output you want to see. An individual flag name may be followed by a level
497 number, with 0 meaning to silence that output, 1 being the default output
498 level, and higher numbers increasing the output of that flag (for those
499 that support higher levels). Use
501 to see all the available flag names, what they output, and what flag names
502 are added for each increase in the verbose level. Some examples:
504 verb( rsync -a --info=progress2 src/ dest/
505 rsync -avv --info=stats2,misc1,flist0 src/ dest/ )
507 Note that bf(--info=name)'s output is affected by the bf(--out-format) and
508 bf(--itemize-changes) (bf(-i)) options. See those options for more
509 information on what is output and when.
511 This option was added to 3.1.0, so an older rsync on the server side might
512 reject your attempts at fine-grained control (if one or more flags needed
513 to be send to the server and the server was too old to understand them).
515 dit(bf(--debug=FLAGS))
516 This option lets you have fine-grained control over the
518 output you want to see. An individual flag name may be followed by a level
519 number, with 0 meaning to silence that output, 1 being the default output
520 level, and higher numbers increasing the output of that flag (for those
521 that support higher levels). Use
523 to see all the available flag names, what they output, and what flag names
524 are added for each increase in the verbose level. Some examples:
526 verb( rsync -avvv --debug=none src/ dest/
527 rsync -avA --del --debug=del2,acl src/ dest/ )
529 This option was added to 3.1.0, so an older rsync on the server side might
530 reject your attempts at fine-grained control (if one or more flags needed
531 to be send to the server and the server was too old to understand them).
533 dit(bf(-q, --quiet)) This option decreases the amount of information you
534 are given during the transfer, notably suppressing information messages
535 from the remote server. This option name is useful when invoking rsync from
538 dit(bf(--no-motd)) This option affects the information that is output
539 by the client at the start of a daemon transfer. This suppresses the
540 message-of-the-day (MOTD) text, but it also affects the list of modules
541 that the daemon sends in response to the "rsync host::" request (due to
542 a limitation in the rsync protocol), so omit this option if you want to
543 request the list of modules from the daemon.
545 dit(bf(-I, --ignore-times)) Normally rsync will skip any files that are
546 already the same size and have the same modification timestamp.
547 This option turns off this "quick check" behavior, causing all files to
550 dit(bf(--size-only)) This modifies rsync's "quick check" algorithm for
551 finding files that need to be transferred, changing it from the default of
552 transferring files with either a changed size or a changed last-modified
553 time to just looking for files that have changed in size. This is useful
554 when starting to use rsync after using another mirroring system which may
555 not preserve timestamps exactly.
557 dit(bf(--modify-window)) When comparing two timestamps, rsync treats the
558 timestamps as being equal if they differ by no more than the modify-window
559 value. This is normally 0 (for an exact match), but you may find it useful
560 to set this to a larger value in some situations. In particular, when
561 transferring to or from an MS Windows FAT filesystem (which represents
562 times with a 2-second resolution), bf(--modify-window=1) is useful
563 (allowing times to differ by up to 1 second).
565 dit(bf(-c, --checksum)) This changes the way rsync checks if the files have
566 been changed and are in need of a transfer. Without this option, rsync
567 uses a "quick check" that (by default) checks if each file's size and time
568 of last modification match between the sender and receiver. This option
569 changes this to compare a 128-bit checksum for each file that has a
570 matching size. Generating the checksums means that both sides will expend
571 a lot of disk I/O reading all the data in the files in the transfer (and
572 this is prior to any reading that will be done to transfer changed files),
573 so this can slow things down significantly.
575 The sending side generates its checksums while it is doing the file-system
576 scan that builds the list of the available files. The receiver generates
577 its checksums when it is scanning for changed files, and will checksum any
578 file that has the same size as the corresponding sender's file: files with
579 either a changed size or a changed checksum are selected for transfer.
581 Note that rsync always verifies that each em(transferred) file was
582 correctly reconstructed on the receiving side by checking a whole-file
583 checksum that is generated as the file is transferred, but that
584 automatic after-the-transfer verification has nothing to do with this
585 option's before-the-transfer "Does this file need to be updated?" check.
587 For protocol 30 and beyond (first supported in 3.0.0), the checksum used is
588 MD5. For older protocols, the checksum used is MD4.
590 dit(bf(-a, --archive)) This is equivalent to bf(-rlptgoD). It is a quick
591 way of saying you want recursion and want to preserve almost
592 everything (with -H being a notable omission).
593 The only exception to the above equivalence is when bf(--files-from) is
594 specified, in which case bf(-r) is not implied.
596 Note that bf(-a) bf(does not preserve hardlinks), because
597 finding multiply-linked files is expensive. You must separately
600 dit(--no-OPTION) You may turn off one or more implied options by prefixing
601 the option name with "no-". Not all options may be prefixed with a "no-":
602 only options that are implied by other options (e.g. bf(--no-D),
603 bf(--no-perms)) or have different defaults in various circumstances
604 (e.g. bf(--no-whole-file), bf(--no-blocking-io), bf(--no-dirs)). You may
605 specify either the short or the long option name after the "no-" prefix
606 (e.g. bf(--no-R) is the same as bf(--no-relative)).
608 For example: if you want to use bf(-a) (bf(--archive)) but don't want
609 bf(-o) (bf(--owner)), instead of converting bf(-a) into bf(-rlptgD), you
610 could specify bf(-a --no-o) (or bf(-a --no-owner)).
612 The order of the options is important: if you specify bf(--no-r -a), the
613 bf(-r) option would end up being turned on, the opposite of bf(-a --no-r).
614 Note also that the side-effects of the bf(--files-from) option are NOT
615 positional, as it affects the default state of several options and slightly
616 changes the meaning of bf(-a) (see the bf(--files-from) option for more
619 dit(bf(-r, --recursive)) This tells rsync to copy directories
620 recursively. See also bf(--dirs) (bf(-d)).
622 Beginning with rsync 3.0.0, the recursive algorithm used is now an
623 incremental scan that uses much less memory than before and begins the
624 transfer after the scanning of the first few directories have been
625 completed. This incremental scan only affects our recursion algorithm, and
626 does not change a non-recursive transfer. It is also only possible when
627 both ends of the transfer are at least version 3.0.0.
629 Some options require rsync to know the full file list, so these options
630 disable the incremental recursion mode. These include: bf(--delete-before),
631 bf(--delete-after), bf(--prune-empty-dirs), and bf(--delay-updates).
632 Because of this, the default delete mode when you specify bf(--delete) is now
633 bf(--delete-during) when both ends of the connection are at least 3.0.0
634 (use bf(--del) or bf(--delete-during) to request this improved deletion mode
635 explicitly). See also the bf(--delete-delay) option that is a better choice
636 than using bf(--delete-after).
638 Incremental recursion can be disabled using the bf(--no-inc-recursive)
639 option or its shorter bf(--no-i-r) alias.
641 dit(bf(-R, --relative)) Use relative paths. This means that the full path
642 names specified on the command line are sent to the server rather than
643 just the last parts of the filenames. This is particularly useful when
644 you want to send several different directories at the same time. For
645 example, if you used this command:
647 quote(tt( rsync -av /foo/bar/baz.c remote:/tmp/))
649 ... this would create a file named baz.c in /tmp/ on the remote
650 machine. If instead you used
652 quote(tt( rsync -avR /foo/bar/baz.c remote:/tmp/))
654 then a file named /tmp/foo/bar/baz.c would be created on the remote
655 machine, preserving its full path. These extra path elements are called
656 "implied directories" (i.e. the "foo" and the "foo/bar" directories in the
659 Beginning with rsync 3.0.0, rsync always sends these implied directories as
660 real directories in the file list, even if a path element is really a
661 symlink on the sending side. This prevents some really unexpected
662 behaviors when copying the full path of a file that you didn't realize had
663 a symlink in its path. If you want to duplicate a server-side symlink,
664 include both the symlink via its path, and referent directory via its real
665 path. If you're dealing with an older rsync on the sending side, you may
666 need to use the bf(--no-implied-dirs) option.
668 It is also possible to limit the amount of path information that is sent as
669 implied directories for each path you specify. With a modern rsync on the
670 sending side (beginning with 2.6.7), you can insert a dot and a slash into
671 the source path, like this:
673 quote(tt( rsync -avR /foo/./bar/baz.c remote:/tmp/))
675 That would create /tmp/bar/baz.c on the remote machine. (Note that the
676 dot must be followed by a slash, so "/foo/." would not be abbreviated.)
677 (2) For older rsync versions, you would need to use a chdir to limit the
678 source path. For example, when pushing files:
680 quote(tt( (cd /foo; rsync -avR bar/baz.c remote:/tmp/) ))
682 (Note that the parens put the two commands into a sub-shell, so that the
683 "cd" command doesn't remain in effect for future commands.)
684 If you're pulling files from an older rsync, use this idiom (but only
685 for a non-daemon transfer):
688 tt( rsync -avR --rsync-path="cd /foo; rsync" \ )nl()
689 tt( remote:bar/baz.c /tmp/)
692 dit(bf(--no-implied-dirs)) This option affects the default behavior of the
693 bf(--relative) option. When it is specified, the attributes of the implied
694 directories from the source names are not included in the transfer. This
695 means that the corresponding path elements on the destination system are
696 left unchanged if they exist, and any missing implied directories are
697 created with default attributes. This even allows these implied path
698 elements to have big differences, such as being a symlink to a directory on
701 For instance, if a command-line arg or a files-from entry told rsync to
702 transfer the file "path/foo/file", the directories "path" and "path/foo"
703 are implied when bf(--relative) is used. If "path/foo" is a symlink to
704 "bar" on the destination system, the receiving rsync would ordinarily
705 delete "path/foo", recreate it as a directory, and receive the file into
706 the new directory. With bf(--no-implied-dirs), the receiving rsync updates
707 "path/foo/file" using the existing path elements, which means that the file
708 ends up being created in "path/bar". Another way to accomplish this link
709 preservation is to use the bf(--keep-dirlinks) option (which will also
710 affect symlinks to directories in the rest of the transfer).
712 When pulling files from an rsync older than 3.0.0, you may need to use this
713 option if the sending side has a symlink in the path you request and you
714 wish the implied directories to be transferred as normal directories.
716 dit(bf(-b, --backup)) With this option, preexisting destination files are
717 renamed as each file is transferred or deleted. You can control where the
718 backup file goes and what (if any) suffix gets appended using the
719 bf(--backup-dir) and bf(--suffix) options.
721 Note that if you don't specify bf(--backup-dir), (1) the
722 bf(--omit-dir-times) option will be implied, and (2) if bf(--delete) is
723 also in effect (without bf(--delete-excluded)), rsync will add a "protect"
724 filter-rule for the backup suffix to the end of all your existing excludes
725 (e.g. bf(-f "P *~")). This will prevent previously backed-up files from being
726 deleted. Note that if you are supplying your own filter rules, you may
727 need to manually insert your own exclude/protect rule somewhere higher up
728 in the list so that it has a high enough priority to be effective (e.g., if
729 your rules specify a trailing inclusion/exclusion of '*', the auto-added
730 rule would never be reached).
732 dit(bf(--backup-dir=DIR)) In combination with the bf(--backup) option, this
733 tells rsync to store all backups in the specified directory on the receiving
734 side. This can be used for incremental backups. You can additionally
735 specify a backup suffix using the bf(--suffix) option
736 (otherwise the files backed up in the specified directory
737 will keep their original filenames).
739 dit(bf(--suffix=SUFFIX)) This option allows you to override the default
740 backup suffix used with the bf(--backup) (bf(-b)) option. The default suffix is a ~
741 if no -bf(-backup-dir) was specified, otherwise it is an empty string.
743 dit(bf(-u, --update)) This forces rsync to skip any files which exist on
744 the destination and have a modified time that is newer than the source
745 file. (If an existing destination file has a modification time equal to the
746 source file's, it will be updated if the sizes are different.)
748 Note that this does not affect the copying of symlinks or other special
749 files. Also, a difference of file format between the sender and receiver
750 is always considered to be important enough for an update, no matter what
751 date is on the objects. In other words, if the source has a directory
752 where the destination has a file, the transfer would occur regardless of
755 This option is a transfer rule, not an exclude, so it doesn't affect the
756 data that goes into the file-lists, and thus it doesn't affect deletions.
757 It just limits the files that the receiver requests to be transferred.
759 dit(bf(--inplace)) This option changes how rsync transfers a file when the
760 file's data needs to be updated: instead of the default method of creating
761 a new copy of the file and moving it into place when it is complete, rsync
762 instead writes the updated data directly to the destination file.
764 This has several effects: (1) in-use binaries cannot be updated (either the
765 OS will prevent this from happening, or binaries that attempt to swap-in
766 their data will misbehave or crash), (2) the file's data will be in an
767 inconsistent state during the transfer, (3) a file's data may be left in an
768 inconsistent state after the transfer if the transfer is interrupted or if
769 an update fails, (4) a file that does not have write permissions can not be
770 updated, and (5) the efficiency of rsync's delta-transfer algorithm may be
771 reduced if some data in the destination file is overwritten before it can
772 be copied to a position later in the file (one exception to this is if you
773 combine this option with bf(--backup), since rsync is smart enough to use
774 the backup file as the basis file for the transfer).
776 WARNING: you should not use this option to update files that are being
777 accessed by others, so be careful when choosing to use this for a copy.
779 This option is useful for transfer of large files with block-based changes
780 or appended data, and also on systems that are disk bound, not network
783 The option implies bf(--partial) (since an interrupted transfer does not delete
784 the file), but conflicts with bf(--partial-dir) and bf(--delay-updates).
785 Prior to rsync 2.6.4 bf(--inplace) was also incompatible with bf(--compare-dest)
788 dit(bf(--append)) This causes rsync to update a file by appending data onto
789 the end of the file, which presumes that the data that already exists on
790 the receiving side is identical with the start of the file on the sending
791 side. If a file needs to be transferred and its size on the receiver is
792 the same or longer than the size on the sender, the file is skipped. This
793 does not interfere with the updating of a file's non-content attributes
794 (e.g. permissions, ownership, etc.) when the file does not need to be
795 transferred, nor does it affect the updating of any non-regular files.
796 Implies bf(--inplace),
797 but does not conflict with bf(--sparse) (since it is always extending a
800 dit(bf(--append-verify)) This works just like the bf(--append) option, but
801 the existing data on the receiving side is included in the full-file
802 checksum verification step, which will cause a file to be resent if the
803 final verification step fails (rsync uses a normal, non-appending
804 bf(--inplace) transfer for the resend).
806 Note: prior to rsync 3.0.0, the bf(--append) option worked like
807 bf(--append-verify), so if you are interacting with an older rsync (or the
808 transfer is using a protocol prior to 30), specifying either append option
809 will initiate an bf(--append-verify) transfer.
811 dit(bf(-d, --dirs)) Tell the sending side to include any directories that
812 are encountered. Unlike bf(--recursive), a directory's contents are not copied
813 unless the directory name specified is "." or ends with a trailing slash
814 (e.g. ".", "dir/.", "dir/", etc.). Without this option or the
815 bf(--recursive) option, rsync will skip all directories it encounters (and
816 output a message to that effect for each one). If you specify both
817 bf(--dirs) and bf(--recursive), bf(--recursive) takes precedence.
819 The bf(--dirs) option is implied by the bf(--files-from) option
820 or the bf(--list-only) option (including an implied
821 bf(--list-only) usage) if bf(--recursive) wasn't specified (so that
822 directories are seen in the listing). Specify bf(--no-dirs) (or bf(--no-d))
823 if you want to turn this off.
825 There is also a backward-compatibility helper option, bf(--old-dirs) (or
826 bf(--old-d)) that tells rsync to use a hack of "-r --exclude='/*/*'" to get
827 an older rsync to list a single directory without recursing.
829 dit(bf(-l, --links)) When symlinks are encountered, recreate the
830 symlink on the destination.
