rsync - faster, flexible replacement for rcp
rsync [OPTION]... SRC [SRC]... DEST
rsync [OPTION]... SRC [SRC]... [USER@]HOST:DEST
rsync [OPTION]... SRC [SRC]... [USER@]HOST::DEST
rsync [OPTION]... SRC [SRC]... rsync://[USER@]HOST[:PORT]/DEST
rsync [OPTION]... SRC
rsync [OPTION]... [USER@]HOST:SRC [DEST]
rsync [OPTION]... [USER@]HOST::SRC [DEST]
rsync [OPTION]... rsync://[USER@]HOST[:PORT]/SRC [DEST]
rsync is a program that behaves in much the same way that rcp does, but has many more options and uses the rsync remote-update protocol to greatly speed up file transfers when the destination file is being updated.
The rsync remote-update protocol allows rsync to transfer just the differences between two sets of files across the network connection, using an efficient checksum-search algorithm described in the technical report that accompanies this package.
Some of the additional features of rsync are:
Rsync copies files either to or from a remote host, or locally on the current host (it does not support copying files between two remote hosts).
There are two different ways for rsync to contact a remote system: using a remote-shell program as the transport (such as ssh or rsh) or contacting an rsync daemon directly via TCP. The remote-shell transport is used whenever the source or destination path contains a single colon (:) separator after a host specification. Contacting an rsync daemon directly happens when the source or destination path contains a double colon (::) separator after a host specification, OR when an rsync:// URL is specified (see also the “USING RSYNC-DAEMON FEATURES VIA A REMOTE-SHELL CONNECTION” section for an exception to this latter rule).
As a special case, if a single source arg is specified without a destination, the files are listed in an output format similar to “ls -l".
As expected, if neither the source or destination path specify a remote host, the copy occurs locally (see also the --list-only option).
See the file README for installation instructions.
Once installed, you can use rsync to any machine that you can access via a remote shell (as well as some that you can access using the rsync daemon-mode protocol). For remote transfers, a modern rsync uses ssh for its communications, but it may have been configured to use a different remote shell by default, such as rsh or remsh.
You can also specify any remote shell you like, either by using the -e command line option, or by setting the RSYNC_RSH environment variable.
Note that rsync must be installed on both the source and destination machines.
You use rsync in the same way you use rcp. You must specify a source and a destination, one of which may be remote.
Perhaps the best way to explain the syntax is with some examples:
rsync -t *.c foo:src/
This would transfer all files matching the pattern *.c from the current directory to the directory src on the machine foo. If any of the files already exist on the remote system then the rsync remote-update protocol is used to update the file by sending only the differences. See the tech report for details.
rsync -avz foo:src/bar /data/tmp
This would recursively transfer all files from the directory src/bar on the machine foo into the /data/tmp/bar directory on the local machine. The files are transferred in “archive” mode, which ensures that symbolic links, devices, attributes, permissions, ownerships, etc. are preserved in the transfer. Additionally, compression will be used to reduce the size of data portions of the transfer.
rsync -avz foo:src/bar/ /data/tmp
A trailing slash on the source changes this behavior to avoid creating an additional directory level at the destination. You can think of a trailing / on a source as meaning “copy the contents of this directory" as opposed to “copy the directory by name", but in both cases the attributes of the containing directory are transferred to the containing directory on the destination. In other words, each of the following commands copies the files in the same way, including their setting of the attributes of /dest/foo:
rsync -av /src/foo /dest
rsync -av /src/foo/ /dest/foo
Note also that host and module references don’t require a trailing slash to copy the contents of the default directory. For example, both of these copy the remote directory’s contents into “/dest":
rsync -av host: /dest
rsync -av host::module /dest
You can also use rsync in local-only mode, where both the source and destination don’t have a ‘:’ in the name. In this case it behaves like an improved copy command.
Finally, you can list all the (listable) modules available from a particular rsync daemon by leaving off the module name:
See the following section for more details.
The syntax for requesting multiple files from a remote host involves using quoted spaces in the SRC. Some examples:
rsync host::’modname/dir1/file1 modname/dir2/file2’ /dest
This would copy file1 and file2 into /dest from an rsync daemon. Each additional arg must include the same “modname/” prefix as the first one, and must be preceded by a single space. All other spaces are assumed to be a part of the filenames.
rsync -av host:’dir1/file1 dir2/file2’ /dest
This would copy file1 and file2 into /dest using a remote shell. This word-splitting is done by the remote shell, so if it doesn’t work it means that the remote shell isn’t configured to split its args based on whitespace (a very rare setting, but not unknown). If you need to transfer a filename that contains whitespace, you’ll need to either escape the whitespace in a way that the remote shell will understand, or use wildcards in place of the spaces. Two examples of this are:
rsync -av host:’file\ name\ with\ spaces’ /dest rsync -av host:file?name?with?spaces /dest
This latter example assumes that your shell passes through unmatched wildcards. If it complains about “no match", put the name in quotes.
It is also possible to use rsync without a remote shell as the transport. In this case you will directly connect to a remote rsync daemon, typically using TCP port 873. (This obviously requires the daemon to be running on the remote system, so refer to the STARTING AN RSYNC DAEMON TO ACCEPT CONNECTIONS section below for information on that.)
Using rsync in this way is the same as using it with a remote shell except that:
An example that copies all the files in a remote module named “src":
rsync -av host::src /dest
Some modules on the remote daemon may require authentication. If so, you will receive a password prompt when you connect. You can avoid the password prompt by setting the environment variable RSYNC_PASSWORD to the password you want to use or using the --password-file option. This may be useful when scripting rsync.
WARNING: On some systems environment variables are visible to all users. On those systems using --password-file is recommended.
You may establish the connection via a web proxy by setting the environment variable RSYNC_PROXY to a hostname:port pair pointing to your web proxy. Note that your web proxy’s configuration must support proxy connections to port 873.
It is sometimes useful to use various features of an rsync daemon (such as named modules) without actually allowing any new socket connections into a system (other than what is already required to allow remoteshell access). Rsync supports connecting to a host using a remote shell and then spawning a single-use “daemon” server that expects to read its config file in the home dir of the remote user. This can be useful if you want to encrypt a daemon-style transfer’s data, but since the daemon is started up fresh by the remote user, you may not be able to use features such as chroot or change the uid used by the daemon. (For another way to encrypt a daemon transfer, consider using ssh to tunnel a local port to a remote machine and configure a normal rsync daemon on that remote host to only allow connections from “localhost".)
