Rsync is an open source file synchronization program. Designed to maintain a mirror of a directory, it is a very advanced tool that can be used to make backups, or to copy files to another disk or host in the event of a hardware or software catastrophe.
The advantages of rsync over alternate transfer methods include:
Rsync has few drawbacks:
Rsync is provided with SystemRescueCd and this documentation is intended to provide useful information and examples for use during the recovery process. After understanding these you will find rsync useful to increase system reliability during normal operations. See “Using rsync regularly to minimize the impact of a disaster”
rsync can be run as server (when started with –daemon) or as client to make copies of files on a local machine or across the network to another host. The recommended options include:
--progressdisplays activity for monitoring progress
--archivepreserves basic attributes (permissions, ownership, type such as symbolic links.
--xattrspreserve eXtended attributes (same as
--aclspreserve Access Control Lists (same as
--hard-linkspreserve hard links (don’t treat them as files) (same as
--compressin flight data ( no effect on destination) (same as
Only one rsync process is involved for local backups. Here is an example which preserves all attributes (including extended ones) and ACLs:
rsync --archive --xattrs --acls --hard-links --progress --compress \ /home/mydir/data/ /backups/data-20080810/
If rsync is installed as a daemon (listening on port tcp/873) on example.com, the client running at myhost can either pull files from the daemon or push files to the daemon.
Here is an example having the client can push a directory to a remote host (preserving attributes, ACLs and hard links (using single letter options):
rsync -aXAHz --progress /home/mydata/ example.com::mybackups/data-20080810/
The client could also download the files from the remote hosts:
rsync -aXAHz --progress example.com::mybackups/data-20080810/ /home/mydata/
Notice two colons between the remote host and the path when the rsync protocol is used.
If the port tcp/873 is blocked (by a firewall) or you need the connection to be
encrypted then you can use rsync through ssh by specifying
the client side.
Here is an example where the clients send to the remote host:
rsync -aAXHz --progress -e ssh /home/mydata/ remote.com:/backups/data-20080810/
rsync -aAXHz --progress -e ssh remote.com:/backups/data-20080810/ /home/mydata/
Observe only one colon between the remote host address and the remote path when rsync connects over ssh.
When recovering large amount of data, either large files, or thousands of small
file, using scp, ftp or http lacks a means of resuming a interrupted transfer.
Manually restarting the transfer is awkward. If the process was interrupted at
99% of a large file, the transfer must be restarted at the beginning, all the
data transmitted again and may mean the transfer never completes. Even with
wget -c a corrupt file may result.
This is also important in those cases where the hard disk only operates while cool. rsync can handle this well by saving files as long as the disk is operating. After the disk heats up and becomes un-responsive rsync will abort. Power off the system and wait an hour or two for the disk to cool down. Power up and rsync can continue saving files where it left off.
Only the files which are not current at the destination will be transferred. By first comparing the source and the destination files only the ones which are different will be transfered. This facilitates the continuation after an interruption and reduces elapsed time and reduces bandwidth requirements for remote transfers.
Rsync is able to resume transfers of files if the connection is lost by
--inplace. This is especially
important with large files.
--partial causes partially transferred files to be retained. By
default rsync removes a partial file when interrupted. If the connection is
lost rsync will stop. When the connection is reestablished and rsync resumes
data will need to be transferred again as rsync will start from the beginning
of the file.
--inplace causes rsync to use original file name at the
destination. By default rsync transfers data to files with a temp name during
the transfer and renames the files once the file is complete. Without
--inplace, on restart rsync will create new temp files and must
restart the transfer from the beginning of the file.
--update must NOT used with
--inplace as the
partial destination file would be considered the same as than the original file
Rsync uses a very efficient algorithm to compare the source file with the destination file and only transfers the different parts of the file. To transfer several versions of the same files on a regular basis, copy the old version (that you have already transferred) to the new destination file, and rsync will skip all the common parts.
With these options, rsync can be interrupted then resumed. Data which has been transferred is preserved.
Here is an example of good command to copy a directory with large files to a remote host:
rsync --archive --partial --inplace --progress --compress \ /home/bigfiles/ MYMIRROR::mybackups/bigfiles/
Check for updates to rsync at http://rsync.samba.org/
rsync has a lot of options. Here are just a few which apply to recovering a failed system.
It is a very important option. Preserves files attributes (permissions, times
and type). As a result, a symbolic link will be copied as a link, without
--archive the contents of the source file are copied in the
destination. Hard links, extended attributes (xattr), and ACLs (Access Control
Lists) are not preserved with this option, see
This option will show the progress on the current transfer
This option shows more details and can be used multiple times to increase the logging
Process only one filesystem. Important when processing a filesystem
/ ) with mounted volumes. Default processes filesystems with mount
points within the source specification.
