Skip to main content

dump command man page

DUMP(8)  System management commands       DUMP(8)

NAME
       dump - ext2/3 filesystem backup
SYNOPSIS
       dump  [-level#] [-ackMnqSuv] [-A file] [-B records] [-b blocksize] [-d
       density] [-D file] [-e inode numbers] [-E file] [-f file]  [-F  script]
       [-h  level]  [-I nr errors] [-jcompression level] [-L label] [-Q file]
       [-s feet] [-T date] [-y] [-zcompression level] files-to-dump

       dump [-W | -w]

DESCRIPTION
       Dump examines files on an ext2/3 filesystem and determines which files
       need to be backed up. These files are copied to the given disk, tape or
       other storage medium for safe keeping (see  the -f  option  below  for
       doing  remote backups). A dump that is larger than the output medium is
       broken into multiple volumes. On most media the size is determined  by
       writing until an end-of-media indication is returned.

       On  media  that cannot reliably return an end-of-media indication (such
       as some cartridge tape drives), each volume is of  a  fixed  size;  the
       actual  size  is determined  by specifying cartridge media, or via the
       tape size, density and/or block count options below.  By default,  the
       same output file name is used for each volume after prompting the oper-
       ator to change media.

       files-to-dump is either a mountpoint of a filesystem or a list of files
       and  directories to  be backed up as a subset of a filesystem. In the
       former case, either the path to a mounted filesystem or the  device  of
       an  unmounted  filesystem  can  be  used.  In  the latter case, certain
       restrictions are placed on the backup: -u is not allowed, the only dump
       level  that  is supported  is 0 and all the files and directories must
       reside on the same filesystem.

OPTIONS
       The following options are supported by dump:

       -level#
     The dump level (any integer). A level 0, full backup, guarantees
     the  entire  file system is copied (but see also the -h option
     below). A level number above 0, incremental backup,  tells  dump
     to copy all files new or modified since the last dump of a lower
     level. The default level is 0. Historically only levels 0 to  9
     were  usable  in dump,  this  version is able to understand any
     integer as a dump level.

       -a     “auto-size”. Bypass all  tape  length  calculations,  and write
     until  an end-of-media indication is returned.  This works best
     for most modern tape drives, and is the  default. Use  of  this
     option is particularly recommended when appending to an existing
     tape, or using a tape drive with hardware compression (where you
     can never be sure about the compression ratio).

       -A archive_file
     Archive  a  dump table-of-contents in the specified archive_file
     to be used by restore(8) to determine whether a file is  in  the
     dump file that is being restored.

       -b blocksize
     The  number  of kilobytes per dump record. The default blocksize
     is 10, unless the -d option has been used to specify a tape den-
     sity  of 6250BPI or more, in which case the default blocksize is
     32. Th maximal value is 1024.  Note however that, since  the  IO
     system slices all requests into chunks of MAXBSIZE (which can be
     as low as 64kB), you can experience problems  with  dump(8)  and
     restore(8)  when using a higher value, depending on your kernel
     and/or libC versions.

       -B records
     The number of 1 kB blocks per volume. Not normally required,  as
     dump  can detect end-of-media. When  the  specified  size  is
     reached, dump waits for you to change the volume.  This option
     overrides the calculation of tape size based on length and den-
     sity. If compression is on this limits  the  size of  the  com-
     pressed  output  per  volume.  Multiple values may be given as a
     single argument separated by commas.  Each value will  be  used
     for  one dump  volume in the order listed; if dump creates more
     volumes than the number of values given, the last value will  be
     used  for the  remaining volumes. This is useful for filling up
     already partially filled media (and then continuing  with  full
     size volumes on empty media) or mixing media of different sizes.

       -c     Change the defaults for use with a cartridge tape drive, with  a
     density  of  8000 bpi,  and a length of 1700 feet. Specifying a
     cartridge drive overrides the end-of-media detection.

       -d density
     Set tape density to density.  The default is 1600BPI. Specifying
     a tape density overrides the end-of-media detection.

       -D file
     Set  the path name of the file storing the information about the
     previous full and incremental dumps.  The default  location  is
     /etc/dumpdates.

       -e inodes
     Exclude  inodes  from  the dump. The inodes parameter is a comma
     separated list of inode numbers (you can use stat(1) to find the
     inode number for a file or directory).

       -E file
     Read  list  of inodes to be excluded from the dump from the text
     file file.  The file file should be an ordinary file  containing
     inode numbers separated by newlines.