From the user’s perspective, a daemon transfer via a remote-shell connection uses nearly the same command-line syntax as a normal rsync-daemon transfer, with the only exception being that you must explicitly set the remote shell program on the command-line with the --rsh=COMMAND option. (Setting the RSYNC_RSH in the environment will not turn on this functionality.) For example:
rsync -av --rsh=ssh host::module /dest
If you need to specify a different remote-shell user, keep in mind that the user@ prefix in front of the host is specifying the rsync-user value (for a module that requires user-based authentication). This means that you must give the ‘-l user’ option to ssh when specifying the remote-shell, as in this example that uses the short version of the --rsh option:
rsync -av -e “ssh -l ssh-user” rsync-user@host::module /dest
The “ssh-user” will be used at the ssh level; the “rsync-user” will be used to log-in to the “module".
In order to connect to an rsync daemon, the remote system needs to have a daemon already running (or it needs to have configured something like inetd to spawn an rsync daemon for incoming connections on a particular port). For full information on how to start a daemon that will handling incoming socket connections, see the rsyncd.conf(5) man page -that is the config file for the daemon, and it contains the full details for how to run the daemon (including stand-alone and inetd configurations).
If you’re using one of the remote-shell transports for the transfer, there is no need to manually start an rsync daemon.
Here are some examples of how I use rsync.
To backup my wife’s home directory, which consists of large MS Word files and mail folders, I use a cron job that runs
rsync -Cavz . arvidsjaur:backup
each night over a PPP connection to a duplicate directory on my machine “arvidsjaur".
To synchronize my samba source trees I use the following Makefile targets:
rsync -avuzb --exclude ‘*~’ samba:samba/ . put:
rsync -Cavuzb . samba:samba/ sync: get put
this allows me to sync with a CVS directory at the other end of the connection. I then do CVS operations on the remote machine, which saves a lot of time as the remote CVS protocol isn’t very efficient.
I mirror a directory between my “old” and “new” ftp sites with the command:
rsync -az -e ssh --delete ~ftp/pub/samba nimbus:"~ftp/pub/tridge"
This is launched from cron every few hours.
Here is a short summary of the options available in rsync. Please refer to the detailed description below for a complete description.
Rsync can also be run as a daemon, in which case the following options are accepted:
rsync uses the GNU long options package. Many of the command line options have two variants, one short and one long. These are shown below, separated by commas. Some options only have a long variant. The ‘=’ for options that take a parameter is optional; whitespace can be used instead.
Note that the names of the transferred files that are output are done using a default --log-format of “%n%L", which tells you just the name of the file and, if the item is a link, where it points. At the single -v level of verbosity, this does not mention when a file gets its attributes changed. If you ask for an itemized list of changed attributes (either --itemize-changes or adding “%i” to the --log-format setting), the output (on the client) increases to mention all items that are changed in any way. See the --log-format option for more details.
Note that rsync always verifies that each transferred file was correctly reconstructed on the receiving side by checking its whole-file checksum, but that automatic after-the-transfer verification has nothing to do with this option’s before-the-transfer “Does this file need to be updated?” check.
Note that -a does not preserve hardlinks, because finding multiply-linked files is expensive. You must separately specify -H.
For example: if you want to use -a (--archive) but don’t want -o (--owner), instead of converting -a into -rlptgD, you could specify -a --no-o (or -a --no-owner).
The order of the options is important: if you specify --no-r -a, the -r option would end up being turned on, the opposite of -a --no-r. Note also that the side-effects of the --files-from option are NOT positional, as it affects the default state of several options and slightly changes the meaning of -a (see the --files-from option for more details).
rsync -av /foo/bar/baz.c remote:/tmp/
rsync -avR /foo/bar/baz.c remote:/tmp/
then a file named /tmp/foo/bar/baz.c would be created on the remote machine -- the full path name is preserved. To limit the amount of path information that is sent, you have a couple options: (1) With a modern rsync on the sending side (beginning with 2.6.7), you can insert a dot and a slash into the source path, like this:
rsync -avR /foo/./bar/baz.c remote:/tmp/
That would create /tmp/bar/baz.c on the remote machine. (Note that the dot must be followed by a slash, so “/foo/.” would not be abbreviated.) (2) For older rsync versions, you would need to use a chdir to limit the source path. For example, when pushing files:
(cd /foo; rsync -avR bar/baz.c remote:/tmp/)
(Note that the parens put the two commands into a sub-shell, so that the “cd” command doesn’t remain in effect for future commands.) If you’re pulling files, use this idiom (which doesn’t work with an rsync daemon):
rsync -avR --rsync-path="cd /foo; rsync” \ remote:bar/baz.c /tmp/
For instance, if a command-line arg or a files-from entry told rsync to transfer the file “path/foo/file", the directories “path” and “path/foo” are implied when --relative is used. If “path/foo” is a symlink to “bar” on the destination system, the receiving rsync would ordinarily delete “path/foo", recreate it as a directory, and receive the file into the new directory. With --no-implied-dirs, the receiving rsync updates “path/foo/file” using the existing path elements, which means that the file ends up being created in “path/bar". Another way to accomplish this link preservation is to use the --keepdirlinks option (which will also affect symlinks to directories in the rest of the transfer).
In a similar but opposite scenario, if the transfer of “path/foo/file” is requested and “path/foo” is a symlink on the sending side, running without --no-implied-dirs would cause rsync to transform “path/foo” on the receiving side into an identical symlink, and then attempt to transfer “path/foo/file", which might fail if the duplicated symlink did not point to a directory on the receiving side. Another way to avoid this sending of a symlink as an implied directory is to use --copyunsafe-links, or --copy-dirlinks (both of which also affect symlinks in the rest of the transfer -- see their descriptions for full details).
Note that if you don’t specify --backup-dir, (1) the --omit-dirtimes option will be implied, and (2) if --delete is also in effect (without --delete-excluded), rsync will add a “protect" filter-rule for the backup suffix to the end of all your existing excludes (e.g. -f “P *~"). This will prevent previously backed-up files from being deleted. Note that if you are supplying your own filter rules, you may need to manually insert your own exclude/protect rule somewhere higher up in the list so that it has a high enough priority to be effective (e.g., if your rules specify a trailing inclusion/exclusion of ‘*’, the auto-added rule would never be reached).