832 dit(bf(-L, --copy-links)) When symlinks are encountered, the item that
833 they point to (the referent) is copied, rather than the symlink. In older
834 versions of rsync, this option also had the side-effect of telling the
835 receiving side to follow symlinks, such as symlinks to directories. In a
836 modern rsync such as this one, you'll need to specify bf(--keep-dirlinks) (bf(-K))
837 to get this extra behavior. The only exception is when sending files to
838 an rsync that is too old to understand bf(-K) -- in that case, the bf(-L) option
839 will still have the side-effect of bf(-K) on that older receiving rsync.
841 dit(bf(--copy-unsafe-links)) This tells rsync to copy the referent of
842 symbolic links that point outside the copied tree. Absolute symlinks
843 are also treated like ordinary files, and so are any symlinks in the
844 source path itself when bf(--relative) is used. This option has no
845 additional effect if bf(--copy-links) was also specified.
847 dit(bf(--safe-links)) This tells rsync to ignore any symbolic links
848 which point outside the copied tree. All absolute symlinks are
849 also ignored. Using this option in conjunction with bf(--relative) may
850 give unexpected results.
852 dit(bf(--munge-links)) This option tells rsync to (1) modify all symlinks on
853 the receiving side in a way that makes them unusable but recoverable (see
854 below), or (2) to unmunge symlinks on the sending side that had been stored in
855 a munged state. This is useful if you don't quite trust the source of the data
856 to not try to slip in a symlink to a unexpected place.
858 The way rsync disables the use of symlinks is to prefix each one with the
859 string "/rsyncd-munged/". This prevents the links from being used as long as
860 that directory does not exist. When this option is enabled, rsync will refuse
861 to run if that path is a directory or a symlink to a directory.
863 The option only affects the client side of the transfer, so if you need it to
864 affect the server, specify it via bf(--remote-option). (Note that in a local
865 transfer, the client side is the sender.)
867 This option has no affect on a daemon, since the daemon configures whether it
868 wants munged symlinks via its "munge symlinks" parameter. See also the
869 "munge-symlinks" perl script in the support directory of the source code.
871 dit(bf(-k, --copy-dirlinks)) This option causes the sending side to treat
872 a symlink to a directory as though it were a real directory. This is
873 useful if you don't want symlinks to non-directories to be affected, as
874 they would be using bf(--copy-links).
876 Without this option, if the sending side has replaced a directory with a
877 symlink to a directory, the receiving side will delete anything that is in
878 the way of the new symlink, including a directory hierarchy (as long as
879 bf(--force) or bf(--delete) is in effect).
881 See also bf(--keep-dirlinks) for an analogous option for the receiving
884 bf(--copy-dirlinks) applies to all symlinks to directories in the source. If
885 you want to follow only a few specified symlinks, a trick you can use is to
886 pass them as additional source args with a trailing slash, using bf(--relative)
887 to make the paths match up right. For example:
889 quote(tt(rsync -r --relative src/./ src/./follow-me/ dest/))
891 This works because rsync calls bf(lstat)(2) on the source arg as given, and the
892 trailing slash makes bf(lstat)(2) follow the symlink, giving rise to a directory
893 in the file-list which overrides the symlink found during the scan of "src/./".
895 dit(bf(-K, --keep-dirlinks)) This option causes the receiving side to treat
896 a symlink to a directory as though it were a real directory, but only if it
897 matches a real directory from the sender. Without this option, the
898 receiver's symlink would be deleted and replaced with a real directory.
900 For example, suppose you transfer a directory "foo" that contains a file
901 "file", but "foo" is a symlink to directory "bar" on the receiver. Without
902 bf(--keep-dirlinks), the receiver deletes symlink "foo", recreates it as a
903 directory, and receives the file into the new directory. With
904 bf(--keep-dirlinks), the receiver keeps the symlink and "file" ends up in
907 One note of caution: if you use bf(--keep-dirlinks), you must trust all
908 the symlinks in the copy! If it is possible for an untrusted user to
909 create their own symlink to any directory, the user could then (on a
910 subsequent copy) replace the symlink with a real directory and affect the
911 content of whatever directory the symlink references. For backup copies,
912 you are better off using something like a bind mount instead of a symlink
913 to modify your receiving hierarchy.
915 See also bf(--copy-dirlinks) for an analogous option for the sending side.
917 dit(bf(-H, --hard-links)) This tells rsync to look for hard-linked files in
918 the transfer and link together the corresponding files on the receiving
919 side. Without this option, hard-linked files in the transfer are treated
920 as though they were separate files.
922 When you are updating a non-empty destination, this option only ensures
923 that files that are hard-linked together on the source are hard-linked
924 together on the destination. It does NOT currently endeavor to break
925 already existing hard links on the destination that do not exist between
926 the source files. Note, however, that if one or more extra-linked files
927 have content changes, they will become unlinked when updated (assuming you
928 are not using the bf(--inplace) option).
930 Note that rsync can only detect hard links between files that are inside
931 the transfer set. If rsync updates a file that has extra hard-link
932 connections to files outside the transfer, that linkage will be broken. If
933 you are tempted to use the bf(--inplace) option to avoid this breakage, be
934 very careful that you know how your files are being updated so that you are
935 certain that no unintended changes happen due to lingering hard links (and
936 see the bf(--inplace) option for more caveats).
938 If incremental recursion is active (see bf(--recursive)), rsync may transfer
939 a missing hard-linked file before it finds that another link for that contents
940 exists elsewhere in the hierarchy. This does not affect the accuracy of
941 the transfer, just its efficiency. One way to avoid this is to disable
942 incremental recursion using the bf(--no-inc-recursive) option.
944 dit(bf(-p, --perms)) This option causes the receiving rsync to set the
945 destination permissions to be the same as the source permissions. (See
946 also the bf(--chmod) option for a way to modify what rsync considers to
947 be the source permissions.)
949 When this option is em(off), permissions are set as follows:
952 it() Existing files (including updated files) retain their existing
953 permissions, though the bf(--executability) option might change just
954 the execute permission for the file.
955 it() New files get their "normal" permission bits set to the source
956 file's permissions masked with the receiving directory's default
957 permissions (either the receiving process's umask, or the permissions
958 specified via the destination directory's default ACL), and
959 their special permission bits disabled except in the case where a new
960 directory inherits a setgid bit from its parent directory.
963 Thus, when bf(--perms) and bf(--executability) are both disabled,
964 rsync's behavior is the same as that of other file-copy utilities,
965 such as bf(cp)(1) and bf(tar)(1).
967 In summary: to give destination files (both old and new) the source
968 permissions, use bf(--perms). To give new files the destination-default
969 permissions (while leaving existing files unchanged), make sure that the
970 bf(--perms) option is off and use bf(--chmod=ugo=rwX) (which ensures that
971 all non-masked bits get enabled). If you'd care to make this latter
972 behavior easier to type, you could define a popt alias for it, such as
973 putting this line in the file ~/.popt (the following defines the bf(-Z) option,
974 and includes --no-g to use the default group of the destination dir):
976 quote(tt( rsync alias -Z --no-p --no-g --chmod=ugo=rwX))
978 You could then use this new option in a command such as this one:
980 quote(tt( rsync -avZ src/ dest/))
982 (Caveat: make sure that bf(-a) does not follow bf(-Z), or it will re-enable
983 the two "--no-*" options mentioned above.)
985 The preservation of the destination's setgid bit on newly-created
986 directories when bf(--perms) is off was added in rsync 2.6.7. Older rsync
987 versions erroneously preserved the three special permission bits for
988 newly-created files when bf(--perms) was off, while overriding the
989 destination's setgid bit setting on a newly-created directory. Default ACL
990 observance was added to the ACL patch for rsync 2.6.7, so older (or
991 non-ACL-enabled) rsyncs use the umask even if default ACLs are present.
992 (Keep in mind that it is the version of the receiving rsync that affects
995 dit(bf(-E, --executability)) This option causes rsync to preserve the
996 executability (or non-executability) of regular files when bf(--perms) is
997 not enabled. A regular file is considered to be executable if at least one
998 'x' is turned on in its permissions. When an existing destination file's
999 executability differs from that of the corresponding source file, rsync
1000 modifies the destination file's permissions as follows:
1003 it() To make a file non-executable, rsync turns off all its 'x'
1005 it() To make a file executable, rsync turns on each 'x' permission that
1006 has a corresponding 'r' permission enabled.
1009 If bf(--perms) is enabled, this option is ignored.
1011 dit(bf(-A, --acls)) This option causes rsync to update the destination
1012 ACLs to be the same as the source ACLs.
1013 The option also implies bf(--perms).
1015 The source and destination systems must have compatible ACL entries for this
1016 option to work properly. See the bf(--fake-super) option for a way to backup
1017 and restore ACLs that are not compatible.
1019 dit(bf(-X, --xattrs)) This option causes rsync to update the remote
1020 extended attributes to be the same as the local ones.
1022 For systems that support extended-attribute namespaces, a copy being done by a
1023 super-user copies all namespaces except system.*. A normal user only copies
1024 the user.* namespace. To be able to backup and restore non-user namespaces as
1025 a normal user, see the bf(--fake-super) option.
1027 dit(bf(--chmod)) This option tells rsync to apply one or more
1028 comma-separated "chmod" strings to the permission of the files in the
1029 transfer. The resulting value is treated as though it was the permissions
1030 that the sending side supplied for the file, which means that this option
1031 can seem to have no effect on existing files if bf(--perms) is not enabled.
1033 In addition to the normal parsing rules specified in the bf(chmod)(1)
1034 manpage, you can specify an item that should only apply to a directory by
1035 prefixing it with a 'D', or specify an item that should only apply to a
1036 file by prefixing it with a 'F'. For example:
1038 quote(--chmod=Dg+s,ug+w,Fo-w,+X)
1040 It is also legal to specify multiple bf(--chmod) options, as each
1041 additional option is just appended to the list of changes to make.
1043 See the bf(--perms) and bf(--executability) options for how the resulting
1044 permission value can be applied to the files in the transfer.
1046 dit(bf(-o, --owner)) This option causes rsync to set the owner of the
1047 destination file to be the same as the source file, but only if the
1048 receiving rsync is being run as the super-user (see also the bf(--super)
1049 and bf(--fake-super) options).
1050 Without this option, the owner of new and/or transferred files are set to
1051 the invoking user on the receiving side.
1053 The preservation of ownership will associate matching names by default, but
1054 may fall back to using the ID number in some circumstances (see also the
1055 bf(--numeric-ids) option for a full discussion).
1057 dit(bf(-g, --group)) This option causes rsync to set the group of the
1058 destination file to be the same as the source file. If the receiving
1059 program is not running as the super-user (or if bf(--no-super) was
1060 specified), only groups that the invoking user on the receiving side
1061 is a member of will be preserved.
1062 Without this option, the group is set to the default group of the invoking
1063 user on the receiving side.
1065 The preservation of group information will associate matching names by
1066 default, but may fall back to using the ID number in some circumstances
1067 (see also the bf(--numeric-ids) option for a full discussion).
1069 dit(bf(--devices)) This option causes rsync to transfer character and
1070 block device files to the remote system to recreate these devices.
1071 This option has no effect if the receiving rsync is not run as the
1072 super-user (see also the bf(--super) and bf(--fake-super) options).
1074 dit(bf(--specials)) This option causes rsync to transfer special files
1075 such as named sockets and fifos.
1077 dit(bf(-D)) The bf(-D) option is equivalent to bf(--devices) bf(--specials).
1079 dit(bf(-t, --times)) This tells rsync to transfer modification times along
1080 with the files and update them on the remote system. Note that if this
1081 option is not used, the optimization that excludes files that have not been
1082 modified cannot be effective; in other words, a missing bf(-t) or bf(-a) will
1083 cause the next transfer to behave as if it used bf(-I), causing all files to be
1084 updated (though rsync's delta-transfer algorithm will make the update fairly efficient
1085 if the files haven't actually changed, you're much better off using bf(-t)).
1087 dit(bf(-O, --omit-dir-times)) This tells rsync to omit directories when
1088 it is preserving modification times (see bf(--times)). If NFS is sharing
1089 the directories on the receiving side, it is a good idea to use bf(-O).
1090 This option is inferred if you use bf(--backup) without bf(--backup-dir).
1092 dit(bf(--super)) This tells the receiving side to attempt super-user
1093 activities even if the receiving rsync wasn't run by the super-user. These
1094 activities include: preserving users via the bf(--owner) option, preserving
1095 all groups (not just the current user's groups) via the bf(--groups)
1096 option, and copying devices via the bf(--devices) option. This is useful
1097 for systems that allow such activities without being the super-user, and
1098 also for ensuring that you will get errors if the receiving side isn't
1099 being run as the super-user. To turn off super-user activities, the
1100 super-user can use bf(--no-super).
1102 dit(bf(--fake-super)) When this option is enabled, rsync simulates
1103 super-user activities by saving/restoring the privileged attributes via
1104 special extended attributes that are attached to each file (as needed). This
1105 includes the file's owner and group (if it is not the default), the file's
1106 device info (device & special files are created as empty text files), and
1107 any permission bits that we won't allow to be set on the real file (e.g.
1108 the real file gets u-s,g-s,o-t for safety) or that would limit the owner's
1109 access (since the real super-user can always access/change a file, the
1110 files we create can always be accessed/changed by the creating user).
1111 This option also handles ACLs (if bf(--acls) was specified) and non-user
1112 extended attributes (if bf(--xattrs) was specified).
1114 This is a good way to backup data without using a super-user, and to store
1115 ACLs from incompatible systems.
1117 The bf(--fake-super) option only affects the side where the option is used.
1118 To affect the remote side of a remote-shell connection, use the
1119 bf(--remote-option) (bf(-M)) option:
1121 quote(tt( rsync -av -M--fake-super /src/ host:/dest/))
1123 For a local copy, this option affects both the source and the destination.
1124 If you wish a local copy to enable this option just for the destination
1125 files, specify bf(-M--fake-super). If you wish a local copy to enable
1126 this option just for the source files, combine bf(--fake-super) with
1129 This option is overridden by both bf(--super) and bf(--no-super).
1131 See also the "fake super" setting in the daemon's rsyncd.conf file.
1133 dit(bf(-S, --sparse)) Try to handle sparse files efficiently so they take
1134 up less space on the destination. Conflicts with bf(--inplace) because it's
1135 not possible to overwrite data in a sparse fashion.
1137 NOTE: Don't use this option when the destination is a Solaris "tmpfs"
1138 filesystem. It doesn't seem to handle seeks over null regions
1139 correctly and ends up corrupting the files.
1141 dit(bf(-n, --dry-run)) This makes rsync perform a trial run that doesn't
1142 make any changes (and produces mostly the same output as a real run). It
1143 is most commonly used in combination with the bf(-v, --verbose) and/or
1144 bf(-i, --itemize-changes) options to see what an rsync command is going
1145 to do before one actually runs it.
1147 The output of bf(--itemize-changes) is supposed to be exactly the same on a
1148 dry run and a subsequent real run (barring intentional trickery and system
1149 call failures); if it isn't, that's a bug. Other output is the same to the
1150 extent practical, but may differ in some areas. Notably, a dry run does not
1151 send the actual data for file transfers, so bf(--progress) has no effect,
1152 the "bytes sent", "bytes received", "literal data", and "matched data"
1153 statistics are too small, and the "speedup" value is equivalent to a run
1154 where no file transfers are needed.
1156 dit(bf(-W, --whole-file)) With this option rsync's delta-transfer algorithm
1157 is not used and the whole file is sent as-is instead. The transfer may be
1158 faster if this option is used when the bandwidth between the source and
1159 destination machines is higher than the bandwidth to disk (especially when the
1160 "disk" is actually a networked filesystem). This is the default when both
1161 the source and destination are specified as local paths, but only if no
1162 batch-writing option is in effect.
1164 dit(bf(-x, --one-file-system)) This tells rsync to avoid crossing a
1165 filesystem boundary when recursing. This does not limit the user's ability
1166 to specify items to copy from multiple filesystems, just rsync's recursion
1167 through the hierarchy of each directory that the user specified, and also
1168 the analogous recursion on the receiving side during deletion. Also keep
1169 in mind that rsync treats a "bind" mount to the same device as being on the
1172 If this option is repeated, rsync omits all mount-point directories from
1173 the copy. Otherwise, it includes an empty directory at each mount-point it
1174 encounters (using the attributes of the mounted directory because those of
1175 the underlying mount-point directory are inaccessible).
