NAME

histbackup - backup a directory using an incremental technique

SYNOPSIS

histbackup [options...] [host:]source backupdir [rsyncopts...]

histbackup [options...] --no-rsync parent

DESCRIPTION

histbackup makes incremental backups of a directory tree in a set of directories named by dates, or date-time if the date already exists. The new backup is prepopulated by hardlinks of each file from the previous backup and then updated with rsync(1). This arranges that the only new content of each new backup is fresh copies of the changed files. In this way each backup directory is a full copy of the source directory but the disc space cost is that of an incremental backup. Because it uses rsync, it is possible efficiently to backup remote directories in this manner. The companion script histbackup-prune keeps the growth under control.

SETUP AND USE

Before deployment, decide who will own and do the backups. Do all the local (backup end) setup as that user. If you're not a sysadmin then that user is probably yourself.

Generally, setup is as follows:

Create the backup area.

Decide where the backup data will be. Make an empty directory to hold it:

        mkdir /path/to/backup/area

Arrange access to the source data

If the data to backup are local to this machine you merely need read access to it. If the data are remote you will need ssh(1) access to the remote host as a user with read access to the data there. Test it:

        ssh user@remote ls -l /path/to/source/directory

Perform a test backup

Histbackup does the right thing when the backup area is empty so just:

        histbackup -x user@remote:/path/to/source/directory /path/to/backup/area

Schedule your backups

If these backups are to happen regularly a crontab(5) entry should be made to do so. Use ``crontab -e'' to make this. Example:

        MAILTO=backup-admin@yoursite
        PATH=$PATH:/opt/css/bin
        0 * * * * histbackup -x user@remote:/path/to/source/directory /path/to/backup/area

That will run a histbackup every day at midnight, delivering the output of the run by email to backup-admin@yoursite.

Note that cron's environment is very spartan (it does not source /etc/profile or ~/.profile) and so you may need to arrange that the CSS package's environment is present. Also, because cron is a batch situation, your ssh access will almost certainly require a special purpose passphraseless key to access the remote host.

OPTIONS

--no-delete: Suppress passing the --delete or --delete-excluded option to rsync.
--no-rsync: Skip the rsync step; just create the new link tree.
--delete: Passed to rsync.
--delete-excluded: Passed to rsync. Also implies --delete.
--no-hbinclude: Do not honour the .hbinclude file in the backup directory.
--rsync-path rrsync: Passed to rsync. Specifies the pathname to the remote rsync executable.
--symlink name: After the backup is done, make a symbolic link named name pointing at the latest backup directory. This is useful so that other programs may trivially reference the latest backup without having to figure out its name. Eg: --symlink LATEST.
-v, --verbose: Turn on verbose mode; also passes -v to rsync.
-x: Trace execution of important commands (rsync, linktree etc).

CUSTOMISATION FILE

If the file backupdir/.hbinclude is present the option ``--include-from=backupdir/.hbinclude will be added to the rsync command invocation. In this way the content to be backed up may be tuned. See rsync(1).

AUTHOR

Cameron Simpson <cs@zip.com.au> 03apr2000