In the current implementation of --update, a difference of file format between the sender and receiver is always considered to be important enough for an update, no matter what date is on the objects. In other words, if the source has a directory or a symlink where the destination has a file, the transfer would occur regardless of the timestamps. This might change in the future (feel free to comment on this on the mailing list if you have an opinion).
This option is useful for transfer of large files with blockbased changes or appended data, and also on systems that are disk bound, not network bound.
The option implies --partial (since an interrupted transfer does not delete the file), but conflicts with --partial-dir and --delay-updates. Prior to rsync 2.6.4 --inplace was also incompatible with --compare-dest and --link-dest.
WARNING: The file’s data will be in an inconsistent state during the transfer (and possibly afterward if the transfer gets interrupted), so you should not use this option to update files that are in use. Also note that rsync will be unable to update a file in-place that is not writable by the receiving user.
Without this option, if the sending side has replaced a directory with a symlink to a directory, the receiving side will delete anything that is in the way of the new symlink, including a directory hierarchy (as long as --force or --delete is in effect).
See also --keep-dirlinks for an analogous option for the receiving side.
For example, suppose you transfer a directory “foo” that contains a file “file", but “foo” is a symlink to directory “bar" on the receiver. Without --keep-dirlinks, the receiver deletes symlink “foo", recreates it as a directory, and receives the file into the new directory. With --keep-dirlinks, the receiver keeps the symlink and “file” ends up in “bar".
See also --copy-dirlinks for an analogous option for the sending side.
Note that rsync can only detect hard links if both parts of the link are in the list of files being sent.
When this option is off, permissions are set as follows:
Thus, when --perms and --executability are both disabled, rsync’s behavior is the same as that of other file-copy utilities, such as cp(1) and tar(1) .
In summary: to give destination files (both old and new) the source permissions, use --perms. To give new files the destination-default permissions (while leaving existing files unchanged), make sure that the --perms option is off and use --chmod=ugo=rwX (which ensures that all non-masked bits get enabled). If you’d care to make this latter behavior easier to type, you could define a popt alias for it, such as putting this line in the file ~/.popt (this defines the -s option, and includes --no-g to use the default group of the destination dir):
rsync alias -s --no-p --no-g --chmod=ugo=rwX
You could then use this new option in a command such as this one:
rsync -asv src/ dest/
(Caveat: make sure that -a does not follow -s, or it will reenable the “--no-*” options.)
The preservation of the destination’s setgid bit on newly-created directories when --perms is off was added in rsync 2.6.7. Older rsync versions erroneously preserved the three special permission bits for newly-created files when --perms was off, while overriding the destination’s setgid bit setting on a newly-created directory. Default ACL observance was added to the ACL patch for rsync 2.6.7, so older (or non-ACL-enabled) rsyncs use the umask even if default ACLs are present. (Keep in mind that it is the version of the receiving rsync that affects these behaviors.)
If --perms is enabled, this option is ignored.
Note also that an optimization of the ACL-sending protocol used by this version makes it incompatible with sending files to an older ACL-enabled rsync unless you double the --acls option (e.g. -AA). This doubling is not needed when pulling files from an older rsync.
In addition to the normal parsing rules specified in the chmod(1) manpage, you can specify an item that should only apply to a directory by prefixing it with a ‘D’, or specify an item that should only apply to a file by prefixing it with a ‘F’. For example:
It is also legal to specify multiple --chmod options, as each additional option is just appended to the list of changes to make.
See the --perms and --executability options for how the resulting permission value can be applied to the files in the transfer.
The preservation of ownership will associate matching names by default, but may fall back to using the ID number in some circumstances (see also the --numeric-ids option for a full discussion).
The preservation of group information will associate matching names by default, but may fall back to using the ID number in some circumstances (see also the --numeric-ids option for a full discussion).
NOTE: Don’t use this option when the destination is a Solaris “tmpfs” filesystem. It doesn’t seem to handle seeks over null regions correctly and ends up corrupting the files.
If this option is repeated, rsync omits all mount-point directories from the copy. Otherwise, it includes an empty directory at each mount-point it encounters (using the attributes of the mounted directory because those of the underlying mount-point directory are inaccessible).
If rsync has been told to collapse symlinks (via --copy-links or --copy-unsafe-links), a symlink to a directory on another device is treated like a mount-point. Symlinks to non-directories are unaffected by this option.
Prior to rsync 2.6.7, this option would have no effect unless --recursive was in effect. Beginning with 2.6.7, deletions will also occur when --dirs (-d) is in effect, but only for directories whose contents are being copied.
This option can be dangerous if used incorrectly! It is a very good idea to run first using the --dry-run option (-n) to see what files would be deleted to make sure important files aren’t listed.
If the sending side detects any I/O errors, then the deletion of any files at the destination will be automatically disabled. This is to prevent temporary filesystem failures (such as NFS errors) on the sending side causing a massive deletion of files on the destination. You can override this with the --ignoreerrors option.
The --delete option may be combined with one of the --deleteWHEN options without conflict, as well as --delete-excluded. However, if none of the --delete-WHEN options are specified, rsync will currently choose the --delete-before algorithm. A future version may change this to choose the --delete-during algorithm. See also --delete-after.
Deleting before the transfer is helpful if the filesystem is tight for space and removing extraneous files would help to make the transfer possible. However, it does introduce a delay before the start of the transfer, and this delay might cause the transfer to timeout (if --timeout was specified).
Note for older rsync versions: --force used to still be required when using --delete-after, and it used to be non-functional unless the --recursive option was also enabled.
The suffixes are as follows: “K” (or “KiB") is a kibibyte (1024), “M” (or “MiB") is a mebibyte (1024*1024), and “G” (or “GiB") is a gibibyte (1024*1024*1024). If you want the multiplier to be 1000 instead of 1024, use “KB", “MB", or “GB". (Note: lower-case is also accepted for all values.) Finally, if the suffix ends in either “+1” or “-1", the value will be offset by one byte in the indicated direction.
Examples: --max-size=1.5mb-1 is 1499999 bytes, and --maxsize=2g+1 is 2147483649 bytes.
If this option is used with [user@]host::module/path, then the remote shell COMMAND will be used to run an rsync daemon on the remote host, and all data will be transmitted through that remote shell connection, rather than through a direct socket connection to a running rsync daemon on the remote host. See the section “USING RSYNC-DAEMON FEATURES VIA A REMOTE-SHELL CONNECTION" above.