1177 If rsync has been told to collapse symlinks (via bf(--copy-links) or
1178 bf(--copy-unsafe-links)), a symlink to a directory on another device is
1179 treated like a mount-point. Symlinks to non-directories are unaffected
1182 dit(bf(--existing, --ignore-non-existing)) This tells rsync to skip
1183 creating files (including directories) that do not exist
1184 yet on the destination. If this option is
1185 combined with the bf(--ignore-existing) option, no files will be updated
1186 (which can be useful if all you want to do is delete extraneous files).
1188 This option is a transfer rule, not an exclude, so it doesn't affect the
1189 data that goes into the file-lists, and thus it doesn't affect deletions.
1190 It just limits the files that the receiver requests to be transferred.
1192 dit(bf(--ignore-existing)) This tells rsync to skip updating files that
1193 already exist on the destination (this does em(not) ignore existing
1194 directories, or nothing would get done). See also bf(--existing).
1196 This option is a transfer rule, not an exclude, so it doesn't affect the
1197 data that goes into the file-lists, and thus it doesn't affect deletions.
1198 It just limits the files that the receiver requests to be transferred.
1200 This option can be useful for those doing backups using the bf(--link-dest)
1201 option when they need to continue a backup run that got interrupted. Since
1202 a bf(--link-dest) run is copied into a new directory hierarchy (when it is
1203 used properly), using bf(--ignore existing) will ensure that the
1204 already-handled files don't get tweaked (which avoids a change in
1205 permissions on the hard-linked files). This does mean that this option
1206 is only looking at the existing files in the destination hierarchy itself.
1208 dit(bf(--remove-source-files)) This tells rsync to remove from the sending
1209 side the files (meaning non-directories) that are a part of the transfer
1210 and have been successfully duplicated on the receiving side.
1212 dit(bf(--delete)) This tells rsync to delete extraneous files from the
1213 receiving side (ones that aren't on the sending side), but only for the
1214 directories that are being synchronized. You must have asked rsync to
1215 send the whole directory (e.g. "dir" or "dir/") without using a wildcard
1216 for the directory's contents (e.g. "dir/*") since the wildcard is expanded
1217 by the shell and rsync thus gets a request to transfer individual files, not
1218 the files' parent directory. Files that are excluded from the transfer are
1219 also excluded from being deleted unless you use the bf(--delete-excluded)
1220 option or mark the rules as only matching on the sending side (see the
1221 include/exclude modifiers in the FILTER RULES section).
1223 Prior to rsync 2.6.7, this option would have no effect unless bf(--recursive)
1224 was enabled. Beginning with 2.6.7, deletions will also occur when bf(--dirs)
1225 (bf(-d)) is enabled, but only for directories whose contents are being copied.
1227 This option can be dangerous if used incorrectly! It is a very good idea to
1228 first try a run using the bf(--dry-run) option (bf(-n)) to see what files are
1229 going to be deleted.
1231 If the sending side detects any I/O errors, then the deletion of any
1232 files at the destination will be automatically disabled. This is to
1233 prevent temporary filesystem failures (such as NFS errors) on the
1234 sending side causing a massive deletion of files on the
1235 destination. You can override this with the bf(--ignore-errors) option.
1237 The bf(--delete) option may be combined with one of the --delete-WHEN options
1238 without conflict, as well as bf(--delete-excluded). However, if none of the
1239 --delete-WHEN options are specified, rsync will choose the
1240 bf(--delete-during) algorithm when talking to rsync 3.0.0 or newer, and
1241 the bf(--delete-before) algorithm when talking to an older rsync. See also
1242 bf(--delete-delay) and bf(--delete-after).
1244 dit(bf(--delete-before)) Request that the file-deletions on the receiving
1245 side be done before the transfer starts.
1246 See bf(--delete) (which is implied) for more details on file-deletion.
1248 Deleting before the transfer is helpful if the filesystem is tight for space
1249 and removing extraneous files would help to make the transfer possible.
1250 However, it does introduce a delay before the start of the transfer,
1251 and this delay might cause the transfer to timeout (if bf(--timeout) was
1252 specified). It also forces rsync to use the old, non-incremental recursion
1253 algorithm that requires rsync to scan all the files in the transfer into
1254 memory at once (see bf(--recursive)).
1256 dit(bf(--delete-during, --del)) Request that the file-deletions on the
1257 receiving side be done incrementally as the transfer happens. The
1258 per-directory delete scan is done right before each directory is checked
1259 for updates, so it behaves like a more efficient bf(--delete-before),
1260 including doing the deletions prior to any per-directory filter files
1261 being updated. This option was first added in rsync version 2.6.4.
1262 See bf(--delete) (which is implied) for more details on file-deletion.
1264 dit(bf(--delete-delay)) Request that the file-deletions on the receiving
1265 side be computed during the transfer (like bf(--delete-during)), and then
1266 removed after the transfer completes. This is useful when combined with
1267 bf(--delay-updates) and/or bf(--fuzzy), and is more efficient than using
1268 bf(--delete-after) (but can behave differently, since bf(--delete-after)
1269 computes the deletions in a separate pass after all updates are done).
1270 If the number of removed files overflows an internal buffer, a
1271 temporary file will be created on the receiving side to hold the names (it
1272 is removed while open, so you shouldn't see it during the transfer). If
1273 the creation of the temporary file fails, rsync will try to fall back to
1274 using bf(--delete-after) (which it cannot do if bf(--recursive) is doing an
1276 See bf(--delete) (which is implied) for more details on file-deletion.
1278 dit(bf(--delete-after)) Request that the file-deletions on the receiving
1279 side be done after the transfer has completed. This is useful if you
1280 are sending new per-directory merge files as a part of the transfer and
1281 you want their exclusions to take effect for the delete phase of the
1282 current transfer. It also forces rsync to use the old, non-incremental
1283 recursion algorithm that requires rsync to scan all the files in the
1284 transfer into memory at once (see bf(--recursive)).
1285 See bf(--delete) (which is implied) for more details on file-deletion.
1287 dit(bf(--delete-excluded)) In addition to deleting the files on the
1288 receiving side that are not on the sending side, this tells rsync to also
1289 delete any files on the receiving side that are excluded (see bf(--exclude)).
1290 See the FILTER RULES section for a way to make individual exclusions behave
1291 this way on the receiver, and for a way to protect files from
1292 bf(--delete-excluded).
1293 See bf(--delete) (which is implied) for more details on file-deletion.
1295 dit(bf(--ignore-missing-args)) When rsync is first processing the explicitly
1296 requested source files (e.g. command-line arguments or bf(--files-from)
1297 entries), it is normally an error if the file cannot be found. This option
1298 suppresses that error, and does not try to transfer the file. This does not
1299 affect subsequent vanished-file errors if a file was initially found to be
1300 present and later is no longer there.
1302 dit(bf(--delete-missing-args)) This option takes the behavior of (the implied)
1303 bf(--ignore-missing-args) option a step farther: each missing arg will become
1304 a deletion request of the corresponding destination file on the receiving side
1305 (should it exist). If the destination file is a non-empty directory, it will
1306 only be successfully deleted if --force or --delete are in effect. Other than
1307 that, this option is independent of any other type of delete processing.
1309 The missing source files are represented by special file-list entries which
1310 display as a "*missing" entry in the bf(--list-only) output.
1312 dit(bf(--ignore-errors)) Tells bf(--delete) to go ahead and delete files
1313 even when there are I/O errors.
1315 dit(bf(--force)) This option tells rsync to delete a non-empty directory
1316 when it is to be replaced by a non-directory. This is only relevant if
1317 deletions are not active (see bf(--delete) for details).
1319 Note for older rsync versions: bf(--force) used to still be required when
1320 using bf(--delete-after), and it used to be non-functional unless the
1321 bf(--recursive) option was also enabled.
1323 dit(bf(--max-delete=NUM)) This tells rsync not to delete more than NUM
1324 files or directories. If that limit is exceeded, a warning is output
1325 and rsync exits with an error code of 25 (new for 3.0.0).
1327 Also new for version 3.0.0, you may specify bf(--max-delete=0) to be warned
1328 about any extraneous files in the destination without removing any of them.
1329 Older clients interpreted this as "unlimited", so if you don't know what
1330 version the client is, you can use the less obvious bf(--max-delete=-1) as
1331 a backward-compatible way to specify that no deletions be allowed (though
1332 older versions didn't warn when the limit was exceeded).
1334 dit(bf(--max-size=SIZE)) This tells rsync to avoid transferring any
1335 file that is larger than the specified SIZE. The SIZE value can be
1336 suffixed with a string to indicate a size multiplier, and
1337 may be a fractional value (e.g. "bf(--max-size=1.5m)").
1339 This option is a transfer rule, not an exclude, so it doesn't affect the
1340 data that goes into the file-lists, and thus it doesn't affect deletions.
1341 It just limits the files that the receiver requests to be transferred.
1343 The suffixes are as follows: "K" (or "KiB") is a kibibyte (1024),
1344 "M" (or "MiB") is a mebibyte (1024*1024), and "G" (or "GiB") is a
1345 gibibyte (1024*1024*1024).
1346 If you want the multiplier to be 1000 instead of 1024, use "KB",
1347 "MB", or "GB". (Note: lower-case is also accepted for all values.)
1348 Finally, if the suffix ends in either "+1" or "-1", the value will
1349 be offset by one byte in the indicated direction.
1351 Examples: --max-size=1.5mb-1 is 1499999 bytes, and --max-size=2g+1 is
1354 dit(bf(--min-size=SIZE)) This tells rsync to avoid transferring any
1355 file that is smaller than the specified SIZE, which can help in not
1356 transferring small, junk files.
1357 See the bf(--max-size) option for a description of SIZE and other information.
1359 dit(bf(-B, --block-size=BLOCKSIZE)) This forces the block size used in
1360 rsync's delta-transfer algorithm to a fixed value. It is normally selected based on
1361 the size of each file being updated. See the technical report for details.
1363 dit(bf(-e, --rsh=COMMAND)) This option allows you to choose an alternative
1364 remote shell program to use for communication between the local and
1365 remote copies of rsync. Typically, rsync is configured to use ssh by
1366 default, but you may prefer to use rsh on a local network.
1368 If this option is used with bf([user@]host::module/path), then the
1369 remote shell em(COMMAND) will be used to run an rsync daemon on the
1370 remote host, and all data will be transmitted through that remote
1371 shell connection, rather than through a direct socket connection to a
1372 running rsync daemon on the remote host. See the section "USING
1373 RSYNC-DAEMON FEATURES VIA A REMOTE-SHELL CONNECTION" above.
1375 Command-line arguments are permitted in COMMAND provided that COMMAND is
1376 presented to rsync as a single argument. You must use spaces (not tabs
1377 or other whitespace) to separate the command and args from each other,
1378 and you can use single- and/or double-quotes to preserve spaces in an
1379 argument (but not backslashes). Note that doubling a single-quote
1380 inside a single-quoted string gives you a single-quote; likewise for
1381 double-quotes (though you need to pay attention to which quotes your
1382 shell is parsing and which quotes rsync is parsing). Some examples:
1385 tt( -e 'ssh -p 2234')nl()
1386 tt( -e 'ssh -o "ProxyCommand nohup ssh firewall nc -w1 %h %p"')nl()
1389 (Note that ssh users can alternately customize site-specific connect
1390 options in their .ssh/config file.)
1392 You can also choose the remote shell program using the RSYNC_RSH
1393 environment variable, which accepts the same range of values as bf(-e).
1395 See also the bf(--blocking-io) option which is affected by this option.
1397 dit(bf(--rsync-path=PROGRAM)) Use this to specify what program is to be run
1398 on the remote machine to start-up rsync. Often used when rsync is not in
1399 the default remote-shell's path (e.g. --rsync-path=/usr/local/bin/rsync).
1400 Note that PROGRAM is run with the help of a shell, so it can be any
1401 program, script, or command sequence you'd care to run, so long as it does
1402 not corrupt the standard-in & standard-out that rsync is using to
1405 One tricky example is to set a different default directory on the remote
1406 machine for use with the bf(--relative) option. For instance:
1408 quote(tt( rsync -avR --rsync-path="cd /a/b && rsync" host:c/d /e/))
1410 dit(bf(-M, --remote-option=OPTION)) This option is used for more advanced
1411 situations where you want certain effects to be limited to one side of the
1412 transfer only. For instance, if you want to pass bf(--log-file=FILE) and
1413 bf(--fake-super) to the remote system, specify it like this:
1415 quote(tt( rsync -av -M --log-file=foo -M--fake-super src/ dest/))
1417 If you want to have an option affect only the local side of a transfer when
1418 it normally affects both sides, send its negation to the remote side. Like
1421 quote(tt( rsync -av -x -M--no-x src/ dest/))
1423 Be cautious using this, as it is possible to toggle an option that will cause
1424 rsync to have a different idea about what data to expect next over the socket,
1425 and that will make it fail in a cryptic fashion.
1427 Note that it is best to use a separate bf(--remote-option) for each option you
1428 want to pass. This makes your useage compatible with the bf(--protect-args)
1429 option. If that option is off, any spaces in your remote options will be split
1430 by the remote shell unless you take steps to protect them.
1432 When performing a local transfer, the "local" side is the sender and the
1433 "remote" side is the receiver.
1435 Note some versions of the popt option-parsing library have a bug in them that
1436 prevents you from using an adjacent arg with an equal in it next to a short
1437 option letter (e.g. tt(-M--log-file=/tmp/foo). If this bug affects your
1438 version of popt, you can use the version of popt that is included with rsync.
1440 dit(bf(-C, --cvs-exclude)) This is a useful shorthand for excluding a
1441 broad range of files that you often don't want to transfer between
1442 systems. It uses a similar algorithm to CVS to determine if
1443 a file should be ignored.
1445 The exclude list is initialized to exclude the following items (these
1446 initial items are marked as perishable -- see the FILTER RULES section):
1448 quote(quote(tt(RCS SCCS CVS CVS.adm RCSLOG cvslog.* tags TAGS .make.state
1449 .nse_depinfo *~ #* .#* ,* _$* *$ *.old *.bak *.BAK *.orig *.rej .del-*
1450 *.a *.olb *.o *.obj *.so *.exe *.Z *.elc *.ln core .svn/ .git/ .bzr/)))
1452 then, files listed in a $HOME/.cvsignore are added to the list and any
1453 files listed in the CVSIGNORE environment variable (all cvsignore names
1454 are delimited by whitespace).
1456 Finally, any file is ignored if it is in the same directory as a
1457 .cvsignore file and matches one of the patterns listed therein. Unlike
1458 rsync's filter/exclude files, these patterns are split on whitespace.
1459 See the bf(cvs)(1) manual for more information.
1461 If you're combining bf(-C) with your own bf(--filter) rules, you should
1462 note that these CVS excludes are appended at the end of your own rules,
1463 regardless of where the bf(-C) was placed on the command-line. This makes them
1464 a lower priority than any rules you specified explicitly. If you want to
1465 control where these CVS excludes get inserted into your filter rules, you
1466 should omit the bf(-C) as a command-line option and use a combination of
1467 bf(--filter=:C) and bf(--filter=-C) (either on your command-line or by
1468 putting the ":C" and "-C" rules into a filter file with your other rules).
1469 The first option turns on the per-directory scanning for the .cvsignore
1470 file. The second option does a one-time import of the CVS excludes
1473 dit(bf(-f, --filter=RULE)) This option allows you to add rules to selectively
1474 exclude certain files from the list of files to be transferred. This is
1475 most useful in combination with a recursive transfer.
1477 You may use as many bf(--filter) options on the command line as you like
1478 to build up the list of files to exclude. If the filter contains whitespace,
1479 be sure to quote it so that the shell gives the rule to rsync as a single
1480 argument. The text below also mentions that you can use an underscore to
1481 replace the space that separates a rule from its arg.