       -f file
     Write the backup to file; file may be a special device file like
     /dev/st0 (a tape drive), /dev/rsd1c (a floppy  disk  drive),  an
     ordinary file,  or - (the standard output). Multiple file names
     may be given as a single argument separated by commas. Each file
     will  be used  for  one dump volume in the order listed; if the
     dump requires more volumes than the number of names  given,  the
     last file name will used for all remaining volumes after prompt-
     ing for media changes. If the name of the file is of  the  form
     host:file or user@host:file dump writes to the named file on the
     remote host (which should already exist, dump doesn’t  create  a
     new  remote  file)  using rmt(8).  The default path name of the
     remote rmt(8) program is /etc/rmt; this can be overridden by the
     environment variable RMT.

       -F script
     Run  script  at  the end of each tape (except for the last one).
     The device name and the current volume number are passed on  the
     command  line.  The script must return 0 if dump should continue
     without asking the user to change the tape,  1  if  dump should
     continue but  ask  the  user to change the tape. Any other exit
     code will cause  dump  to abort. For  security reasons,  dump
     reverts  back  to the real user ID and the real group ID before
     running the script.

       -h level
     Honor the user nodump flag UF_NODUMP only for dumps at or above
     the  given  level.  The default honor level is 1, so that incre-
     mental backups omit such files but full backups retain them.

       -I nr errors
     By default, dump will ignore the first 32 read  errors  on  the
     file  system  before  asking  for operator intervention. You can
     change this using this flag to any value. This  is  useful  when
     running  dump  on an active filesystem where read errors simply
     indicate an  inconsistency  between  the mapping  and  dumping
     passes.

     A value of 0 means that all read errors will be ignored.

       -jcompression level
     Compress every  block  to  be  written  on the tape using bzlib
     library. This option will work only when dumping to  a  file  or
     pipe  or, when  dumping to  a tape drive, if the tape drive is
     capable of writing variable length  blocks.  You will  need  at
     least  the  0.4b24  version  of restore in order to extract com-
     pressed tapes. Tapes written using compression will not be  com-
     patible with the BSD tape format. The (optional) parameter spec-
     ifies the compression level bzlib will use. The default compres-
     sion  level  is 2. If the optional parameter is specified, there
     should be no white space between the  option  letter  and  the
     parameter.

       -k     Use  Kerberos  authentication  to talk  to remote tape servers.
     (Only available if this option was enabled when  dump  was  com-
     piled.)

       -L label
     The  user-supplied  text string label  is placed into the dump
     header, where tools like restore(8) and file(8) can  access  it.
     Note that this label is limited to be at most LBLSIZE (currently
     16) characters, which must include the terminating \0.

       -m     If this flag is specified, dump will  optimise  the  output  for
     inodes  having been changed but not modified since the last dump
     (’changed’ and ’modified’ have the meaning defined in stat(2) ).
     For  those  inodes, dump will save only the metadata, instead of
     saving the entire inode contents.   Inodes  which  are either
     directories  or have been modified since the last dump are saved
     in a regular way. Uses of this flag must be consistent,  meaning
     that either every dump in an incremental dump set have the flag,
     or no one has it.

     If you use this option, be aware that many programs that unpack
     files  from archives (e.g. tar, rpm, unzip, dpkg) may set files’
     mtimes to dates in the past.  Files installed in this  way  may
     not be dumped correctly using "dump -m" if the modified mtime is
     earlier than the previous level dump.

     Tapes written using such ’metadata only’ inodes will not be com-
     patible with the BSD tape format or older versions of restore.

       -M     Enable  the  multi-volume feature. The name specified with f is
     treated as a prefix and dump writes in sequence to  <prefix>001,
     <prefix>002  etc. This can be useful when dumping to files on an
     ext2 partition, in order to bypass the 2GB file size limitation.

       -n     Whenever dump requires operator attention, notify all operators
     in the group operator by means similar to a wall(1).

       -q     Make dump abort immediately  whenever  operator attention  is
     required, without  prompting  in case  of  write  errors, tape
     changes etc.

       -Q file
     Enable the Quick File Access support. Tape  positions  for  each
     inode are stored into the file file which is used by restore (if
     called with parameter -Q and the filename) to directly  position
     the tape at the file restore is currently working on. This saves
     hours when restoring single files from large backups, saves  the
     tapes and the drive’s head.

     It is recommended to set up the st driver to return logical tape
     positions rather than physical before calling dump/restore  with
     parameter -Q.  Since not all tape devices support physical tape
     positions those tape devices return an error during dump/restore
     when  the st  driver  is set  to the default physical setting.
     Please see the st(4) man page, option MTSETDRVBUFFER  ,  or  the
     mt(1)  man page, on how to set the driver to return logical tape
     positions.