Command-line arguments are permitted in COMMAND provided that COMMAND is presented to rsync as a single argument. You must use spaces (not tabs or other whitespace) to separate the command and args from each other, and you can use single- and/or double-quotes to preserve spaces in an argument (but not backslashes). Note that doubling a single-quote inside a singlequoted string gives you a single-quote; likewise for doublequotes (though you need to pay attention to which quotes your shell is parsing and which quotes rsync is parsing). Some examples:
(Note that ssh users can alternately customize site-specific connect options in their .ssh/config file.)
You can also choose the remote shell program using the RSYNC_RSH environment variable, which accepts the same range of values as -e.
See also the --blocking-io option which is affected by this option.
One tricky example is to set a different default directory on the remote machine for use with the --relative option. For instance:
rsync -avR --rsync-path="cd /a/b && rsync” hst:c/d /e/
The exclude list is initialized to:
RCS SCCS CVS CVS.adm RCSLOG cvslog.* tags TAGS .make.state .nse_depinfo *~ #* .#* ,* _$* *$ *.old *.bak *.BAK *.orig *.rej .del-* *.a *.olb *.o *.obj *.so *.exe *.Z *.elc *.ln core .svn/
then files listed in a $HOME/.cvsignore are added to the list and any files listed in the CVSIGNORE environment variable (all cvsignore names are delimited by whitespace).
Finally, any file is ignored if it is in the same directory as a .cvsignore file and matches one of the patterns listed therein. Unlike rsync’s filter/exclude files, these patterns are split on whitespace. See the cvs(1) manual for more information.
If you’re combining -C with your own --filter rules, you should note that these CVS excludes are appended at the end of your own rules, regardless of where the -C was placed on the commandline. This makes them a lower priority than any rules you specified explicitly. If you want to control where these CVS excludes get inserted into your filter rules, you should omit the -C as a command-line option and use a combination of --filter=:C and --filter=-C (either on your command-line or by putting the “:C” and “-C” rules into a filter file with your other rules). The first option turns on the per-directory scanning for the .cvsignore file. The second option does a one-time import of the CVS excludes mentioned above.
You may use as many --filter options on the command line as you like to build up the list of files to exclude.
See the FILTER RULES section for detailed information on this option.
This tells rsync to look for per-directory .rsync-filter files that have been sprinkled through the hierarchy and use their rules to filter the files in the transfer. If -F is repeated, it is a shorthand for this rule:
This filters out the .rsync-filter files themselves from the transfer.
See the FILTER RULES section for detailed information on how these options work.
See the FILTER RULES section for detailed information on this option.
See the FILTER RULES section for detailed information on this option.
The file names that are read from the FILE are all relative to the source dir -- any leading slashes are removed and no “.." references are allowed to go higher than the source dir. For example, take this command:
rsync -a --files-from=/tmp/foo /usr remote:/backup
If /tmp/foo contains the string “bin” (or even “/bin"), the /usr/bin directory will be created as /backup/bin on the remote host. If it contains “bin/” (note the trailing slash), the immediate contents of the directory would also be sent (without needing to be explicitly mentioned in the file -- this began in version 2.6.4). In both cases, if the -r option was enabled, that dir’s entire hierarchy would also be transferred (keep in mind that -r needs to be specified explicitly with --files-from, since it is not implied by -a). Also note that the effect of the (enabled by default) --relative option is to duplicate only the path info that is read from the file -- it does not force the duplication of the source-spec path (/usr in this case).
In addition, the --files-from file can be read from the remote host instead of the local host if you specify a “host:” in front of the file (the host must match one end of the transfer). As a short-cut, you can specify just a prefix of “:” to mean “use the remote end of the transfer". For example:
rsync -a --files-from=:/path/file-list src:/ /tmp/copy
This would copy all the files specified in the /path/file-list file that was located on the remote “src” host.
This option is most often used when the receiving disk partition does not have enough free space to hold a copy of the largest file in the transfer. In this case (i.e. when the scratch directory in on a different disk partition), rsync will not be able to rename each received temporary file over the top of the associated destination file, but instead must copy it into place. Rsync does this by copying the file over the top of the destination file, which means that the destination file will contain truncated data during this copy. If this were not done this way (even if the destination file were first removed, the data locally copied to a temporary file in the destination directory, and then renamed into place) it would be possible for the old file to continue taking up disk space (if someone had it open), and thus there might not be enough room to fit the new version on the disk at the same time.
If you are using this option for reasons other than a shortage of disk space, you may wish to combine it with the --delayupdates option, which will ensure that all copied files get put into subdirectories in the destination hierarchy, awaiting the end of the transfer. If you don’t have enough room to duplicate all the arriving files on the destination partition, another way to tell rsync that you aren’t overly concerned about disk space is to use the --partial-dir option with a relative path; because this tells rsync that it is OK to stash off a copy of a single file in a subdir in the destination hierarchy, rsync will use the partial-dir as a staging area to bring over the copied file, and then rename it into place from there. (Specifying a --partial-dir with an absolute path does not have this side-effect.)
Note that the use of the --delete option might get rid of any potential fuzzy-match files, so either use --delete-after or specify some filename exclusions if you need to prevent this.
Beginning in version 2.6.4, multiple --compare-dest directories may be provided, which will cause rsync to search the list in the order specified for an exact match. If a match is found that differs only in attributes, a local copy is made and the attributes updated. If a match is not found, a basis file from one of the DIRs will be selected to try to speed up the transfer.
If DIR is a relative path, it is relative to the destination directory. See also --copy-dest and --link-dest.
Multiple --copy-dest directories may be provided, which will cause rsync to search the list in the order specified for an unchanged file. If a match is not found, a basis file from one of the DIRs will be selected to try to speed up the transfer.
If DIR is a relative path, it is relative to the destination directory. See also --compare-dest and --link-dest.
rsync -av --link-dest=$PWD/prior_dir host:src_dir/ new_dir/
Beginning in version 2.6.4, multiple --link-dest directories may be provided, which will cause rsync to search the list in the order specified for an exact match. If a match is found that differs only in attributes, a local copy is made and the attributes updated. If a match is not found, a basis file from one of the DIRs will be selected to try to speed up the transfer.