1483 See the FILTER RULES section for detailed information on this option.
1485 dit(bf(-F)) The bf(-F) option is a shorthand for adding two bf(--filter) rules to
1486 your command. The first time it is used is a shorthand for this rule:
1488 quote(tt( --filter='dir-merge /.rsync-filter'))
1490 This tells rsync to look for per-directory .rsync-filter files that have
1491 been sprinkled through the hierarchy and use their rules to filter the
1492 files in the transfer. If bf(-F) is repeated, it is a shorthand for this
1495 quote(tt( --filter='exclude .rsync-filter'))
1497 This filters out the .rsync-filter files themselves from the transfer.
1499 See the FILTER RULES section for detailed information on how these options
1502 dit(bf(--exclude=PATTERN)) This option is a simplified form of the
1503 bf(--filter) option that defaults to an exclude rule and does not allow
1504 the full rule-parsing syntax of normal filter rules.
1506 See the FILTER RULES section for detailed information on this option.
1508 dit(bf(--exclude-from=FILE)) This option is related to the bf(--exclude)
1509 option, but it specifies a FILE that contains exclude patterns (one per line).
1510 Blank lines in the file and lines starting with ';' or '#' are ignored.
1511 If em(FILE) is bf(-), the list will be read from standard input.
1513 dit(bf(--include=PATTERN)) This option is a simplified form of the
1514 bf(--filter) option that defaults to an include rule and does not allow
1515 the full rule-parsing syntax of normal filter rules.
1517 See the FILTER RULES section for detailed information on this option.
1519 dit(bf(--include-from=FILE)) This option is related to the bf(--include)
1520 option, but it specifies a FILE that contains include patterns (one per line).
1521 Blank lines in the file and lines starting with ';' or '#' are ignored.
1522 If em(FILE) is bf(-), the list will be read from standard input.
1524 dit(bf(--files-from=FILE)) Using this option allows you to specify the
1525 exact list of files to transfer (as read from the specified FILE or bf(-)
1526 for standard input). It also tweaks the default behavior of rsync to make
1527 transferring just the specified files and directories easier:
1530 it() The bf(--relative) (bf(-R)) option is implied, which preserves the path
1531 information that is specified for each item in the file (use
1532 bf(--no-relative) or bf(--no-R) if you want to turn that off).
1533 it() The bf(--dirs) (bf(-d)) option is implied, which will create directories
1534 specified in the list on the destination rather than noisily skipping
1535 them (use bf(--no-dirs) or bf(--no-d) if you want to turn that off).
1536 it() The bf(--archive) (bf(-a)) option's behavior does not imply bf(--recursive)
1537 (bf(-r)), so specify it explicitly, if you want it.
1538 it() These side-effects change the default state of rsync, so the position
1539 of the bf(--files-from) option on the command-line has no bearing on how
1540 other options are parsed (e.g. bf(-a) works the same before or after
1541 bf(--files-from), as does bf(--no-R) and all other options).
1544 The filenames that are read from the FILE are all relative to the
1545 source dir -- any leading slashes are removed and no ".." references are
1546 allowed to go higher than the source dir. For example, take this
1549 quote(tt( rsync -a --files-from=/tmp/foo /usr remote:/backup))
1551 If /tmp/foo contains the string "bin" (or even "/bin"), the /usr/bin
1552 directory will be created as /backup/bin on the remote host. If it
1553 contains "bin/" (note the trailing slash), the immediate contents of
1554 the directory would also be sent (without needing to be explicitly
1555 mentioned in the file -- this began in version 2.6.4). In both cases,
1556 if the bf(-r) option was enabled, that dir's entire hierarchy would
1557 also be transferred (keep in mind that bf(-r) needs to be specified
1558 explicitly with bf(--files-from), since it is not implied by bf(-a)).
1560 that the effect of the (enabled by default) bf(--relative) option is to
1561 duplicate only the path info that is read from the file -- it does not
1562 force the duplication of the source-spec path (/usr in this case).
1564 In addition, the bf(--files-from) file can be read from the remote host
1565 instead of the local host if you specify a "host:" in front of the file
1566 (the host must match one end of the transfer). As a short-cut, you can
1567 specify just a prefix of ":" to mean "use the remote end of the
1568 transfer". For example:
1570 quote(tt( rsync -a --files-from=:/path/file-list src:/ /tmp/copy))
1572 This would copy all the files specified in the /path/file-list file that
1573 was located on the remote "src" host.
1575 dit(bf(-0, --from0)) This tells rsync that the rules/filenames it reads from a
1576 file are terminated by a null ('\0') character, not a NL, CR, or CR+LF.
1577 This affects bf(--exclude-from), bf(--include-from), bf(--files-from), and any
1578 merged files specified in a bf(--filter) rule.
1579 It does not affect bf(--cvs-exclude) (since all names read from a .cvsignore
1580 file are split on whitespace).
1582 If the bf(--iconv) and bf(--protect-args) options are specified and the
1583 bf(--files-from) filenames are being sent from one host to another, the
1584 filenames will be translated from the sending host's charset to the
1585 receiving host's charset.
1587 dit(bf(-s, --protect-args)) This option sends all filenames and some options to
1588 the remote rsync without allowing the remote shell to interpret them. This
1589 means that spaces are not split in names, and any non-wildcard special
1590 characters are not translated (such as ~, $, ;, &, etc.). Wildcards are
1591 expanded on the remote host by rsync (instead of the shell doing it).
1593 If you use this option with bf(--iconv), the args will also be translated
1594 from the local to the remote character-set. The translation happens before
1595 wild-cards are expanded. See also the bf(--files-from) option.
1597 dit(bf(-T, --temp-dir=DIR)) This option instructs rsync to use DIR as a
1598 scratch directory when creating temporary copies of the files transferred
1599 on the receiving side. The default behavior is to create each temporary
1600 file in the same directory as the associated destination file.
1602 This option is most often used when the receiving disk partition does not
1603 have enough free space to hold a copy of the largest file in the transfer.
1604 In this case (i.e. when the scratch directory is on a different disk
1605 partition), rsync will not be able to rename each received temporary file
1606 over the top of the associated destination file, but instead must copy it
1607 into place. Rsync does this by copying the file over the top of the
1608 destination file, which means that the destination file will contain
1609 truncated data during this copy. If this were not done this way (even if
1610 the destination file were first removed, the data locally copied to a
1611 temporary file in the destination directory, and then renamed into place)
1612 it would be possible for the old file to continue taking up disk space (if
1613 someone had it open), and thus there might not be enough room to fit the
1614 new version on the disk at the same time.
1616 If you are using this option for reasons other than a shortage of disk
1617 space, you may wish to combine it with the bf(--delay-updates) option,
1618 which will ensure that all copied files get put into subdirectories in the
1619 destination hierarchy, awaiting the end of the transfer. If you don't
1620 have enough room to duplicate all the arriving files on the destination
1621 partition, another way to tell rsync that you aren't overly concerned
1622 about disk space is to use the bf(--partial-dir) option with a relative
1623 path; because this tells rsync that it is OK to stash off a copy of a
1624 single file in a subdir in the destination hierarchy, rsync will use the
1625 partial-dir as a staging area to bring over the copied file, and then
1626 rename it into place from there. (Specifying a bf(--partial-dir) with
1627 an absolute path does not have this side-effect.)
1629 dit(bf(-y, --fuzzy)) This option tells rsync that it should look for a
1630 basis file for any destination file that is missing. The current algorithm
1631 looks in the same directory as the destination file for either a file that
1632 has an identical size and modified-time, or a similarly-named file. If
1633 found, rsync uses the fuzzy basis file to try to speed up the transfer.
1635 Note that the use of the bf(--delete) option might get rid of any potential
1636 fuzzy-match files, so either use bf(--delete-after) or specify some
1637 filename exclusions if you need to prevent this.
1639 dit(bf(--compare-dest=DIR)) This option instructs rsync to use em(DIR) on
1640 the destination machine as an additional hierarchy to compare destination
1641 files against doing transfers (if the files are missing in the destination
1642 directory). If a file is found in em(DIR) that is identical to the
1643 sender's file, the file will NOT be transferred to the destination
1644 directory. This is useful for creating a sparse backup of just files that
1645 have changed from an earlier backup.
1647 Beginning in version 2.6.4, multiple bf(--compare-dest) directories may be
1648 provided, which will cause rsync to search the list in the order specified
1650 If a match is found that differs only in attributes, a local copy is made
1651 and the attributes updated.
1652 If a match is not found, a basis file from one of the em(DIR)s will be
1653 selected to try to speed up the transfer.
1655 If em(DIR) is a relative path, it is relative to the destination directory.
1656 See also bf(--copy-dest) and bf(--link-dest).
1658 dit(bf(--copy-dest=DIR)) This option behaves like bf(--compare-dest), but
1659 rsync will also copy unchanged files found in em(DIR) to the destination
1660 directory using a local copy.
1661 This is useful for doing transfers to a new destination while leaving
1662 existing files intact, and then doing a flash-cutover when all files have
1663 been successfully transferred.
1665 Multiple bf(--copy-dest) directories may be provided, which will cause
1666 rsync to search the list in the order specified for an unchanged file.
1667 If a match is not found, a basis file from one of the em(DIR)s will be
1668 selected to try to speed up the transfer.
1670 If em(DIR) is a relative path, it is relative to the destination directory.
1671 See also bf(--compare-dest) and bf(--link-dest).
1673 dit(bf(--link-dest=DIR)) This option behaves like bf(--copy-dest), but
1674 unchanged files are hard linked from em(DIR) to the destination directory.
1675 The files must be identical in all preserved attributes (e.g. permissions,
1676 possibly ownership) in order for the files to be linked together.
1679 quote(tt( rsync -av --link-dest=$PWD/prior_dir host:src_dir/ new_dir/))
1681 If file's aren't linking, double-check their attributes. Also check if some
1682 attributes are getting forced outside of rsync's control, such a mount option
1683 that squishes root to a single user, or mounts a removable drive with generic
1684 ownership (such as OS X's "Ignore ownership on this volume" option).
1686 Beginning in version 2.6.4, multiple bf(--link-dest) directories may be
1687 provided, which will cause rsync to search the list in the order specified
1689 If a match is found that differs only in attributes, a local copy is made
1690 and the attributes updated.
1691 If a match is not found, a basis file from one of the em(DIR)s will be
1692 selected to try to speed up the transfer.
1694 This option works best when copying into an empty destination hierarchy, as
1695 rsync treats existing files as definitive (so it never looks in the link-dest
1696 dirs when a destination file already exists), and as malleable (so it might
1697 change the attributes of a destination file, which affects all the hard-linked
1700 Note that if you combine this option with bf(--ignore-times), rsync will not
1701 link any files together because it only links identical files together as a
1702 substitute for transferring the file, never as an additional check after the
1705 If em(DIR) is a relative path, it is relative to the destination directory.
1706 See also bf(--compare-dest) and bf(--copy-dest).
1708 Note that rsync versions prior to 2.6.1 had a bug that could prevent
1709 bf(--link-dest) from working properly for a non-super-user when bf(-o) was
1710 specified (or implied by bf(-a)). You can work-around this bug by avoiding
1711 the bf(-o) option when sending to an old rsync.
1713 dit(bf(-z, --compress)) With this option, rsync compresses the file data
1714 as it is sent to the destination machine, which reduces the amount of data
1715 being transmitted -- something that is useful over a slow connection.
1717 Note that this option typically achieves better compression ratios than can
1718 be achieved by using a compressing remote shell or a compressing transport
1719 because it takes advantage of the implicit information in the matching data
1720 blocks that are not explicitly sent over the connection.
1722 See the bf(--skip-compress) option for the default list of file suffixes
1723 that will not be compressed.
1725 dit(bf(--compress-level=NUM)) Explicitly set the compression level to use
1726 (see bf(--compress)) instead of letting it default. If NUM is non-zero,
1727 the bf(--compress) option is implied.
1729 dit(bf(--skip-compress=LIST)) Override the list of file suffixes that will
1730 not be compressed. The bf(LIST) should be one or more file suffixes
1731 (without the dot) separated by slashes (/).
1733 You may specify an empty string to indicate that no file should be skipped.
1735 Simple character-class matching is supported: each must consist of a list
1736 of letters inside the square brackets (e.g. no special classes, such as
1737 "[:alpha:]", are supported, and '-' has no special meaning).
1739 The characters asterisk (*) and question-mark (?) have no special meaning.
1741 Here's an example that specifies 6 suffixes to skip (since 1 of the 5 rules
1742 matches 2 suffixes):
1744 verb( --skip-compress=gz/jpg/mp[34]/7z/bz2)
1746 The default list of suffixes that will not be compressed is this (in this
1773 This list will be replaced by your bf(--skip-compress) list in all but one
1774 situation: a copy from a daemon rsync will add your skipped suffixes to
1775 its list of non-compressing files (and its list may be configured to a
1778 dit(bf(--numeric-ids)) With this option rsync will transfer numeric group
1779 and user IDs rather than using user and group names and mapping them
1782 By default rsync will use the username and groupname to determine
1783 what ownership to give files. The special uid 0 and the special group
1784 0 are never mapped via user/group names even if the bf(--numeric-ids)
1785 option is not specified.
1787 If a user or group has no name on the source system or it has no match
1788 on the destination system, then the numeric ID
1789 from the source system is used instead. See also the comments on the
1790 "use chroot" setting in the rsyncd.conf manpage for information on how
1791 the chroot setting affects rsync's ability to look up the names of the
1792 users and groups and what you can do about it.
1794 dit(bf(--usermap=STRING, --groupmap=STRING)) These options allow you to
1795 specify users and groups that should be mapped to other values by the
1796 receiving side. The bf(STRING) is one or more bf(FROM):bf(TO) pairs of
1797 values separated by commas. Any matching bf(FROM) value from the sender is
1798 replaced with a bf(TO) value from the receiver. You may specify usernames
1799 or user IDs for the bf(FROM) and bf(TO) values, and the bf(FROM) value may
1800 also be a wild-card string, which will be matched against the sender's
1801 names (wild-cards do NOT match against ID numbers, though see below for
1802 why a '*' matches everything). You may instead specify a range of ID
1803 numbers via an inclusive range: LOW-HIGH. For example:
1805 verb( --usermap=0-99:nobody,wayne:admin,*:normal --groupmap=usr:1,1:usr)
1807 The first match in the list is the one that is used. You should specify
1808 all your user mappings using a single bf(--usermap) option, and/or all
1809 your group mappings using a single bf(--groupmap) option.
1811 Note that the sender's name for the 0 user and group are not transmitted
1812 to the receiver, so you should either match these values using a 0, or use
1813 the names in effect on the receiving side (typically "root"). All other
1814 bf(FROM) names match those in use on the sending side. All bf(TO) names
1815 match those in use on the receiving side.
1817 Any IDs that do not have a name on the sending side are treated as having an
1818 empty name for the purpose of matching. This allows them to be matched via
1819 a "*" or using an empty name. For instance:
1821 verb( --usermap=:nobody --groupmap=*:nobody)
1823 When the bf(--numeric-ids) option is used, the sender does not send any
1824 names, so all the IDs are treated as having an empty name. This means that
1825 you will need to specify numeric bf(FROM) values if you want to map these
1826 nameless IDs to different values.
1828 For the bf(--usermap) option to have any effect, the bf(-o) (bf(--owner))
1829 option must be used (or implied), and the receiver will need to be running
1830 as a super-user (see also the bf(--fake-super) option). For the bf(--groupmap)
1831 option to have any effect, the bf(-g) (bf(--groups)) option must be used
1832 (or implied), and the receiver will need to have permissions to set that
1835 dit(bf(--chown=USER:GROUP)) This option forces all files to be owned by USER
1836 with group GROUP. This is a simpler interface than using bf(--usermap) and
1837 bf(--groupmap) directly, but it is implemented using those options internally,
1838 so you cannot mix them. If either the USER or GROUP is empty, no mapping for
1839 the omitted user/group will occur. If GROUP is empty, the trailing colon may
1840 be omitted, but if USER is empty, a leading colon must be supplied.