     Before calling restore with parameter -Q, always make  sure  the
     st  driver  is set to return the same type of tape position used
     during the call to dump. Otherwise restore may be confused.

     This option can be used when dumping to local tapes (see above)
     or to local files.

       -s feet
     Attempt  to  calculate the amount of tape needed at a particular
     density. If this amount is exceeded,  dump  prompts  for a  new
     tape. It is recommended to be a bit conservative on this option.
     The default tape length is 2300 feet. Specifying the  tape  size
     overrides end-of-media detection.

       -S     Size  estimate.  Determine the amount of space that is needed to
     perform the dump without actually doing it,  and  display  the
     estimated number of  bytes  it will take. This is useful with
     incremental dumps to determine how many volumes of media will be
     needed.

       -T date
     Use the specified date as the starting time for the dump instead
     of the time determined from looking  in  /etc/dumpdates  .   The
     format  of  date is the same as that of ctime(3) followed by an
     rfc822 timezone specification: either a plus or minus sign  fol-
     lowed  by two digits for the number of hours and two digits for
     the minutes.  For example, -0800 for eight hours west of Green-
     wich  or +0230 for two hours and a half east of Greenwich. This
     timezone offset takes into account  daylight  savings  time  (if
     applicable  to  the timezone): UTC offsets when daylight savings
     time is in effect will be different than offsets when  daylight
     savings time is not in effect. For backward compatibility, if no
     timezone is specified, a local time is assumed.  This option  is
     useful  for automated dump scripts that wish to dump over a spe-
     cific period of time. The -T option is mutually  exclusive  from
     the -u option.

       -u     Update the file /etc/dumpdates after a successful dump. The for-
     mat of /etc/dumpdates is readable by people, consisting  of  one
     free  format  record  per line: filesystem name, increment level
     and ctime(3) format dump date  followed  by  a  rfc822  timezone
     specification  (see  the -u option for details). If no timezone
     offset is specified, times are interpreted  as  local.  Whenever
     the  file is written, all dates in the file are converted to the
     local time zone, without changing the UTC times. There  may  be
     only one entry per filesystem at each level. The file /etc/dump-
     dates may be edited to change any of the fields, if necessary.

       -v     The -v (verbose) makes dump to  print  extra  information which
     could be helpful in debug sessions.

       -W     Dump  tells  the operator  what file systems need to be dumped.
     This information is gleaned from the  files  /etc/dumpdates  and
     /etc/fstab.   The -W  option  causes dump to print out, for all
     file systems in /etc/dumpdates , and recognized file systems  in
     /etc/mtab and /etc/fstab.  the most recent dump date and level,
     and highlights those that should be dumped. If the -W option  is
     set,  all other options are ignored, and dump exits immediately.

       -w     Is like -W, but prints only recognized filesystems in  /etc/mtab
     and /etc/fstab which need to be dumped.

       -y     Compress every  block  to  be written to the tape using the lzo
     library. This doesn’t compress as well as the zlib library  but
     it’s  much faster.  This option will work only when dumping to a
     file or pipe or, when dumping to a tape drive, if the tape drive
     is  capable of writing variable length blocks.  You will need at
     least the 0.4b34 version of restore in  order  to extract  com-
     pressed  tapes. Tapes written using compression will not be com-
     patible with the BSD tape format.

       -zcompression level
     Compress every block to  be  written  on the  tape  using  zlib
     library. This  option  will work only when dumping to a file or
     pipe or, when dumping to a tape drive,  if  the  tape  drive  is
     capable  of  writing  variable  length  blocks. You will need at
     least the 0.4b22 version of restore in  order  to extract  com-
     pressed  tapes. Tapes written using compression will not be com-
     patible with the BSD tape format. The (optional) parameter spec-
     ifies  the compression level zlib will use. The default compres-
     sion level is 2. If the optional parameter is  specified, there
     should  be  no  white  space  between  the option letter and the
     parameter.

       Dump requires operator intervention on these conditions: end  of tape,
       end  of dump, tape write error, tape open error or disk read error (if
       there is more than a threshold of nr errors). In addition  to  alerting
       all  operators  implied by the -n key, dump interacts with the operator
       on dump’s control terminal at times when dump can no longer proceed, or
       if  something  is  grossly  wrong.  All questions  dump poses must be
       answered by typing “yes” or “no”, appropriately.

       Since making a dump involves a lot of time and effort for  full dumps,
       dump  checkpoints  itself  at the start of each tape volume. If writing
       that volume fails for some reason, dump will, with operator permission,
       restart itself from the checkpoint after the old tape has been rewound
       and removed, and a new tape has been mounted.