If DIR is a relative path, it is relative to the destination directory. See also --compare-dest and --copy-dest.
Note that rsync versions prior to 2.6.1 had a bug that could prevent --link-dest from working properly for a non-super-user when -o was specified (or implied by -a). You can work-around this bug by avoiding the -o option when sending to an old rsync.
Note that this option typically achieves better compression ratios than can be achieved by using a compressing remote shell or a compressing transport because it takes advantage of the implicit information in the matching data blocks that are not explicitly sent over the connection.
By default rsync will use the username and groupname to determine what ownership to give files. The special uid 0 and the special group 0 are never mapped via user/group names even if the --numeric-ids option is not specified.
If a user or group has no name on the source system or it has no match on the destination system, then the numeric ID from the source system is used instead. See also the comments on the “use chroot” setting in the rsyncd.conf manpage for information on how the chroot setting affects rsync’s ability to look up the names of the users and groups and what you can do about it.
The “%i” escape has a cryptic output that is 9 letters long. The general format is like the string YXcstpogz, where Y is replaced by the type of update being done, X is replaced by the file-type, and the other letters represent attributes that may be output if they are being modified.
The update types that replace the Y are as follows:
The file-types that replace the X are: f for a file, a d for a directory, an L for a symlink, a D for a device, and a S for a special file (e.g. named sockets and fifos).
The other letters in the string above are the actual letters that will be output if the associated attribute for the item is being updated or a “.” for no change. Three exceptions to this are: (1) a newly created item replaces each letter with a “+", (2) an identical item replaces the dots with spaces, and (3) an unknown attribute replaces each letter with a “?” (this can happen when talking to an older rsync).
The attribute that is associated with each letter is as follows:
One other output is possible: when deleting files, the “%i" will output the string “*deleting” for each item that is being removed (assuming that you are talking to a recent enough rsync that it logs deletions instead of outputting them as a verbose message).
Specifying this option will mention each file, dir, etc. that gets updated in a significant way (a transferred file, a recreated symlink/device, or a touched directory) unless the itemizechanges escape (%i) is included in the string, in which case the logging of names increases to mention any item that is changed in any way (as long as the receiving side is at least 2.6.4). See the --itemize-changes option for a description of the output of “%i".
The --verbose option implies a format of “%n%L", but you can use --log-format without --verbose if you like, or you can override the format of its per-file output using this option.
Rsync will output the log-format string prior to a file’s transfer unless one of the transfer-statistic escapes is requested, in which case the logging is done at the end of the file’s transfer. When this late logging is in effect and --progress is also specified, rsync will also output the name of the file being transferred prior to its progress information (followed, of course, by the log-format output).
The current statistics are as follows:
The escape idiom that started in 2.6.7 is to output a literal backslash (\) and a hash (#), followed by exactly 3 octal digits. For example, a newline would output as “\#012". A literal backslash that is in a filename is not escaped unless it is followed by a hash and 3 digits (0-9).
Note that if --whole-file is specified (or implied), any partial-dir file that is found for a file that is being updated will simply be removed (since rsync is sending files without using the incremental rsync algorithm).
Rsync will create the DIR if it is missing (just the last dir -not the whole path). This makes it easy to use a relative path (such as “--partial-dir=.rsync-partial") to have rsync create the partial-directory in the destination file’s directory when needed, and then remove it again when the partial file is deleted.
If the partial-dir value is not an absolute path, rsync will add an exclude rule at the end of all your existing excludes. This will prevent the sending of any partial-dir files that may exist on the sending side, and will also prevent the untimely deletion of partial-dir items on the receiving side. An example: the above --partial-dir option would add the equivalent of “--exclude=.rsync-partial/” at the end of any other filter rules.
If you are supplying your own exclude rules, you may need to add your own exclude/hide/protect rule for the partial-dir because (1) the auto-added rule may be ineffective at the end of your other rules, or (2) you may wish to override rsync’s exclude choice. For instance, if you want to make rsync clean-up any left-over partial-dirs that may be lying around, you should specify --delete-after and add a “risk” filter rule, e.g. -f ‘R .rsync-partial/’. (Avoid using --delete-before or --delete-during unless you don’t need rsync to use any of the left-over partial-dir data during the current run.)
IMPORTANT: the --partial-dir should not be writable by other users or it is a security risk. E.g. AVOID “/tmp".
You can also set the partial-dir value the RSYNC_PARTIAL_DIR environment variable. Setting this in the environment does not force --partial to be enabled, but rather it affects where partial files go when --partial is specified. For instance, instead of using --partial-dir=.rsync-tmp along with --progress, you could set RSYNC_PARTIAL_DIR=.rsync-tmp in your environment and then just use the -P option to turn on the use of the .rsync-tmp dir for partial transfers. The only times that the --partial option does not look for this environment value are (1) when --inplace was specified (since --inplace conflicts with --partial-dir), and (2) when --delay-updates was specified (see below).
For the purposes of the daemon-config’s “refuse options” setting, --partial-dir does not imply --partial. This is so that a refusal of the --partial option can be used to disallow the overwriting of destination files with a partial transfer, while still allowing the safer idiom provided by --partial-dir.
This option uses more memory on the receiving side (one bit per file transferred) and also requires enough free disk space on the receiving side to hold an additional copy of all the updated files. Note also that you should not use an absolute path to --partial-dir unless (1) there is no chance of any of the files in the transfer having the same name (since all the updated files will be put into a single directory if the path is absolute) and (2) there are no mount points in the hierarchy (since the delayed updates will fail if they can’t be renamed into place).
See also the “atomic-rsync” perl script in the “support” subdir for an update algorithm that is even more atomic (it uses --link-dest and a parallel hierarchy of files).
Because the file-list is actually being pruned, this option also affects what directories get deleted when a delete is active. However, keep in mind that excluded files and directories can prevent existing items from being deleted (because an exclude hides source files and protects destination files).
You can prevent the pruning of certain empty directories from the file-list by using a global “protect” filter. For instance, this option would ensure that the directory “emptydir” was kept in the file-list:
Here’s an example that copies all .pdf files in a hierarchy, only creating the necessary destination directories to hold the .pdf files, and ensures that any superfluous files and directories in the destination are removed (note the hide filter of non-directories being used instead of an exclude):
rsync -avm --del --include=’*.pdf’ -f ‘hide,! */’ src/ dest
If you didn’t want to remove superfluous destination files, the more time-honored options of “--include=’*/’ --exclude=’*’" would work fine in place of the hide-filter (if that is more natural to you).