1842 If you specify "--chown=foo:bar, this is exactly the same as specifying
1843 "--usermap=*:foo --groupmap=*:bar", only easier.
1845 dit(bf(--timeout=TIMEOUT)) This option allows you to set a maximum I/O
1846 timeout in seconds. If no data is transferred for the specified time
1847 then rsync will exit. The default is 0, which means no timeout.
1849 dit(bf(--contimeout)) This option allows you to set the amount of time
1850 that rsync will wait for its connection to an rsync daemon to succeed.
1851 If the timeout is reached, rsync exits with an error.
1853 dit(bf(--address)) By default rsync will bind to the wildcard address when
1854 connecting to an rsync daemon. The bf(--address) option allows you to
1855 specify a specific IP address (or hostname) to bind to. See also this
1856 option in the bf(--daemon) mode section.
1858 dit(bf(--port=PORT)) This specifies an alternate TCP port number to use
1859 rather than the default of 873. This is only needed if you are using the
1860 double-colon (::) syntax to connect with an rsync daemon (since the URL
1861 syntax has a way to specify the port as a part of the URL). See also this
1862 option in the bf(--daemon) mode section.
1864 dit(bf(--sockopts)) This option can provide endless fun for people
1865 who like to tune their systems to the utmost degree. You can set all
1866 sorts of socket options which may make transfers faster (or
1867 slower!). Read the man page for the code(setsockopt()) system call for
1868 details on some of the options you may be able to set. By default no
1869 special socket options are set. This only affects direct socket
1870 connections to a remote rsync daemon. This option also exists in the
1871 bf(--daemon) mode section.
1873 dit(bf(--blocking-io)) This tells rsync to use blocking I/O when launching
1874 a remote shell transport. If the remote shell is either rsh or remsh,
1875 rsync defaults to using
1876 blocking I/O, otherwise it defaults to using non-blocking I/O. (Note that
1877 ssh prefers non-blocking I/O.)
1879 dit(bf(-i, --itemize-changes)) Requests a simple itemized list of the
1880 changes that are being made to each file, including attribute changes.
1881 This is exactly the same as specifying bf(--out-format='%i %n%L').
1882 If you repeat the option, unchanged files will also be output, but only
1883 if the receiving rsync is at least version 2.6.7 (you can use bf(-vv)
1884 with older versions of rsync, but that also turns on the output of other
1887 The "%i" escape has a cryptic output that is 11 letters long. The general
1888 format is like the string bf(YXcstpoguax), where bf(Y) is replaced by the
1889 type of update being done, bf(X) is replaced by the file-type, and the
1890 other letters represent attributes that may be output if they are being
1893 The update types that replace the bf(Y) are as follows:
1896 it() A bf(<) means that a file is being transferred to the remote host
1898 it() A bf(>) means that a file is being transferred to the local host
1900 it() A bf(c) means that a local change/creation is occurring for the item
1901 (such as the creation of a directory or the changing of a symlink, etc.).
1902 it() A bf(h) means that the item is a hard link to another item (requires
1904 it() A bf(.) means that the item is not being updated (though it might
1905 have attributes that are being modified).
1906 it() A bf(*) means that the rest of the itemized-output area contains
1907 a message (e.g. "deleting").
1910 The file-types that replace the bf(X) are: bf(f) for a file, a bf(d) for a
1911 directory, an bf(L) for a symlink, a bf(D) for a device, and a bf(S) for a
1912 special file (e.g. named sockets and fifos).
1914 The other letters in the string above are the actual letters that
1915 will be output if the associated attribute for the item is being updated or
1916 a "." for no change. Three exceptions to this are: (1) a newly created
1917 item replaces each letter with a "+", (2) an identical item replaces the
1918 dots with spaces, and (3) an unknown attribute replaces each letter with
1919 a "?" (this can happen when talking to an older rsync).
1921 The attribute that is associated with each letter is as follows:
1924 it() A bf(c) means either that a regular file has a different checksum
1925 (requires bf(--checksum)) or that a symlink, device, or special file has
1927 Note that if you are sending files to an rsync prior to 3.0.1, this
1928 change flag will be present only for checksum-differing regular files.
1929 it() A bf(s) means the size of a regular file is different and will be updated
1930 by the file transfer.
1931 it() A bf(t) means the modification time is different and is being updated
1932 to the sender's value (requires bf(--times)). An alternate value of bf(T)
1933 means that the modification time will be set to the transfer time, which happens
1934 when a file/symlink/device is updated without bf(--times) and when a
1935 symlink is changed and the receiver can't set its time.
1936 (Note: when using an rsync 3.0.0 client, you might see the bf(s) flag combined
1937 with bf(t) instead of the proper bf(T) flag for this time-setting failure.)
1938 it() A bf(p) means the permissions are different and are being updated to
1939 the sender's value (requires bf(--perms)).
1940 it() An bf(o) means the owner is different and is being updated to the
1941 sender's value (requires bf(--owner) and super-user privileges).
1942 it() A bf(g) means the group is different and is being updated to the
1943 sender's value (requires bf(--group) and the authority to set the group).
1944 it() The bf(u) slot is reserved for future use.
1945 it() The bf(a) means that the ACL information changed.
1946 it() The bf(x) means that the extended attribute information changed.
1949 One other output is possible: when deleting files, the "%i" will output
1950 the string "*deleting" for each item that is being removed (assuming that
1951 you are talking to a recent enough rsync that it logs deletions instead of
1952 outputting them as a verbose message).
1954 dit(bf(--out-format=FORMAT)) This allows you to specify exactly what the
1955 rsync client outputs to the user on a per-update basis. The format is a
1956 text string containing embedded single-character escape sequences prefixed
1957 with a percent (%) character. A default format of "%n%L" is assumed if
1958 either bf(--info=name) or bf(-v) is specified (this tells you just the name
1959 of the file and, if the item is a link, where it points). For a full list
1960 of the possible escape characters, see the "log format" setting in the
1961 rsyncd.conf manpage.
1963 Specifying the bf(--out-format) option implies the bf(--info=name) option,
1964 which will mention each file, dir, etc. that gets updated in a significant
1965 way (a transferred file, a recreated symlink/device, or a touched
1966 directory). In addition, if the itemize-changes escape (%i) is included in
1967 the string (e.g. if the bf(--itemize-changes) option was used), the logging
1968 of names increases to mention any item that is changed in any way (as long
1969 as the receiving side is at least 2.6.4). See the bf(--itemize-changes)
1970 option for a description of the output of "%i".
1972 Rsync will output the out-format string prior to a file's transfer unless
1973 one of the transfer-statistic escapes is requested, in which case the
1974 logging is done at the end of the file's transfer. When this late logging
1975 is in effect and bf(--progress) is also specified, rsync will also output
1976 the name of the file being transferred prior to its progress information
1977 (followed, of course, by the out-format output).
1979 dit(bf(--log-file=FILE)) This option causes rsync to log what it is doing
1980 to a file. This is similar to the logging that a daemon does, but can be
1981 requested for the client side and/or the server side of a non-daemon
1982 transfer. If specified as a client option, transfer logging will be
1983 enabled with a default format of "%i %n%L". See the bf(--log-file-format)
1984 option if you wish to override this.
1986 Here's a example command that requests the remote side to log what is
1989 verb( rsync -av --remote-option=--log-file=/tmp/rlog src/ dest/)
1991 This is very useful if you need to debug why a connection is closing
1994 dit(bf(--log-file-format=FORMAT)) This allows you to specify exactly what
1995 per-update logging is put into the file specified by the bf(--log-file) option
1996 (which must also be specified for this option to have any effect). If you
1997 specify an empty string, updated files will not be mentioned in the log file.
1998 For a list of the possible escape characters, see the "log format" setting
1999 in the rsyncd.conf manpage.
2001 The default FORMAT used if bf(--log-file) is specified and this option is not
2004 dit(bf(--stats)) This tells rsync to print a verbose set of statistics
2005 on the file transfer, allowing you to tell how effective rsync's delta-transfer
2006 algorithm is for your data. This option is equivalent to bf(--info=stats2)
2007 if combined with 0 or 1 bf(-v) options, or bf(--info=stats3) if combined
2008 with 2 or more bf(-v) options.
2010 The current statistics are as follows: quote(itemization(
2011 it() bf(Number of files) is the count of all "files" (in the generic
2012 sense), which includes directories, symlinks, etc.
2013 it() bf(Number of files transferred) is the count of normal files that
2014 were updated via rsync's delta-transfer algorithm, which does not include created
2015 dirs, symlinks, etc.
2016 it() bf(Total file size) is the total sum of all file sizes in the transfer.
2017 This does not count any size for directories or special files, but does
2018 include the size of symlinks.
2019 it() bf(Total transferred file size) is the total sum of all files sizes
2020 for just the transferred files.
2021 it() bf(Literal data) is how much unmatched file-update data we had to
2022 send to the receiver for it to recreate the updated files.
2023 it() bf(Matched data) is how much data the receiver got locally when
2024 recreating the updated files.
2025 it() bf(File list size) is how big the file-list data was when the sender
2026 sent it to the receiver. This is smaller than the in-memory size for the
2027 file list due to some compressing of duplicated data when rsync sends the
2029 it() bf(File list generation time) is the number of seconds that the
2030 sender spent creating the file list. This requires a modern rsync on the
2031 sending side for this to be present.
2032 it() bf(File list transfer time) is the number of seconds that the sender
2033 spent sending the file list to the receiver.
2034 it() bf(Total bytes sent) is the count of all the bytes that rsync sent
2035 from the client side to the server side.
2036 it() bf(Total bytes received) is the count of all non-message bytes that
2037 rsync received by the client side from the server side. "Non-message"
2038 bytes means that we don't count the bytes for a verbose message that the
2039 server sent to us, which makes the stats more consistent.
2042 dit(bf(-8, --8-bit-output)) This tells rsync to leave all high-bit characters
2043 unescaped in the output instead of trying to test them to see if they're
2044 valid in the current locale and escaping the invalid ones. All control
2045 characters (but never tabs) are always escaped, regardless of this option's
2048 The escape idiom that started in 2.6.7 is to output a literal backslash (\)
2049 and a hash (#), followed by exactly 3 octal digits. For example, a newline
2050 would output as "\#012". A literal backslash that is in a filename is not
2051 escaped unless it is followed by a hash and 3 digits (0-9).
2053 dit(bf(-h, --human-readable)) Output numbers in a more human-readable format.
2054 There are 3 possible levels: (1) output numbers with a separator between each
2055 set of 3 digits (either a comma or a period, depending on if the decimal point
2056 is represented by a period or a comma); (2) output numbers in units of 1000
2057 (with a character suffix for larger units -- see below); (3) output numbers in
2060 The default is human-readable level 1. Each bf(-h) option increases the level
2061 by one. You can take the level down to 0 (to output numbers as pure digits) by
2062 specifing the bf(--no-human-readable) (bf(--no-h)) option.
2064 The unit letters that are appended in levels 2 and 3 are: K (kilo), M (mega),
2065 G (giga), or T (tera). For example, a 1234567-byte file would output as 1.23M
2066 in level-2 (assuming that a period is your local decimal point).
2068 Backward compatibility note: versions of rsync prior to 3.1.0 do not support
2069 human-readable level 1, and they default to level 0. Thus, specifying one or
2070 two bf(-h) options behaves the same in old and new versions as long as you
2071 didn't specify a bf(--no-h) option prior to one or more bf(-h) options.
2073 dit(bf(--partial)) By default, rsync will delete any partially
2074 transferred file if the transfer is interrupted. In some circumstances
2075 it is more desirable to keep partially transferred files. Using the
2076 bf(--partial) option tells rsync to keep the partial file which should
2077 make a subsequent transfer of the rest of the file much faster.
2079 dit(bf(--partial-dir=DIR)) A better way to keep partial files than the
2080 bf(--partial) option is to specify a em(DIR) that will be used to hold the
2081 partial data (instead of writing it out to the destination file).
2082 On the next transfer, rsync will use a file found in this
2083 dir as data to speed up the resumption of the transfer and then delete it
2084 after it has served its purpose.
2086 Note that if bf(--whole-file) is specified (or implied), any partial-dir
2087 file that is found for a file that is being updated will simply be removed
2089 rsync is sending files without using rsync's delta-transfer algorithm).
2091 Rsync will create the em(DIR) if it is missing (just the last dir -- not
2092 the whole path). This makes it easy to use a relative path (such as
2093 "bf(--partial-dir=.rsync-partial)") to have rsync create the
2094 partial-directory in the destination file's directory when needed, and then
2095 remove it again when the partial file is deleted.
2097 If the partial-dir value is not an absolute path, rsync will add an exclude
2098 rule at the end of all your existing excludes. This will prevent the
2099 sending of any partial-dir files that may exist on the sending side, and
2100 will also prevent the untimely deletion of partial-dir items on the
2101 receiving side. An example: the above bf(--partial-dir) option would add
2102 the equivalent of "bf(-f '-p .rsync-partial/')" at the end of any other
2105 If you are supplying your own exclude rules, you may need to add your own
2106 exclude/hide/protect rule for the partial-dir because (1) the auto-added
2107 rule may be ineffective at the end of your other rules, or (2) you may wish
2108 to override rsync's exclude choice. For instance, if you want to make
2109 rsync clean-up any left-over partial-dirs that may be lying around, you
2110 should specify bf(--delete-after) and add a "risk" filter rule, e.g.
2111 bf(-f 'R .rsync-partial/'). (Avoid using bf(--delete-before) or
2112 bf(--delete-during) unless you don't need rsync to use any of the
2113 left-over partial-dir data during the current run.)
2115 IMPORTANT: the bf(--partial-dir) should not be writable by other users or it
2116 is a security risk. E.g. AVOID "/tmp".
2118 You can also set the partial-dir value the RSYNC_PARTIAL_DIR environment
2119 variable. Setting this in the environment does not force bf(--partial) to be
2120 enabled, but rather it affects where partial files go when bf(--partial) is
2121 specified. For instance, instead of using bf(--partial-dir=.rsync-tmp)
2122 along with bf(--progress), you could set RSYNC_PARTIAL_DIR=.rsync-tmp in your
2123 environment and then just use the bf(-P) option to turn on the use of the
2124 .rsync-tmp dir for partial transfers. The only times that the bf(--partial)
2125 option does not look for this environment value are (1) when bf(--inplace) was
2126 specified (since bf(--inplace) conflicts with bf(--partial-dir)), and (2) when
2127 bf(--delay-updates) was specified (see below).
2129 For the purposes of the daemon-config's "refuse options" setting,
2130 bf(--partial-dir) does em(not) imply bf(--partial). This is so that a
2131 refusal of the bf(--partial) option can be used to disallow the overwriting
2132 of destination files with a partial transfer, while still allowing the
2133 safer idiom provided by bf(--partial-dir).
2135 dit(bf(--delay-updates)) This option puts the temporary file from each
2136 updated file into a holding directory until the end of the
2137 transfer, at which time all the files are renamed into place in rapid
2138 succession. This attempts to make the updating of the files a little more
2139 atomic. By default the files are placed into a directory named ".~tmp~" in
2140 each file's destination directory, but if you've specified the
2141 bf(--partial-dir) option, that directory will be used instead. See the
2142 comments in the bf(--partial-dir) section for a discussion of how this
2143 ".~tmp~" dir will be excluded from the transfer, and what you can do if
2144 you want rsync to cleanup old ".~tmp~" dirs that might be lying around.
2145 Conflicts with bf(--inplace) and bf(--append).
2147 This option uses more memory on the receiving side (one bit per file
2148 transferred) and also requires enough free disk space on the receiving
2149 side to hold an additional copy of all the updated files. Note also that
2150 you should not use an absolute path to bf(--partial-dir) unless (1)
2152 chance of any of the files in the transfer having the same name (since all
2153 the updated files will be put into a single directory if the path is
2155 and (2) there are no mount points in the hierarchy (since the
2156 delayed updates will fail if they can't be renamed into place).
2158 See also the "atomic-rsync" perl script in the "support" subdir for an
2159 update algorithm that is even more atomic (it uses bf(--link-dest) and a
2160 parallel hierarchy of files).
