BACKUP(1) BACKUP(1) NAME backup - backup and recover files SYNOPSIS backup recover [ option ... ] file ... backup grep [ option ... ] pattern ... backup fetch [ option ... ] [ file ... ] backup stats [ option ... ] backup backup [ file ... ] backup munge backup mount [ option ... ] mountpt DESCRIPTION The backup programs save and restore archival copies of files in an optical disk store on a central system (see backup(5)). Backup occurs automatically daily (see backup(8)) and upon specific request via backup backup. Backup grep shows backup copy names for specific files, and backup fetch restores data from specific backup copies. Backup recover is a combination of these two; it fetches the most recent copy. All the backup programs describe their options when presented with a bad option such as -?. Backup recover retrieves files by name. The names should be full pathnames rooted at /n/; if not, backup tries to guess names that begin with /n/. Directories should be recovered before their contents. Regular files that are linked together will stay linked if they are recovered together. The options for recover are: -o dir The argument is restored as an entry in the directory dir. -v Verbose (enforced). -F Restore directories as files containing a null- terminated list of element names. -r Recursively recover any subdirectories. -d Create any missing intermediate directories. -Dold=new Replace the prefix old of the original filename with new to form the new output filename. -m The names are backup copy names, as determined from backup grep, not original filenames. -fdevice BACKUP(1) BACKUP(1) Use device rather than /dev/worm0 for the WORM. Device may be on another machine: machine!device. An initial w implies a WORM device; a j implies a jukebox. A numeric device means /dev/wormdevice. -e Cause the worm fetch server on the backup system to terminate gracefully. -i Append .n to the output name for each file where n is an increasing integer. This is useful for recovering multiple copies of the same file. A diagnostic like need disk backup2a means you need to mount the A side of the cartridge labeled backup2. Backup grep searches for names of backed up files that match the strings patterns. If the pattern is a literal (no -e) that looks like a filename, it reports the filename cate- nated with // and the time of the most recent backup copy. If the pattern is a literal that looks like the output under option -d, it reports the name of the corresponding backup copy. The options are: -d Print file change times (ctime, see stat(2)) as inte- gers rather than as dates. -e Interpret patterns as regular expressions given in the notation of regexp(3). Warning: this option can execute extremely slowly; it is almost always better to use gre(1) on on the backup machine; see backup(5). -a Print all names in the database. -V Treat pattern as a literal filename and list all ver- sions of the file. -<n Only list entries with a date less than or equal to n. If n is not a simple integer date, it is interpreted as by timec(3). ->n Only list entries with a date greater than or equal to n. -D Print the most recent entry for every file name start- ing with pattern, taking into account any cutoff date, but turning off option -e. Backup fetch takes from its arguments or from standard input backup copy names as reported by backup grep (such as v2345/987) and restores the corresponding files. It accepts the same options as backup recover except -m; -v is really optional. Irrelevant prefixes are stripped from backup copy names. Thus the output of the backup grep command can be used directly. Backup stats provides statistics about the files backed up. By default, it looks for all systems and all users and gives a grand total. The options are -i Give information per system or user rather than a BACKUP(1) BACKUP(1) total. -s systems -u users With option -i, restrict the total to the systems or users named in comma-separated lists. The name `*' expands to all systems or all users. -d Print average number of files and bytes for the last 1 day, 7 days and 30 days. Backup backup backs up files. If no file names are given, they are taken from standard input. File names are inter- preted as in backup recover. The files are safely on the backup system when the command exits but will normally take a day to get into the backup database. Backup munge causes the backup system to process any received files. When this terminates (assuming no errors), the files have been put onto backup media and have been absorbed into the database. Backup mount is an experimental way to access backed up files. The specified part of the backup files (set by -Droot or / by default) is mounted at mountpt. There is one option -d date Make the mounted hierarchy reflect the state at the given date. The mounting can be reversed with umount; see mount(8). EXAMPLES backup stats -i -s '*' Get totals for all systems. /n/bowell/etc/passwd\`` backup fetch `backup grep -d \`backup grep -d What backup recover does for you. backup recover /n/coma/usr/rob/fortunes cd /n/coma/usr/rob; backup recover fortunes Two ways to get the latest available copy of /n/coma/usr/rob/fortunes. backup grep -V /n/coma/usr/rob/fortunes List all available copies of /n/coma/usr/rob/fortunes with their dates. backup recover -m -o /tmp /n/wild/usr/backup/v/v919/678 backup recover -m -o /tmp v919/678 Two ways to recover a specific backup copy and place the result in /tmp. /n/wild/usr/backup/v/v919/678 is the name of the backup copy; the file will be restored BACKUP(1) BACKUP(1) to its home machine, not to wild. backup grep -V /n/coma/usr/rob/fortunes | backup fetch -i - o . Recover all the versions of the fortunes file into fortunes.1, fortunes.2, ... in the current directory. FILES home of all datafiles and executables (on client machines) SEE ALSO worm(8), backup(5), backup(8) BUGS Recovery via symbolic links may not work; use the non-linked pathname.