When the file is transferring, the data looks like this:
782448 63% 110.64kB/s 0:00:04
This tells you the current file size, the percentage of the transfer that is complete, the current calculated file-completion rate (including both data over the wire and data being matched locally), and the estimated time remaining in this transfer.
After a file is complete, the data looks like this:
1238099 100% 146.38kB/s 0:00:08 (5, 57.1% of 396)
This tells you the final file size, that it’s 100% complete, the final transfer rate for the file, the amount of elapsed time it took to transfer the file, and the addition of a total-transfer summary in parentheses. These additional numbers tell you how many files have been updated, and what percent of the total number of files has been scanned.
rsync -av --list-only foo* dest/
Note that you can feel free to write the batch directly to some portable media: if this media fills to capacity before the end of the transfer, you can just apply that partial transfer to the destination and repeat the whole process to get the rest of the changes (as long as you don’t mind a partially updated destination system while the multi-update cycle is happening).
Also note that you only save bandwidth when pushing changes to a remote system because this allows the batched data to be diverted from the sender into the batch file without having to flow over the wire to the receiver (when pulling, the sender is remote, and thus can’t write the batch).
The options allowed when starting an rsync daemon are as follows:
If standard input is a socket then rsync will assume that it is being run via inetd, otherwise it will detach from the current terminal and become a background daemon. The daemon will read the config file (rsyncd.conf) on each connect made by a client and respond to requests accordingly. See the rsyncd.conf(5) man page for more details.
The filter rules allow for flexible selection of which files to transfer (include) and which files to skip (exclude). The rules either directly specify include/exclude patterns or they specify a way to acquire more include/exclude patterns (e.g. to read them from a file).
As the list of files/directories to transfer is built, rsync checks each name to be transferred against the list of include/exclude patterns in turn, and the first matching pattern is acted on: if it is an exclude pattern, then that file is skipped; if it is an include pattern then that filename is not skipped; if no matching pattern is found, then the filename is not skipped.
Rsync builds an ordered list of filter rules as specified on the command-line. Filter rules have the following syntax:
You have your choice of using either short or long RULE names, as described below. If you use a short-named rule, the ‘,’ separating the RULE from the MODIFIERS is optional. The PATTERN or FILENAME that follows (when present) must come after either a single space or an underscore (_). Here are the available rule prefixes:
exclude, - specifies an exclude pattern. include, + specifies an include pattern. merge, . specifies a merge-file to read for more rules. dir-merge, : specifies a per-directory merge-file. hide, H specifies a pattern for hiding files from the transfer. show, S files that match the pattern are not hidden. protect, P specifies a pattern for protecting files from deletion. risk, R files that match the pattern are not protected. clear, ! clears the current include/exclude list (takes no arg)
When rules are being read from a file, empty lines are ignored, as are comment lines that start with a “#".
Note that the --include/--exclude command-line options do not allow the full range of rule parsing as described above -- they only allow the specification of include/exclude patterns plus a “!” token to clear the list (and the normal comment parsing when rules are read from a file). If a pattern does not begin with “- “ (dash, space) or “+ “ (plus, space), then the rule will be interpreted as if “+ “ (for an include option) or “- “ (for an exclude option) were prefixed to the string. A --filter option, on the other hand, must always contain either a short or long rule name at the start of the rule.
Note also that the --filter, --include, and --exclude options take one rule/pattern each. To add multiple ones, you can repeat the options on the command-line, use the merge-file syntax of the --filter option, or the --include-from/--exclude-from options.
You can include and exclude files by specifying patterns using the “+", “-", etc. filter rules (as introduced in the FILTER RULES section above). The include/exclude rules each specify a pattern that is matched against the names of the files that are going to be transferred. These patterns can take several forms:
Note that, when using the --recursive (-r) option (which is implied by -a), every subcomponent of every path is visited from the top down, so include/exclude patterns get applied recursively to each subcomponent’s full name (e.g. to include “/foo/bar/baz” the subcomponents “/foo” and “/foo/bar” must not be excluded). The exclude patterns actually shortcircuit the directory traversal stage when rsync finds the files to send. If a pattern excludes a particular parent directory, it can render a deeper include pattern ineffectual because rsync did not descend through that excluded section of the hierarchy. This is particularly important when using a trailing ‘*’ rule. For instance, this won’t work:
This fails because the parent directory “some” is excluded by the ‘*’ rule, so rsync never visits any of the files in the “some” or “some/path” directories. One solution is to ask for all directories in the hierarchy to be included by using a single rule: “+ */” (put it somewhere before the “- *” rule), and perhaps use the --prune-emptydirs option. Another solution is to add specific include rules for all the parent dirs that need to be visited. For instance, this set of rules works fine:
Here are some examples of exclude/include matching:
You can merge whole files into your filter rules by specifying either a merge (.) or a dir-merge (:) filter rule (as introduced in the FILTER RULES section above).
There are two kinds of merged files -- single-instance (’.’) and perdirectory (’:’). A single-instance merge file is read one time, and its rules are incorporated into the filter list in the place of the “." rule. For per-directory merge files, rsync will scan every directory that it traverses for the named file, merging its contents when the file exists into the current list of inherited rules. These per-directory rule files must be created on the sending side because it is the sending side that is being scanned for the available files to transfer. These rule files may also need to be transferred to the receiving side if you want them to affect what files don’t get deleted (see PER-DIRECTORY RULES AND DELETE below).
dir-merge,n- .non-inherited-per-dir-excludes :n- .non-inherited-per-dir-excludes
The following modifiers are accepted after a merge or dir-merge rule:
The following modifiers are accepted after a “+” or “-":
Per-directory rules are inherited in all subdirectories of the directory where the merge-file was found unless the ‘n’ modifier was used. Each subdirectory’s rules are prefixed to the inherited per-directory rules from its parents, which gives the newest rules a higher priority than the inherited rules. The entire set of dir-merge rules are grouped together in the spot where the merge-file was specified, so it is possible to override dir-merge rules via a rule that got specified earlier in the list of global rules. When the list-clearing rule ("!") is read from a per-directory file, it only clears the inherited rules for the current merge file.