2162 dit(bf(-m, --prune-empty-dirs)) This option tells the receiving rsync to get
2163 rid of empty directories from the file-list, including nested directories
2164 that have no non-directory children. This is useful for avoiding the
2165 creation of a bunch of useless directories when the sending rsync is
2166 recursively scanning a hierarchy of files using include/exclude/filter
2169 Note that the use of transfer rules, such as the bf(--min-size) option, does
2170 not affect what goes into the file list, and thus does not leave directories
2171 empty, even if none of the files in a directory match the transfer rule.
2173 Because the file-list is actually being pruned, this option also affects
2174 what directories get deleted when a delete is active. However, keep in
2175 mind that excluded files and directories can prevent existing items from
2176 being deleted due to an exclude both hiding source files and protecting
2177 destination files. See the perishable filter-rule option for how to avoid
2180 You can prevent the pruning of certain empty directories from the file-list
2181 by using a global "protect" filter. For instance, this option would ensure
2182 that the directory "emptydir" was kept in the file-list:
2184 quote( --filter 'protect emptydir/')
2186 Here's an example that copies all .pdf files in a hierarchy, only creating
2187 the necessary destination directories to hold the .pdf files, and ensures
2188 that any superfluous files and directories in the destination are removed
2189 (note the hide filter of non-directories being used instead of an exclude):
2191 quote( rsync -avm --del --include='*.pdf' -f 'hide,! */' src/ dest)
2193 If you didn't want to remove superfluous destination files, the more
2194 time-honored options of "bf(--include='*/' --exclude='*')" would work fine
2195 in place of the hide-filter (if that is more natural to you).
2197 dit(bf(--progress)) This option tells rsync to print information
2198 showing the progress of the transfer. This gives a bored user
2200 With a modern rsync this is the same as specifying
2201 bf(--info=flist2,name,progress), but any user-supplied settings for those
2202 info flags takes precedence (e.g. "--info=flist0 --progress").
2204 While rsync is transferring a regular file, it updates a progress line that
2207 verb( 782448 63% 110.64kB/s 0:00:04)
2209 In this example, the receiver has reconstructed 782448 bytes or 63% of the
2210 sender's file, which is being reconstructed at a rate of 110.64 kilobytes
2211 per second, and the transfer will finish in 4 seconds if the current rate
2212 is maintained until the end.
2214 These statistics can be misleading if rsync's delta-transfer algorithm is
2215 in use. For example, if the sender's file consists of the basis file
2216 followed by additional data, the reported rate will probably drop
2217 dramatically when the receiver gets to the literal data, and the transfer
2218 will probably take much longer to finish than the receiver estimated as it
2219 was finishing the matched part of the file.
2221 When the file transfer finishes, rsync replaces the progress line with a
2222 summary line that looks like this:
2224 verb( 1,238,099 100% 146.38kB/s 0:00:08 (xfr#5, to-chk=169/396))
2226 In this example, the file was 1,238,099 bytes long in total, the average rate
2227 of transfer for the whole file was 146.38 kilobytes per second over the 8
2228 seconds that it took to complete, it was the 5th transfer of a regular file
2229 during the current rsync session, and there are 169 more files for the
2230 receiver to check (to see if they are up-to-date or not) remaining out of
2231 the 396 total files in the file-list.
2233 In an incremental recursion scan, rsync won't know the total number of files
2234 in the file-list until it reaches the ends of the scan, but since it starts to
2235 transfer files during the scan, it will display a line with the text "ir-chk"
2236 (for incremental recursion check) instead of "to-chk" until the point that it
2237 knows the full size of the list, at which point it will switch to using
2238 "to-chk". Thus, seeing "ir-chk" lets you know that the total count of files
2239 in the file list is still going to increase (and each time it does, the count
2240 of files left to check will increase by the number of the files added to the
2243 dit(bf(-P)) The bf(-P) option is equivalent to bf(--partial) bf(--progress). Its
2244 purpose is to make it much easier to specify these two options for a long
2245 transfer that may be interrupted.
2247 There is also a bf(--info=progress2) option that outputs statistics based
2248 on the whole transfer, rather than individual files. Use this flag without
2249 outputting a filename (e.g. avoid bf(-v) or specify bf(--info=name0) if you
2250 want to see how the transfer is doing without scrolling the screen with a
2251 lot of names. (You don't need to specify the bf(--progress) option in
2252 order to use bf(--info=progress2).)
2254 dit(bf(--password-file)) This option allows you to provide a password in a
2255 file for accessing an rsync daemon. The file must not be world readable.
2256 It should contain just the password as the first line of the file (all
2257 other lines are ignored).
2259 This option does not supply a password to a remote shell transport such as
2260 ssh; to learn how to do that, consult the remote shell's documentation.
2261 When accessing an rsync daemon using a remote shell as the transport, this
2262 option only comes into effect after the remote shell finishes its
2263 authentication (i.e. if you have also specified a password in the daemon's
2266 dit(bf(--list-only)) This option will cause the source files to be listed
2267 instead of transferred. This option is inferred if there is a single source
2268 arg and no destination specified, so its main uses are: (1) to turn a copy
2269 command that includes a
2270 destination arg into a file-listing command, or (2) to be able to specify
2271 more than one source arg (note: be sure to include the destination).
2272 Caution: keep in mind that a source arg with a wild-card is expanded by the
2273 shell into multiple args, so it is never safe to try to list such an arg
2274 without using this option. For example:
2276 verb( rsync -av --list-only foo* dest/)
2278 Compatibility note: when requesting a remote listing of files from an rsync
2279 that is version 2.6.3 or older, you may encounter an error if you ask for a
2280 non-recursive listing. This is because a file listing implies the bf(--dirs)
2281 option w/o bf(--recursive), and older rsyncs don't have that option. To
2282 avoid this problem, either specify the bf(--no-dirs) option (if you don't
2283 need to expand a directory's content), or turn on recursion and exclude
2284 the content of subdirectories: bf(-r --exclude='/*/*').
2286 dit(bf(--bwlimit=KBPS)) This option allows you to specify a maximum
2287 transfer rate in kilobytes per second. This option is most effective when
2288 using rsync with large files (several megabytes and up). Due to the nature
2289 of rsync transfers, blocks of data are sent, then if rsync determines the
2290 transfer was too fast, it will wait before sending the next data block. The
2291 result is an average transfer rate equaling the specified limit. A value
2292 of zero specifies no limit.
2294 dit(bf(--write-batch=FILE)) Record a file that can later be applied to
2295 another identical destination with bf(--read-batch). See the "BATCH MODE"
2296 section for details, and also the bf(--only-write-batch) option.
2298 dit(bf(--only-write-batch=FILE)) Works like bf(--write-batch), except that
2299 no updates are made on the destination system when creating the batch.
2300 This lets you transport the changes to the destination system via some
2301 other means and then apply the changes via bf(--read-batch).
2303 Note that you can feel free to write the batch directly to some portable
2304 media: if this media fills to capacity before the end of the transfer, you
2305 can just apply that partial transfer to the destination and repeat the
2306 whole process to get the rest of the changes (as long as you don't mind a
2307 partially updated destination system while the multi-update cycle is
2310 Also note that you only save bandwidth when pushing changes to a remote
2311 system because this allows the batched data to be diverted from the sender
2312 into the batch file without having to flow over the wire to the receiver
2313 (when pulling, the sender is remote, and thus can't write the batch).
2315 dit(bf(--read-batch=FILE)) Apply all of the changes stored in FILE, a
2316 file previously generated by bf(--write-batch).
2317 If em(FILE) is bf(-), the batch data will be read from standard input.
2318 See the "BATCH MODE" section for details.
2320 dit(bf(--protocol=NUM)) Force an older protocol version to be used. This
2321 is useful for creating a batch file that is compatible with an older
2322 version of rsync. For instance, if rsync 2.6.4 is being used with the
2323 bf(--write-batch) option, but rsync 2.6.3 is what will be used to run the
2324 bf(--read-batch) option, you should use "--protocol=28" when creating the
2325 batch file to force the older protocol version to be used in the batch
2326 file (assuming you can't upgrade the rsync on the reading system).
2328 dit(bf(--iconv=CONVERT_SPEC)) Rsync can convert filenames between character
2329 sets using this option. Using a CONVERT_SPEC of "." tells rsync to look up
2330 the default character-set via the locale setting. Alternately, you can
2331 fully specify what conversion to do by giving a local and a remote charset
2332 separated by a comma in the order bf(--iconv=LOCAL,REMOTE), e.g.
2333 bf(--iconv=utf8,iso88591). This order ensures that the option
2334 will stay the same whether you're pushing or pulling files.
2335 Finally, you can specify either bf(--no-iconv) or a CONVERT_SPEC of "-"
2336 to turn off any conversion.
2337 The default setting of this option is site-specific, and can also be
2338 affected via the RSYNC_ICONV environment variable.
2340 For a list of what charset names your local iconv library supports, you can
2343 If you specify the bf(--protect-args) option (bf(-s)), rsync will translate
2344 the filenames you specify on the command-line that are being sent to the
2345 remote host. See also the bf(--files-from) option.
2347 Note that rsync does not do any conversion of names in filter files
2348 (including include/exclude files). It is up to you to ensure that you're
2349 specifying matching rules that can match on both sides of the transfer.
2350 For instance, you can specify extra include/exclude rules if there are
2351 filename differences on the two sides that need to be accounted for.
2353 When you pass an bf(--iconv) option to an rsync daemon that allows it, the
2354 daemon uses the charset specified in its "charset" configuration parameter
2355 regardless of the remote charset you actually pass. Thus, you may feel free to
2356 specify just the local charset for a daemon transfer (e.g. bf(--iconv=utf8)).
2358 dit(bf(-4, --ipv4) or bf(-6, --ipv6)) Tells rsync to prefer IPv4/IPv6
2359 when creating sockets. This only affects sockets that rsync has direct
2360 control over, such as the outgoing socket when directly contacting an
2361 rsync daemon. See also these options in the bf(--daemon) mode section.
2363 If rsync was complied without support for IPv6, the bf(--ipv6) option
2364 will have no effect. The bf(--version) output will tell you if this
2367 dit(bf(--checksum-seed=NUM)) Set the checksum seed to the integer
2368 NUM. This 4 byte checksum seed is included in each block and file
2369 checksum calculation. By default the checksum seed is generated
2370 by the server and defaults to the current code(time()). This option
2371 is used to set a specific checksum seed, which is useful for
2372 applications that want repeatable block and file checksums, or
2373 in the case where the user wants a more random checksum seed.
2374 Setting NUM to 0 causes rsync to use the default of code(time())
2378 manpagesection(DAEMON OPTIONS)
2380 The options allowed when starting an rsync daemon are as follows:
2383 dit(bf(--daemon)) This tells rsync that it is to run as a daemon. The
2384 daemon you start running may be accessed using an rsync client using
2385 the bf(host::module) or bf(rsync://host/module/) syntax.
2387 If standard input is a socket then rsync will assume that it is being
2388 run via inetd, otherwise it will detach from the current terminal and
2389 become a background daemon. The daemon will read the config file
2390 (rsyncd.conf) on each connect made by a client and respond to
2391 requests accordingly. See the bf(rsyncd.conf)(5) man page for more
2394 dit(bf(--address)) By default rsync will bind to the wildcard address when
2395 run as a daemon with the bf(--daemon) option. The bf(--address) option
2396 allows you to specify a specific IP address (or hostname) to bind to. This
2397 makes virtual hosting possible in conjunction with the bf(--config) option.
2398 See also the "address" global option in the rsyncd.conf manpage.
2400 dit(bf(--bwlimit=KBPS)) This option allows you to specify a maximum
2401 transfer rate in kilobytes per second for the data the daemon sends.
2402 The client can still specify a smaller bf(--bwlimit) value, but their
2403 requested value will be rounded down if they try to exceed it. See the
2404 client version of this option (above) for some extra details.
2406 dit(bf(--config=FILE)) This specifies an alternate config file than
2407 the default. This is only relevant when bf(--daemon) is specified.
2408 The default is /etc/rsyncd.conf unless the daemon is running over
2409 a remote shell program and the remote user is not the super-user; in that case
2410 the default is rsyncd.conf in the current directory (typically $HOME).
2412 dit(bf(-M, --dparam=OVERRIDE)) This option can be used to set a daemon-config
2413 parameter when starting up rsync in daemon mode. It is equivalent to adding
2414 the parameter at the end of the global settings prior to the first module's
2415 definition. The parameter names can be specified without spaces, if you so
2416 desire. For instance:
2418 verb( rsync --daemon -M pidfile=/path/rsync.pid )
2420 dit(bf(--no-detach)) When running as a daemon, this option instructs
2421 rsync to not detach itself and become a background process. This
2422 option is required when running as a service on Cygwin, and may also
2423 be useful when rsync is supervised by a program such as
2424 bf(daemontools) or AIX's bf(System Resource Controller).
2425 bf(--no-detach) is also recommended when rsync is run under a
2426 debugger. This option has no effect if rsync is run from inetd or
2429 dit(bf(--port=PORT)) This specifies an alternate TCP port number for the
2430 daemon to listen on rather than the default of 873. See also the "port"
2431 global option in the rsyncd.conf manpage.
2433 dit(bf(--log-file=FILE)) This option tells the rsync daemon to use the
2434 given log-file name instead of using the "log file" setting in the config
2437 dit(bf(--log-file-format=FORMAT)) This option tells the rsync daemon to use the
2438 given FORMAT string instead of using the "log format" setting in the config
2439 file. It also enables "transfer logging" unless the string is empty, in which
2440 case transfer logging is turned off.
2442 dit(bf(--sockopts)) This overrides the bf(socket options) setting in the
2443 rsyncd.conf file and has the same syntax.
2445 dit(bf(-v, --verbose)) This option increases the amount of information the
2446 daemon logs during its startup phase. After the client connects, the
2447 daemon's verbosity level will be controlled by the options that the client
2448 used and the "max verbosity" setting in the module's config section.
2450 dit(bf(-4, --ipv4) or bf(-6, --ipv6)) Tells rsync to prefer IPv4/IPv6
2451 when creating the incoming sockets that the rsync daemon will use to
2452 listen for connections. One of these options may be required in older
2453 versions of Linux to work around an IPv6 bug in the kernel (if you see
2454 an "address already in use" error when nothing else is using the port,
2455 try specifying bf(--ipv6) or bf(--ipv4) when starting the daemon).
2457 If rsync was complied without support for IPv6, the bf(--ipv6) option
2458 will have no effect. The bf(--version) output will tell you if this
2461 dit(bf(-h, --help)) When specified after bf(--daemon), print a short help
2462 page describing the options available for starting an rsync daemon.
2465 manpagesection(FILTER RULES)
2467 The filter rules allow for flexible selection of which files to transfer
2468 (include) and which files to skip (exclude). The rules either directly
2469 specify include/exclude patterns or they specify a way to acquire more
2470 include/exclude patterns (e.g. to read them from a file).
2472 As the list of files/directories to transfer is built, rsync checks each
2473 name to be transferred against the list of include/exclude patterns in
2474 turn, and the first matching pattern is acted on: if it is an exclude
2475 pattern, then that file is skipped; if it is an include pattern then that
2476 filename is not skipped; if no matching pattern is found, then the
2477 filename is not skipped.
2479 Rsync builds an ordered list of filter rules as specified on the
2480 command-line. Filter rules have the following syntax:
2483 tt(RULE [PATTERN_OR_FILENAME])nl()
2484 tt(RULE,MODIFIERS [PATTERN_OR_FILENAME])nl()
2487 You have your choice of using either short or long RULE names, as described
2488 below. If you use a short-named rule, the ',' separating the RULE from the
2489 MODIFIERS is optional. The PATTERN or FILENAME that follows (when present)
2490 must come after either a single space or an underscore (_).