       Dump tells the operator what is going on at periodic intervals, includ-
       ing  usually low estimates of the number of blocks to write, the number
       of tapes it will take, the time to completion, and the time to the tape
       change. The  output  is verbose, so that others know that the terminal
       controlling dump is busy, and will be for some time.

       In the event of a catastrophic disk event, the time required to restore
       all  the necessary backup tapes or files to disk can be kept to a mini-
       mum by staggering the incremental dumps. An efficient method  of stag-
       gering incremental dumps to minimize the number of tapes follows:

       —      Always start with a level 0 backup, for example:
    /sbin/dump -0u -f /dev/st0 /usr/src

     This  should  be done at set intervals, say once a month or once
     every two months, and on a set of fresh tapes that is saved for-
     ever.

       —      After  a level  0,  dumps of active file systems are taken on a
     daily basis, using a modified Tower  of  Hanoi  algorithm,  with
     this sequence of dump levels:
    3 2 5 4 7 6 9 8 9 9 ...

     For the daily dumps, it should be possible to use a fixed number
     of tapes for each day, used on a weekly basis. Each  week,  a
     level  1 dump  is  taken,  and the daily Hanoi sequence repeats
     beginning with 3. For weekly dumps, another fixed set  of tapes
     per dumped file system is used, also on a cyclical basis.

       After  several  months  or  so, the  daily and weekly tapes should get
       rotated out of the dump cycle and fresh tapes brought in.

       (The 4.3BSD option syntax is implemented for backward compatibility but
       is not documented here.)

ENVIRONMENT
       TAPE   If  no  -f option was specified, dump will use the device speci-
     fied via TAPE as the dump device. TAPE may be of the form tape-
     name, host:tapename, or user@host:tapename.

       RMT    The environment variable RMT will be used to determine the path-
     name of the remote rmt(8) program.

       RSH    Dump uses the contents of this variable to determine the name of
     the  remote shell command to use when doing remote backups (rsh,
     ssh etc.). If this variable is not set, rcmd(3)  will  be used,
     but only root will be able to do remote backups.

FILES
       /dev/st0
     default tape unit to dump to

       /etc/dumpdates
     dump date records

       /etc/fstab
     dump table: file systems and frequency

       /etc/mtab
     dump table: mounted file systems

       /etc/group
     to find group operator

SEE ALSO
       fstab(5), restore(8), rmt(8)

DIAGNOSTICS
       Many, and verbose.

COMPATIBILITY
       The  format  of the /etc/dumpdates file has changed in release 0.4b34,
       however, the file will be read  correctly  with either pre-0.4b34  or
       0.4b34  and  later  versions of dump provided that the machine on which
       dump is run did not change timezones (which should  be  a  fairly  rare
       occurrence).

EXIT STATUS
       Dump  exits  with  zero status on success. Startup errors are indicated
       with an exit code of 1; abnormal termination is indicated with an  exit
       code of 3.

BUGS
       It  might be considered a bug that this version of dump can only handle
       ext2/3 filesystems.  Specifically, it does not work with FAT  filesys-
       tems.

       Fewer  than  32 read errors (change this with -I) on the filesystem are
       ignored. If noticing read errors is important, the output from dump can
       be parsed to look for lines that contain the text ’read error’.

       When  a read  error occurs, dump prints out the corresponding physical
       disk block and sector number and the ext2/3 logical  block  number.  It
       doesn’t print out the corresponding file name or even the inode number.
       The user has to use debugfs(8), commands ncheck and icheck to translate
       the  ext2blk number printed out by dump into an inode number, then into
       a file name.

       Each reel requires a new process, so parent processes for reels already
       written just hang around until the entire tape is written.

       The estimated number of tapes is not correct if compression is on.

       It  would  be  nice if dump knew about the dump sequence, kept track of
       the tapes scribbled on, told the operator which tape to mount when, and
       provided more assistance for the operator running restore.

       Dump  cannot  do remote backups without being run as root, due to its
       security history.  Presently, it works if you set it  setuid  (like  it
       used  to be), but this might constitute a security risk. Note that you
       can set RSH to use a remote shell program instead.

AUTHOR
       The dump/restore backup suite was ported to  Linux’s  Second  Extended
       File System by Remy Card <card@Linux.EU.Org>. He maintained the initial
       versions of dump (up and including 0.4b4, released in January 1997).

       Starting  with 0.4b5,  the new   maintainer   is Stelian   Pop
       <stelian@popies.net>.

AVAILABILITY
       The  dump/restore  backup  suite is available from <http://dump.source-
       forge.net>

HISTORY
       A dump command appeared in Version 6 AT&T UNIX.

BSD       version 0.4b41 of January 2, 2006       DUMP(8)

Comments

Popular posts from this blog