Another way to prevent a single rule from a dir-merge file from being inherited is to anchor it with a leading slash. Anchored rules in a per-directory merge-file are relative to the merge-file’s directory, so a pattern “/foo” would only match the file “foo” in the directory where the dir-merge filter file was found.
Here’s an example filter file which you’d specify via --filter=". file":
This will merge the contents of the /home/user/.global-filter file at the start of the list and also turns the “.rules” filename into a perdirectory filter file. All rules read in prior to the start of the directory scan follow the global anchoring rules (i.e. a leading slash matches at the root of the transfer).
If a per-directory merge-file is specified with a path that is a parent directory of the first transfer directory, rsync will scan all the parent dirs from that starting point to the transfer directory for the indicated per-directory file. For instance, here is a common filter (see -F):
That rule tells rsync to scan for the file .rsync-filter in all directories from the root down through the parent directory of the transfer prior to the start of the normal directory scan of the file in the directories that are sent as a part of the transfer. (Note: for an rsync daemon, the root is always the same as the module’s “path".)
Some examples of this pre-scanning for per-directory files:
rsync -avF /src/path/ /dest/dir
rsync -av --filter=’: ../../.rsync-filter’ /src/path/ /dest/dir rsync -av --filter=’: .rsync-filter’ /src/path/ /dest/dir
The first two commands above will look for “.rsync-filter” in “/” and “/src” before the normal scan begins looking for the file in “/src/path” and its subdirectories. The last command avoids the parent-dir scan and only looks for the “.rsync-filter” files in each directory that is a part of the transfer.
If you want to include the contents of a “.cvsignore” in your patterns, you should use the rule “:C", which creates a dir-merge of the .cvsignore file, but parsed in a CVS-compatible manner. You can use this to affect where the --cvs-exclude (-C) option’s inclusion of the perdirectory .cvsignore file gets placed into your rules by putting the “:C” wherever you like in your filter rules. Without this, rsync would add the dir-merge rule for the .cvsignore file at the end of all your other rules (giving it a lower priority than your command-line rules). For example:
cat <<EOT | rsync -avC --filter=’. -’ a/ b
rsync -avC --include=foo.o -f :C --exclude=’*.old’ a/ b
Both of the above rsync commands are identical. Each one will merge all the per-directory .cvsignore rules in the middle of the list rather than at the end. This allows their dir-specific rules to supersede the rules that follow the :C instead of being subservient to all your rules. To affect the other CVS exclude rules (i.e. the default list of exclusions, the contents of $HOME/.cvsignore, and the value of $CVSIGNORE) you should omit the -C command-line option and instead insert a “-C” rule into your filter rules; e.g. “--filter=-C".
You can clear the current include/exclude list by using the “!” filter rule (as introduced in the FILTER RULES section above). The “current" list is either the global list of rules (if the rule is encountered while parsing the filter options) or a set of per-directory rules (which are inherited in their own sub-list, so a subdirectory can use this to clear out the parent’s rules).
As mentioned earlier, global include/exclude patterns are anchored at the “root of the transfer” (as opposed to per-directory patterns, which are anchored at the merge-file’s directory). If you think of the transfer as a subtree of names that are being sent from sender to receiver, the transfer-root is where the tree starts to be duplicated in the destination directory. This root governs where patterns that start with a / match.
Because the matching is relative to the transfer-root, changing the trailing slash on a source path or changing your use of the --relative option affects the path you need to use in your matching (in addition to changing how much of the file tree is duplicated on the destination host). The following examples demonstrate this.
Let’s say that we want to match two source files, one with an absolute path of “/home/me/foo/bar", and one with a path of “/home/you/bar/baz". Here is how the various command choices differ for a 2-source transfer:
Example cmd: rsync -a /home/me /home/you /dest
+/- pattern: /me/foo/bar
+/- pattern: /you/bar/baz
Target file: /dest/me/foo/bar
Target file: /dest/you/bar/baz
Example cmd: rsync -a /home/me/ /home/you/ /dest
Example cmd: rsync -a --relative /home/me/ /home/you /dest
Example cmd: cd /home; rsync -a --relative me/foo you/ /dest
The easiest way to see what name you should filter is to just look at the output when using --verbose and put a / in front of the name (use the --dry-run option if you’re not yet ready to copy any files).
Without a delete option, per-directory rules are only relevant on the sending side, so you can feel free to exclude the merge files themselves without affecting the transfer. To make this easy, the ‘e’ modifier adds this exclude for you, as seen in these two equivalent commands:
rsync -av --filter=’: .excl’ --exclude=.excl host:src/dir /dest rsync -av --filter=’:e .excl’ host:src/dir /dest
However, if you want to do a delete on the receiving side AND you want some files to be excluded from being deleted, you’ll need to be sure that the receiving side knows what files to exclude. The easiest way is to include the per-directory merge files in the transfer and use --delete-after, because this ensures that the receiving side gets all the same exclude rules as the sending side before it tries to delete anything:
rsync -avF --delete-after host:src/dir /dest
However, if the merge files are not a part of the transfer, you’ll need to either specify some global exclude rules (i.e. specified on the command line), or you’ll need to maintain your own per-directory merge files on the receiving side. An example of the first is this (assume that the remote .rules files exclude themselves):
rsync -av --filter=’: .rules’ --filter=’. /my/extra.rules’ --delete host:src/dir /dest
In the above example the extra.rules file can affect both sides of the transfer, but (on the sending side) the rules are subservient to the rules merged from the .rules files because they were specified after the per-directory merge rule.
In one final example, the remote side is excluding the .rsync-filter files from the transfer, but we want to use our own .rsync-filter files to control what gets deleted on the receiving side. To do this we must specifically exclude the per-directory merge files (so that they don’t get deleted) and then put rules into the local files to control what else should not get deleted. Like one of these commands:
rsync -av --filter=’:e /.rsync-filter’ --delete \
rsync -avFF --delete host:src/dir /dest
Batch mode can be used to apply the same set of updates to many identical systems. Suppose one has a tree which is replicated on a number of hosts. Now suppose some changes have been made to this source tree and those changes need to be propagated to the other hosts. In order to do this using batch mode, rsync is run with the write-batch option to apply the changes made to the source tree to one of the destination trees. The write-batch option causes the rsync client to store in a “batch file” all the information needed to repeat this operation against other, identical destination trees.
To apply the recorded changes to another destination tree, run rsync with the read-batch option, specifying the name of the same batch file, and the destination tree. Rsync updates the destination tree using the information stored in the batch file.