2491 Here are the available rule prefixes:
2494 bf(exclude, -) specifies an exclude pattern. nl()
2495 bf(include, +) specifies an include pattern. nl()
2496 bf(merge, .) specifies a merge-file to read for more rules. nl()
2497 bf(dir-merge, :) specifies a per-directory merge-file. nl()
2498 bf(hide, H) specifies a pattern for hiding files from the transfer. nl()
2499 bf(show, S) files that match the pattern are not hidden. nl()
2500 bf(protect, P) specifies a pattern for protecting files from deletion. nl()
2501 bf(risk, R) files that match the pattern are not protected. nl()
2502 bf(clear, !) clears the current include/exclude list (takes no arg) nl()
2505 When rules are being read from a file, empty lines are ignored, as are
2506 comment lines that start with a "#".
2508 Note that the bf(--include)/bf(--exclude) command-line options do not allow the
2509 full range of rule parsing as described above -- they only allow the
2510 specification of include/exclude patterns plus a "!" token to clear the
2511 list (and the normal comment parsing when rules are read from a file).
2513 does not begin with "- " (dash, space) or "+ " (plus, space), then the
2514 rule will be interpreted as if "+ " (for an include option) or "- " (for
2515 an exclude option) were prefixed to the string. A bf(--filter) option, on
2516 the other hand, must always contain either a short or long rule name at the
2519 Note also that the bf(--filter), bf(--include), and bf(--exclude) options take one
2520 rule/pattern each. To add multiple ones, you can repeat the options on
2521 the command-line, use the merge-file syntax of the bf(--filter) option, or
2522 the bf(--include-from)/bf(--exclude-from) options.
2524 manpagesection(INCLUDE/EXCLUDE PATTERN RULES)
2526 You can include and exclude files by specifying patterns using the "+",
2527 "-", etc. filter rules (as introduced in the FILTER RULES section above).
2528 The include/exclude rules each specify a pattern that is matched against
2529 the names of the files that are going to be transferred. These patterns
2530 can take several forms:
2533 it() if the pattern starts with a / then it is anchored to a
2534 particular spot in the hierarchy of files, otherwise it is matched
2535 against the end of the pathname. This is similar to a leading ^ in
2536 regular expressions.
2537 Thus "/foo" would match a name of "foo" at either the "root of the
2538 transfer" (for a global rule) or in the merge-file's directory (for a
2539 per-directory rule).
2540 An unqualified "foo" would match a name of "foo" anywhere in the
2541 tree because the algorithm is applied recursively from the
2542 top down; it behaves as if each path component gets a turn at being the
2543 end of the filename. Even the unanchored "sub/foo" would match at
2544 any point in the hierarchy where a "foo" was found within a directory
2545 named "sub". See the section on ANCHORING INCLUDE/EXCLUDE PATTERNS for
2546 a full discussion of how to specify a pattern that matches at the root
2548 it() if the pattern ends with a / then it will only match a
2549 directory, not a regular file, symlink, or device.
2550 it() rsync chooses between doing a simple string match and wildcard
2551 matching by checking if the pattern contains one of these three wildcard
2552 characters: '*', '?', and '[' .
2553 it() a '*' matches any path component, but it stops at slashes.
2554 it() use '**' to match anything, including slashes.
2555 it() a '?' matches any character except a slash (/).
2556 it() a '[' introduces a character class, such as [a-z] or [[:alpha:]].
2557 it() in a wildcard pattern, a backslash can be used to escape a wildcard
2558 character, but it is matched literally when no wildcards are present.
2559 it() if the pattern contains a / (not counting a trailing /) or a "**",
2560 then it is matched against the full pathname, including any leading
2561 directories. If the pattern doesn't contain a / or a "**", then it is
2562 matched only against the final component of the filename.
2563 (Remember that the algorithm is applied recursively so "full filename"
2564 can actually be any portion of a path from the starting directory on
2566 it() a trailing "dir_name/***" will match both the directory (as if
2567 "dir_name/" had been specified) and everything in the directory
2568 (as if "dir_name/**" had been specified). This behavior was added in
2572 Note that, when using the bf(--recursive) (bf(-r)) option (which is implied by
2573 bf(-a)), every subcomponent of every path is visited from the top down, so
2574 include/exclude patterns get applied recursively to each subcomponent's
2575 full name (e.g. to include "/foo/bar/baz" the subcomponents "/foo" and
2576 "/foo/bar" must not be excluded).
2577 The exclude patterns actually short-circuit the directory traversal stage
2578 when rsync finds the files to send. If a pattern excludes a particular
2579 parent directory, it can render a deeper include pattern ineffectual
2580 because rsync did not descend through that excluded section of the
2581 hierarchy. This is particularly important when using a trailing '*' rule.
2582 For instance, this won't work:
2585 tt(+ /some/path/this-file-will-not-be-found)nl()
2586 tt(+ /file-is-included)nl()
2590 This fails because the parent directory "some" is excluded by the '*'
2591 rule, so rsync never visits any of the files in the "some" or "some/path"
2592 directories. One solution is to ask for all directories in the hierarchy
2593 to be included by using a single rule: "+ */" (put it somewhere before the
2594 "- *" rule), and perhaps use the bf(--prune-empty-dirs) option. Another
2595 solution is to add specific include rules for all
2596 the parent dirs that need to be visited. For instance, this set of rules
2601 tt(+ /some/path/)nl()
2602 tt(+ /some/path/this-file-is-found)nl()
2603 tt(+ /file-also-included)nl()
2607 Here are some examples of exclude/include matching:
2610 it() "- *.o" would exclude all names matching *.o
2611 it() "- /foo" would exclude a file (or directory) named foo in the
2612 transfer-root directory
2613 it() "- foo/" would exclude any directory named foo
2614 it() "- /foo/*/bar" would exclude any file named bar which is at two
2615 levels below a directory named foo in the transfer-root directory
2616 it() "- /foo/**/bar" would exclude any file named bar two
2617 or more levels below a directory named foo in the transfer-root directory
2618 it() The combination of "+ */", "+ *.c", and "- *" would include all
2619 directories and C source files but nothing else (see also the
2620 bf(--prune-empty-dirs) option)
2621 it() The combination of "+ foo/", "+ foo/bar.c", and "- *" would include
2622 only the foo directory and foo/bar.c (the foo directory must be
2623 explicitly included or it would be excluded by the "*")
2626 The following modifiers are accepted after a "+" or "-":
2629 it() A bf(/) specifies that the include/exclude rule should be matched
2630 against the absolute pathname of the current item. For example,
2631 "-/ /etc/passwd" would exclude the passwd file any time the transfer
2632 was sending files from the "/etc" directory, and "-/ subdir/foo"
2633 would always exclude "foo" when it is in a dir named "subdir", even
2634 if "foo" is at the root of the current transfer.
2635 it() A bf(!) specifies that the include/exclude should take effect if
2636 the pattern fails to match. For instance, "-! */" would exclude all
2638 it() A bf(C) is used to indicate that all the global CVS-exclude rules
2639 should be inserted as excludes in place of the "-C". No arg should
2641 it() An bf(s) is used to indicate that the rule applies to the sending
2642 side. When a rule affects the sending side, it prevents files from
2643 being transferred. The default is for a rule to affect both sides
2644 unless bf(--delete-excluded) was specified, in which case default rules
2645 become sender-side only. See also the hide (H) and show (S) rules,
2646 which are an alternate way to specify sending-side includes/excludes.
2647 it() An bf(r) is used to indicate that the rule applies to the receiving
2648 side. When a rule affects the receiving side, it prevents files from
2649 being deleted. See the bf(s) modifier for more info. See also the
2650 protect (P) and risk (R) rules, which are an alternate way to
2651 specify receiver-side includes/excludes.
2652 it() A bf(p) indicates that a rule is perishable, meaning that it is
2653 ignored in directories that are being deleted. For instance, the bf(-C)
2654 option's default rules that exclude things like "CVS" and "*.o" are
2655 marked as perishable, and will not prevent a directory that was removed
2656 on the source from being deleted on the destination.
2659 manpagesection(MERGE-FILE FILTER RULES)
2661 You can merge whole files into your filter rules by specifying either a
2662 merge (.) or a dir-merge (:) filter rule (as introduced in the FILTER RULES
2665 There are two kinds of merged files -- single-instance ('.') and
2666 per-directory (':'). A single-instance merge file is read one time, and
2667 its rules are incorporated into the filter list in the place of the "."
2668 rule. For per-directory merge files, rsync will scan every directory that
2669 it traverses for the named file, merging its contents when the file exists
2670 into the current list of inherited rules. These per-directory rule files
2671 must be created on the sending side because it is the sending side that is
2672 being scanned for the available files to transfer. These rule files may
2673 also need to be transferred to the receiving side if you want them to
2674 affect what files don't get deleted (see PER-DIRECTORY RULES AND DELETE
2680 tt(merge /etc/rsync/default.rules)nl()
2681 tt(. /etc/rsync/default.rules)nl()
2682 tt(dir-merge .per-dir-filter)nl()
2683 tt(dir-merge,n- .non-inherited-per-dir-excludes)nl()
2684 tt(:n- .non-inherited-per-dir-excludes)nl()
2687 The following modifiers are accepted after a merge or dir-merge rule:
2690 it() A bf(-) specifies that the file should consist of only exclude
2691 patterns, with no other rule-parsing except for in-file comments.
2692 it() A bf(+) specifies that the file should consist of only include
2693 patterns, with no other rule-parsing except for in-file comments.
2694 it() A bf(C) is a way to specify that the file should be read in a
2695 CVS-compatible manner. This turns on 'n', 'w', and '-', but also
2696 allows the list-clearing token (!) to be specified. If no filename is
2697 provided, ".cvsignore" is assumed.
2698 it() A bf(e) will exclude the merge-file name from the transfer; e.g.
2699 "dir-merge,e .rules" is like "dir-merge .rules" and "- .rules".
2700 it() An bf(n) specifies that the rules are not inherited by subdirectories.
2701 it() A bf(w) specifies that the rules are word-split on whitespace instead
2702 of the normal line-splitting. This also turns off comments. Note: the
2703 space that separates the prefix from the rule is treated specially, so
2704 "- foo + bar" is parsed as two rules (assuming that prefix-parsing wasn't
2706 it() You may also specify any of the modifiers for the "+" or "-" rules
2707 (above) in order to have the rules that are read in from the file
2708 default to having that modifier set (except for the bf(!) modifier, which
2709 would not be useful). For instance, "merge,-/ .excl" would
2710 treat the contents of .excl as absolute-path excludes,
2711 while "dir-merge,s .filt" and ":sC" would each make all their
2712 per-directory rules apply only on the sending side. If the merge rule
2713 specifies sides to affect (via the bf(s) or bf(r) modifier or both),
2714 then the rules in the file must not specify sides (via a modifier or
2715 a rule prefix such as bf(hide)).
2718 Per-directory rules are inherited in all subdirectories of the directory
2719 where the merge-file was found unless the 'n' modifier was used. Each
2720 subdirectory's rules are prefixed to the inherited per-directory rules
2721 from its parents, which gives the newest rules a higher priority than the
2722 inherited rules. The entire set of dir-merge rules are grouped together in
2723 the spot where the merge-file was specified, so it is possible to override
2724 dir-merge rules via a rule that got specified earlier in the list of global
2725 rules. When the list-clearing rule ("!") is read from a per-directory
2726 file, it only clears the inherited rules for the current merge file.
2728 Another way to prevent a single rule from a dir-merge file from being inherited is to
2729 anchor it with a leading slash. Anchored rules in a per-directory
2730 merge-file are relative to the merge-file's directory, so a pattern "/foo"
2731 would only match the file "foo" in the directory where the dir-merge filter
2734 Here's an example filter file which you'd specify via bf(--filter=". file":)
2737 tt(merge /home/user/.global-filter)nl()
2739 tt(dir-merge .rules)nl()
2744 This will merge the contents of the /home/user/.global-filter file at the
2745 start of the list and also turns the ".rules" filename into a per-directory
2746 filter file. All rules read in prior to the start of the directory scan
2747 follow the global anchoring rules (i.e. a leading slash matches at the root
2750 If a per-directory merge-file is specified with a path that is a parent
2751 directory of the first transfer directory, rsync will scan all the parent
2752 dirs from that starting point to the transfer directory for the indicated
2753 per-directory file. For instance, here is a common filter (see bf(-F)):
2755 quote(tt(--filter=': /.rsync-filter'))
2757 That rule tells rsync to scan for the file .rsync-filter in all
2758 directories from the root down through the parent directory of the
2759 transfer prior to the start of the normal directory scan of the file in
2760 the directories that are sent as a part of the transfer. (Note: for an
2761 rsync daemon, the root is always the same as the module's "path".)
2763 Some examples of this pre-scanning for per-directory files:
2766 tt(rsync -avF /src/path/ /dest/dir)nl()
2767 tt(rsync -av --filter=': ../../.rsync-filter' /src/path/ /dest/dir)nl()
2768 tt(rsync -av --filter=': .rsync-filter' /src/path/ /dest/dir)nl()
2771 The first two commands above will look for ".rsync-filter" in "/" and
2772 "/src" before the normal scan begins looking for the file in "/src/path"
2773 and its subdirectories. The last command avoids the parent-dir scan
2774 and only looks for the ".rsync-filter" files in each directory that is
2775 a part of the transfer.
2777 If you want to include the contents of a ".cvsignore" in your patterns,
2778 you should use the rule ":C", which creates a dir-merge of the .cvsignore
2779 file, but parsed in a CVS-compatible manner. You can
2780 use this to affect where the bf(--cvs-exclude) (bf(-C)) option's inclusion of the
2781 per-directory .cvsignore file gets placed into your rules by putting the
2782 ":C" wherever you like in your filter rules. Without this, rsync would
2783 add the dir-merge rule for the .cvsignore file at the end of all your other
2784 rules (giving it a lower priority than your command-line rules). For
2788 tt(cat <<EOT | rsync -avC --filter='. -' a/ b)nl()
2793 tt(rsync -avC --include=foo.o -f :C --exclude='*.old' a/ b)nl()
2796 Both of the above rsync commands are identical. Each one will merge all
2797 the per-directory .cvsignore rules in the middle of the list rather than
2798 at the end. This allows their dir-specific rules to supersede the rules
2799 that follow the :C instead of being subservient to all your rules. To
2800 affect the other CVS exclude rules (i.e. the default list of exclusions,
2801 the contents of $HOME/.cvsignore, and the value of $CVSIGNORE) you should
2802 omit the bf(-C) command-line option and instead insert a "-C" rule into
2803 your filter rules; e.g. "bf(--filter=-C)".
2805 manpagesection(LIST-CLEARING FILTER RULE)
2807 You can clear the current include/exclude list by using the "!" filter
2808 rule (as introduced in the FILTER RULES section above). The "current"
2809 list is either the global list of rules (if the rule is encountered while
2810 parsing the filter options) or a set of per-directory rules (which are
2811 inherited in their own sub-list, so a subdirectory can use this to clear
2812 out the parent's rules).
2814 manpagesection(ANCHORING INCLUDE/EXCLUDE PATTERNS)
2816 As mentioned earlier, global include/exclude patterns are anchored at the
2817 "root of the transfer" (as opposed to per-directory patterns, which are
2818 anchored at the merge-file's directory). If you think of the transfer as
2819 a subtree of names that are being sent from sender to receiver, the
2820 transfer-root is where the tree starts to be duplicated in the destination
2821 directory. This root governs where patterns that start with a / match.
2823 Because the matching is relative to the transfer-root, changing the
2824 trailing slash on a source path or changing your use of the bf(--relative)
2825 option affects the path you need to use in your matching (in addition to
2826 changing how much of the file tree is duplicated on the destination
2827 host). The following examples demonstrate this.
2829 Let's say that we want to match two source files, one with an absolute
2830 path of "/home/me/foo/bar", and one with a path of "/home/you/bar/baz".