Rsync can compress the data that are transferred to reduce network activity. The destination file will be the same as the original. Use it for remote transfers when synching files which are uncompressed and will have a large compression ratio (eg: large text files, CD images or raw partition images). This option is not efficient on files which are already compressed: zip, gz, bz2, taz, jpeg, pdf, … This will cause a significant increase in the CPU usage on both sourcing and destination systems.
Use these options to transfer large files and insure that the transfer will resume at a restart point in case of connection failure. See the sections about Transferring large files for more details.
Use this option to exclude files or directories from the transfer. For instance temporary files or cache directories.
When synchronizing to an existing destination, files in the destination directory will be removed if they are not in the source directory (The files have been deleted). By default files no longer in the source remain in the destination. Consider the scenario in which files are versioned by date and are deleted as a new one is created. The default will not remove old versions and they will accumulate. This will cause the destination to not be a “mirror” of the source and the space required for the destination will increase with each run. Processing a single run with this option will delete all accumulated files.
Defines this run to be an update of an existing destination. Use this option if
files have been modified in the destination directory. For instance, if you
are migrating data from an old server to a new one, and if people have already
started working on the new server. Use
--update to prevent overwriting
changes with an older version.
This option is useful to move data to the destination. By default, rsync makes a
copy leaving the original intact. With
--remove-source-files, the source
files to be deleted if the transfer was successful.
By default rsync uses the file modification-time and size to to decide whether
or not the file needs to be transferred. A file that exists in the destination
directory with the same name, date and size is considered to be the same. This
is accurate in nearly all cases. The comparison is very quick because only the
file attributes are read. In rare cases this is incorrect. There are some
utilities which can modify a file and retain the original modification time. If
the modification does not change the size of the file (very rare) rsync will
erroneously consider the source and destination files to be identical and not
transfer the file. Specify
--checksum to force rsync to calculate a
checksum of the files and compare this checksum to determine if the files are
the same or not. Be warned that this will add significant CPU time and I/O
activity at both the source and destination since the entire contents of both
files must be read to generate the checksum.
The return status must be checked to determine whether or not the transfer was successful. When rsync returns 0, it means that the transfer was successful. Any other value indicates an error.
Some errors which may indicate a recoverable problem include:
The rsync client will only transfer data to an rsync server.
The configuration default location is either
/etc/rsync/rsyncd.conf. After changing the configuration, send a HUP
kill -HUP <pid-of-rsync>.
Several options are available to restrict connections:
Here is an example of basic secured configuration:
# ======================/etc/rsyncd.conf====================== pid file = /var/run/rsyncd.pid read only = yes uid = root gid = root [share1] path = /mnt/share1 read only = yes hosts allow = myhost, 10.88.45.0/24 [backups] path = /var/tmp/catalyst/tmp read only = no hosts allow = mybiggie, 10.88.45.0/24 [rootfs] path = / read only = yes hosts allow = myhost, 10.88.45.0/24 [upload] path = /upload read only = no hosts allow = 172.16.0.0/16
The rsync daemon can be run on an MS Windows system by using cygwin which provides a Linux environment. Minimal information is provided here to get you started. It is not emulated programs: it is just the software we use on linux has been compiled to run on windows, so it is a windows executable which is speaking to the windows kernel directly
To install cygwin, run the setup.exe program found on the cygwin website
The hard disks as seen as
/cygdrive/d/, … .
Click on the cygwin icon to run a bash shell.
You can also install rsync as a daemon on cygwin. To install services in cygwin,
you can use a special program named
cygrunsrv.exe. It installs a cygwin
service as a normal windows service, so that it can be automatically started at
windows boot time. That way you don’t have to start the daemon by hand.
Here is the command to use:
cygrunsrv.exe -I "cygrsyncd" -p /usr/bin/rsync.exe -a "--config=/etc/rsyncd.conf --daemon --no-detach"
Then, start the new service run
services.msc and start the service named “cygrsyncd”
The previous sections of this document focus on the features and operation of rsync to help recover a system.
Here we present additional information we thought would be useful.
rsync is a utility that maintains a mirror of a filesystem. Updated regularly, the mirror can be a major asset for recovering a failed system.
Here are some addition notes for using rsync:
The rsync daemon can be installed in a dedicated listen mode (tcp port 873). Having xinetd configured to listen on the behalf of rsync uses less memory and is more secure. It also means rsync is started and the configuration file read on each connection so changes will effected on the next connection.
rsync traps intr so pressing ^C stops when the current file is completed. Use the less common quit ^\ to have it quit now.