For convenience, one additional file is creating when the write-batch option is used. This file’s name is created by appending “.sh” to the batch filename. The .sh file contains a command-line suitable for updating a destination tree using that batch file. It can be executed using a Bourne (or Bourne-like) shell, optionally passing in an alternate destination tree pathname which is then used instead of the original path. This is useful when the destination tree path differs from the original destination tree path.
Generating the batch file once saves having to perform the file status, checksum, and data block generation more than once when updating multiple destination trees. Multicast transport protocols can be used to transfer the batch update files in parallel to many hosts at once, instead of sending the same data to every host individually.
$ rsync --write-batch=foo -a host:/source/dir/ /adest/dir/
$ scp foo* remote:
$ ssh remote ./foo.sh /bdest/dir/
$ rsync --write-batch=foo -a /source/dir/ /adest/dir/ $ ssh remote rsync --read-batch=- -a /bdest/dir/ <foo
In these examples, rsync is used to update /adest/dir/ from /source/dir/ and the information to repeat this operation is stored in “foo” and “foo.sh". The host “remote” is then updated with the batched data going into the directory /bdest/dir. The differences between the two examples reveals some of the flexibility you have in how you deal with batches:
The read-batch option expects the destination tree that it is updating to be identical to the destination tree that was used to create the batch update fileset. When a difference between the destination trees is encountered the update might be discarded with a warning (if the file appears to be up-to-date already) or the file-update may be attempted and then, if the file fails to verify, the update discarded with an error. This means that it should be safe to re-run a readbatch operation if the command got interrupted. If you wish to force the batched-update to always be attempted regardless of the file’s size and date, use the -I option (when reading the batch). If an error occurs, the destination tree will probably be in a partially updated state. In that case, rsync can be used in its regular (non-batch) mode of operation to fix up the destination tree.
The rsync version used on all destinations must be at least as new as the one used to generate the batch file. Rsync will die with an error if the protocol version in the batch file is too new for the batchreading rsync to handle. See also the --protocol option for a way to have the creating rsync generate a batch file that an older rsync can understand. (Note that batch files changed format in version 2.6.3, so mixing versions older than that with newer versions will not work.)
When reading a batch file, rsync will force the value of certain options to match the data in the batch file if you didn’t set them to the same as the batch-writing command. Other options can (and should) be changed. For instance --write-batch changes to --read-batch, --files-from is dropped, and the --filter/--include/--exclude options are not needed unless one of the --delete options is specified.
The code that creates the BATCH.sh file transforms any filter/include/exclude options into a single list that is appended as a “here” document to the shell script file. An advanced user can use this to modify the exclude list if a change in what gets deleted by --delete is desired. A normal user can ignore this detail and just use the shell script as an easy way to run the appropriate --read-batch command for the batched data.
The original batch mode in rsync was based on “rsync+", but the latest version uses a new implementation.
Three basic behaviors are possible when rsync encounters a symbolic link in the source directory.
By default, symbolic links are not transferred at all. A message “skipping non-regular” file is emitted for any symlinks that exist.
If --links is specified, then symlinks are recreated with the same target on the destination. Note that --archive implies --links.
If --copy-links is specified, then symlinks are “collapsed” by copying their referent, rather than the symlink.
rsync also distinguishes “safe” and “unsafe” symbolic links. An example where this might be used is a web site mirror that wishes ensure the rsync module they copy does not include symbolic links to /etc/passwd in the public section of the site. Using --copy-unsafelinks will cause any links to be copied as the file they point to on the destination. Using --safe-links will cause unsafe links to be omitted altogether. (Note that you must specify --links for --safelinks to have any effect.)
Symbolic links are considered unsafe if they are absolute symlinks (start with /), empty, or if they contain enough “..” components to ascend from the directory being copied.
Here’s a summary of how the symlink options are interpreted. The list is in order of precedence, so if your combination of options isn’t mentioned, use the first line that is a complete subset of your options:
rsync occasionally produces error messages that may seem a little cryptic. The one that seems to cause the most confusion is “protocol version mismatch -- is your shell clean?".
This message is usually caused by your startup scripts or remote shell facility producing unwanted garbage on the stream that rsync is using for its transport. The way to diagnose this problem is to run your remote shell like this:
ssh remotehost /bin/true > out.dat
then look at out.dat. If everything is working correctly then out.dat should be a zero length file. If you are getting the above error from rsync then you will probably find that out.dat contains some text or data. Look at the contents and try to work out what is producing it. The most common cause is incorrectly configured shell startup scripts (such as .cshrc or .profile) that contain output statements for noninteractive logins.
If you are having trouble debugging filter patterns, then try specifying the -vv option. At this level of verbosity rsync will show why each individual file is included or excluded.
The CVSIGNORE environment variable supplements any ignore patterns in .cvsignore files. See the --cvs-exclude option for more details.
USER or LOGNAME
The USER or LOGNAME environment variables are used to determine the default username sent to an rsync daemon. If neither is set, the username defaults to “nobody".
/etc/rsyncd.conf or rsyncd.conf
times are transferred as *nix time_t values
When transferring to FAT filesystems rsync may re-sync unmodified files. See the comments on the --modify-window option.
file permissions, devices, etc. are transferred as native numerical values
see also the comments on the --delete option
Please report bugs! See the website at http://rsync.samba.org/
This man page is current for version 2.6.8 of rsync.
rsync is distributed under the GNU public license. See the file COPYING for details.
A WEB site is available at http://rsync.samba.org/. The site includes an FAQ-O-Matic which may cover questions unanswered by this manual page.
The primary ftp site for rsync is ftp://rsync.samba.org/pub/rsync.
We would be delighted to hear from you if you like this program.
This program uses the excellent zlib compression library written by Jean-loup Gailly and Mark Adler.
Thanks to Richard Brent, Brendan Mackay, Bill Waite, Stephen Rothwell and David Bell for helpful suggestions, patches and testing of rsync. I’ve probably missed some people, my apologies if I have.
Especial thanks also to: David Dykstra, Jos Backus, Sebastian Krahmer, Martin Pool, Wayne Davison, J.W. Schultz.
rsync was originally written by Andrew Tridgell and Paul Mackerras. Many people have later contributed to it.
Mailing lists for support and development are available at http://lists.samba.org
Table of Contents