2831 Here is how the various command choices differ for a 2-source transfer:
2834 Example cmd: rsync -a /home/me /home/you /dest nl()
2835 +/- pattern: /me/foo/bar nl()
2836 +/- pattern: /you/bar/baz nl()
2837 Target file: /dest/me/foo/bar nl()
2838 Target file: /dest/you/bar/baz nl()
2842 Example cmd: rsync -a /home/me/ /home/you/ /dest nl()
2843 +/- pattern: /foo/bar (note missing "me") nl()
2844 +/- pattern: /bar/baz (note missing "you") nl()
2845 Target file: /dest/foo/bar nl()
2846 Target file: /dest/bar/baz nl()
2850 Example cmd: rsync -a --relative /home/me/ /home/you /dest nl()
2851 +/- pattern: /home/me/foo/bar (note full path) nl()
2852 +/- pattern: /home/you/bar/baz (ditto) nl()
2853 Target file: /dest/home/me/foo/bar nl()
2854 Target file: /dest/home/you/bar/baz nl()
2858 Example cmd: cd /home; rsync -a --relative me/foo you/ /dest nl()
2859 +/- pattern: /me/foo/bar (starts at specified path) nl()
2860 +/- pattern: /you/bar/baz (ditto) nl()
2861 Target file: /dest/me/foo/bar nl()
2862 Target file: /dest/you/bar/baz nl()
2865 The easiest way to see what name you should filter is to just
2866 look at the output when using bf(--verbose) and put a / in front of the name
2867 (use the bf(--dry-run) option if you're not yet ready to copy any files).
2869 manpagesection(PER-DIRECTORY RULES AND DELETE)
2871 Without a delete option, per-directory rules are only relevant on the
2872 sending side, so you can feel free to exclude the merge files themselves
2873 without affecting the transfer. To make this easy, the 'e' modifier adds
2874 this exclude for you, as seen in these two equivalent commands:
2877 tt(rsync -av --filter=': .excl' --exclude=.excl host:src/dir /dest)nl()
2878 tt(rsync -av --filter=':e .excl' host:src/dir /dest)nl()
2881 However, if you want to do a delete on the receiving side AND you want some
2882 files to be excluded from being deleted, you'll need to be sure that the
2883 receiving side knows what files to exclude. The easiest way is to include
2884 the per-directory merge files in the transfer and use bf(--delete-after),
2885 because this ensures that the receiving side gets all the same exclude
2886 rules as the sending side before it tries to delete anything:
2888 quote(tt(rsync -avF --delete-after host:src/dir /dest))
2890 However, if the merge files are not a part of the transfer, you'll need to
2891 either specify some global exclude rules (i.e. specified on the command
2892 line), or you'll need to maintain your own per-directory merge files on
2893 the receiving side. An example of the first is this (assume that the
2894 remote .rules files exclude themselves):
2896 verb(rsync -av --filter=': .rules' --filter='. /my/extra.rules'
2897 --delete host:src/dir /dest)
2899 In the above example the extra.rules file can affect both sides of the
2900 transfer, but (on the sending side) the rules are subservient to the rules
2901 merged from the .rules files because they were specified after the
2902 per-directory merge rule.
2904 In one final example, the remote side is excluding the .rsync-filter
2905 files from the transfer, but we want to use our own .rsync-filter files
2906 to control what gets deleted on the receiving side. To do this we must
2907 specifically exclude the per-directory merge files (so that they don't get
2908 deleted) and then put rules into the local files to control what else
2909 should not get deleted. Like one of these commands:
2911 verb( rsync -av --filter=':e /.rsync-filter' --delete \
2913 rsync -avFF --delete host:src/dir /dest)
2915 manpagesection(BATCH MODE)
2917 Batch mode can be used to apply the same set of updates to many
2918 identical systems. Suppose one has a tree which is replicated on a
2919 number of hosts. Now suppose some changes have been made to this
2920 source tree and those changes need to be propagated to the other
2921 hosts. In order to do this using batch mode, rsync is run with the
2922 write-batch option to apply the changes made to the source tree to one
2923 of the destination trees. The write-batch option causes the rsync
2924 client to store in a "batch file" all the information needed to repeat
2925 this operation against other, identical destination trees.
2927 Generating the batch file once saves having to perform the file
2928 status, checksum, and data block generation more than once when
2929 updating multiple destination trees. Multicast transport protocols can
2930 be used to transfer the batch update files in parallel to many hosts
2931 at once, instead of sending the same data to every host individually.
2933 To apply the recorded changes to another destination tree, run rsync
2934 with the read-batch option, specifying the name of the same batch
2935 file, and the destination tree. Rsync updates the destination tree
2936 using the information stored in the batch file.
2938 For your convenience, a script file is also created when the write-batch
2939 option is used: it will be named the same as the batch file with ".sh"
2940 appended. This script file contains a command-line suitable for updating a
2941 destination tree using the associated batch file. It can be executed using
2942 a Bourne (or Bourne-like) shell, optionally passing in an alternate
2943 destination tree pathname which is then used instead of the original
2944 destination path. This is useful when the destination tree path on the
2945 current host differs from the one used to create the batch file.
2950 tt($ rsync --write-batch=foo -a host:/source/dir/ /adest/dir/)nl()
2951 tt($ scp foo* remote:)nl()
2952 tt($ ssh remote ./foo.sh /bdest/dir/)nl()
2956 tt($ rsync --write-batch=foo -a /source/dir/ /adest/dir/)nl()
2957 tt($ ssh remote rsync --read-batch=- -a /bdest/dir/ <foo)nl()
2960 In these examples, rsync is used to update /adest/dir/ from /source/dir/
2961 and the information to repeat this operation is stored in "foo" and
2962 "foo.sh". The host "remote" is then updated with the batched data going
2963 into the directory /bdest/dir. The differences between the two examples
2964 reveals some of the flexibility you have in how you deal with batches:
2967 it() The first example shows that the initial copy doesn't have to be
2968 local -- you can push or pull data to/from a remote host using either the
2969 remote-shell syntax or rsync daemon syntax, as desired.
2970 it() The first example uses the created "foo.sh" file to get the right
2971 rsync options when running the read-batch command on the remote host.
2972 it() The second example reads the batch data via standard input so that
2973 the batch file doesn't need to be copied to the remote machine first.
2974 This example avoids the foo.sh script because it needed to use a modified
2975 bf(--read-batch) option, but you could edit the script file if you wished to
2976 make use of it (just be sure that no other option is trying to use
2977 standard input, such as the "bf(--exclude-from=-)" option).
2982 The read-batch option expects the destination tree that it is updating
2983 to be identical to the destination tree that was used to create the
2984 batch update fileset. When a difference between the destination trees
2985 is encountered the update might be discarded with a warning (if the file
2986 appears to be up-to-date already) or the file-update may be attempted
2987 and then, if the file fails to verify, the update discarded with an
2988 error. This means that it should be safe to re-run a read-batch operation
2989 if the command got interrupted. If you wish to force the batched-update to
2990 always be attempted regardless of the file's size and date, use the bf(-I)
2991 option (when reading the batch).
2992 If an error occurs, the destination tree will probably be in a
2993 partially updated state. In that case, rsync can
2994 be used in its regular (non-batch) mode of operation to fix up the
2997 The rsync version used on all destinations must be at least as new as the
2998 one used to generate the batch file. Rsync will die with an error if the
2999 protocol version in the batch file is too new for the batch-reading rsync
3000 to handle. See also the bf(--protocol) option for a way to have the
3001 creating rsync generate a batch file that an older rsync can understand.
3002 (Note that batch files changed format in version 2.6.3, so mixing versions
3003 older than that with newer versions will not work.)
3005 When reading a batch file, rsync will force the value of certain options
3006 to match the data in the batch file if you didn't set them to the same
3007 as the batch-writing command. Other options can (and should) be changed.
3008 For instance bf(--write-batch) changes to bf(--read-batch),
3009 bf(--files-from) is dropped, and the
3010 bf(--filter)/bf(--include)/bf(--exclude) options are not needed unless
3011 one of the bf(--delete) options is specified.
3013 The code that creates the BATCH.sh file transforms any filter/include/exclude
3014 options into a single list that is appended as a "here" document to the
3015 shell script file. An advanced user can use this to modify the exclude
3016 list if a change in what gets deleted by bf(--delete) is desired. A normal
3017 user can ignore this detail and just use the shell script as an easy way
3018 to run the appropriate bf(--read-batch) command for the batched data.
3020 The original batch mode in rsync was based on "rsync+", but the latest
3021 version uses a new implementation.
3023 manpagesection(SYMBOLIC LINKS)
3025 Three basic behaviors are possible when rsync encounters a symbolic
3026 link in the source directory.
3028 By default, symbolic links are not transferred at all. A message
3029 "skipping non-regular" file is emitted for any symlinks that exist.
3031 If bf(--links) is specified, then symlinks are recreated with the same
3032 target on the destination. Note that bf(--archive) implies
3035 If bf(--copy-links) is specified, then symlinks are "collapsed" by
3036 copying their referent, rather than the symlink.
3038 rsync also distinguishes "safe" and "unsafe" symbolic links. An
3039 example where this might be used is a web site mirror that wishes
3040 ensure the rsync module they copy does not include symbolic links to
3041 bf(/etc/passwd) in the public section of the site. Using
3042 bf(--copy-unsafe-links) will cause any links to be copied as the file
3043 they point to on the destination. Using bf(--safe-links) will cause
3044 unsafe links to be omitted altogether. (Note that you must specify
3045 bf(--links) for bf(--safe-links) to have any effect.)
3047 Symbolic links are considered unsafe if they are absolute symlinks
3048 (start with bf(/)), empty, or if they contain enough ".."
3049 components to ascend from the directory being copied.
3051 Here's a summary of how the symlink options are interpreted. The list is
3052 in order of precedence, so if your combination of options isn't mentioned,
3053 use the first line that is a complete subset of your options:
3055 dit(bf(--copy-links)) Turn all symlinks into normal files (leaving no
3056 symlinks for any other options to affect).
3058 dit(bf(--links --copy-unsafe-links)) Turn all unsafe symlinks into files
3059 and duplicate all safe symlinks.
3061 dit(bf(--copy-unsafe-links)) Turn all unsafe symlinks into files, noisily
3062 skip all safe symlinks.
3064 dit(bf(--links --safe-links)) Duplicate safe symlinks and skip unsafe
3067 dit(bf(--links)) Duplicate all symlinks.
3069 manpagediagnostics()
3071 rsync occasionally produces error messages that may seem a little
3072 cryptic. The one that seems to cause the most confusion is "protocol
3073 version mismatch -- is your shell clean?".
3075 This message is usually caused by your startup scripts or remote shell
3076 facility producing unwanted garbage on the stream that rsync is using
3077 for its transport. The way to diagnose this problem is to run your
3078 remote shell like this:
3080 quote(tt(ssh remotehost /bin/true > out.dat))
3082 then look at out.dat. If everything is working correctly then out.dat
3083 should be a zero length file. If you are getting the above error from
3084 rsync then you will probably find that out.dat contains some text or
3085 data. Look at the contents and try to work out what is producing
3086 it. The most common cause is incorrectly configured shell startup
3087 scripts (such as .cshrc or .profile) that contain output statements
3088 for non-interactive logins.
3090 If you are having trouble debugging filter patterns, then
3091 try specifying the bf(-vv) option. At this level of verbosity rsync will
3092 show why each individual file is included or excluded.
3094 manpagesection(EXIT VALUES)
3098 dit(bf(1)) Syntax or usage error
3099 dit(bf(2)) Protocol incompatibility
3100 dit(bf(3)) Errors selecting input/output files, dirs
3101 dit(bf(4)) Requested action not supported: an attempt
3102 was made to manipulate 64-bit files on a platform that cannot support
3103 them; or an option was specified that is supported by the client and
3105 dit(bf(5)) Error starting client-server protocol
3106 dit(bf(6)) Daemon unable to append to log-file
3107 dit(bf(10)) Error in socket I/O
3108 dit(bf(11)) Error in file I/O
3109 dit(bf(12)) Error in rsync protocol data stream
3110 dit(bf(13)) Errors with program diagnostics
3111 dit(bf(14)) Error in IPC code
3112 dit(bf(20)) Received SIGUSR1 or SIGINT
3113 dit(bf(21)) Some error returned by code(waitpid())
3114 dit(bf(22)) Error allocating core memory buffers
3115 dit(bf(23)) Partial transfer due to error
3116 dit(bf(24)) Partial transfer due to vanished source files
3117 dit(bf(25)) The --max-delete limit stopped deletions
3118 dit(bf(30)) Timeout in data send/receive
3119 dit(bf(35)) Timeout waiting for daemon connection
3122 manpagesection(ENVIRONMENT VARIABLES)
3125 dit(bf(CVSIGNORE)) The CVSIGNORE environment variable supplements any
3126 ignore patterns in .cvsignore files. See the bf(--cvs-exclude) option for
3128 dit(bf(RSYNC_ICONV)) Specify a default bf(--iconv) setting using this
3129 environment variable.
3130 dit(bf(RSYNC_RSH)) The RSYNC_RSH environment variable allows you to
3131 override the default shell used as the transport for rsync. Command line
3132 options are permitted after the command name, just as in the bf(-e) option.
3133 dit(bf(RSYNC_PROXY)) The RSYNC_PROXY environment variable allows you to
3134 redirect your rsync client to use a web proxy when connecting to a
3135 rsync daemon. You should set RSYNC_PROXY to a hostname:port pair.
3136 dit(bf(RSYNC_PASSWORD)) Setting RSYNC_PASSWORD to the required
3137 password allows you to run authenticated rsync connections to an rsync
3138 daemon without user intervention. Note that this does not supply a
3139 password to a remote shell transport such as ssh; to learn how to do that,
3140 consult the remote shell's documentation.
3141 dit(bf(USER) or bf(LOGNAME)) The USER or LOGNAME environment variables
3142 are used to determine the default username sent to an rsync daemon.
3143 If neither is set, the username defaults to "nobody".
3144 dit(bf(HOME)) The HOME environment variable is used to find the user's
3145 default .cvsignore file.
3150 /etc/rsyncd.conf or rsyncd.conf
3158 times are transferred as *nix time_t values
3160 When transferring to FAT filesystems rsync may re-sync
3162 See the comments on the bf(--modify-window) option.
3164 file permissions, devices, etc. are transferred as native numerical
3167 see also the comments on the bf(--delete) option
3169 Please report bugs! See the web site at
3170 url(http://rsync.samba.org/)(http://rsync.samba.org/)
3172 manpagesection(VERSION)
3174 This man page is current for version 3.0.3 of rsync.
3176 manpagesection(INTERNAL OPTIONS)
3178 The options bf(--server) and bf(--sender) are used internally by rsync,
3179 and should never be typed by a user under normal circumstances. Some
3180 awareness of these options may be needed in certain scenarios, such as
3181 when setting up a login that can only run an rsync command. For instance,
3182 the support directory of the rsync distribution has an example script
3183 named rrsync (for restricted rsync) that can be used with a restricted
3186 manpagesection(CREDITS)
3188 rsync is distributed under the GNU public license. See the file
3189 COPYING for details.
3191 A WEB site is available at
3192 url(http://rsync.samba.org/)(http://rsync.samba.org/). The site
3193 includes an FAQ-O-Matic which may cover questions unanswered by this
3196 The primary ftp site for rsync is
3197 url(ftp://rsync.samba.org/pub/rsync)(ftp://rsync.samba.org/pub/rsync).
3199 We would be delighted to hear from you if you like this program.
3200 Please contact the mailing-list at rsync@lists.samba.org.
3202 This program uses the excellent zlib compression library written by
3203 Jean-loup Gailly and Mark Adler.
3205 manpagesection(THANKS)
3207 Especial thanks go out to: John Van Essen, Matt McCutchen, Wesley W. Terpstra,
3208 David Dykstra, Jos Backus, Sebastian Krahmer, Martin Pool, and our
3209 gone-but-not-forgotten compadre, J.W. Schultz.
3211 Thanks also to Richard Brent, Brendan Mackay, Bill Waite, Stephen Rothwell
3212 and David Bell. I've probably missed some people, my apologies if I have.
3216 rsync was originally written by Andrew Tridgell and Paul Mackerras.
3217 Many people have later contributed to it. It is currently maintained
3220 Mailing lists for support and development are available at
3221 url(http://lists.samba.org)(lists.samba.org)