1 \input texinfo @c -*-texinfo-*-
2 @comment %**start of header
5 @settitle GNU tar @value{VERSION}
13 @include rendition.texi
16 @c Put everything in one index (arbitrarily chosen to be the concept index).
26 This manual is for @acronym{GNU} @command{tar} (version
27 @value{VERSION}, @value{UPDATED}), which creates and extracts files
30 Copyright @copyright{} 1992, 1994, 1995, 1996, 1997, 1999, 2000, 2001,
31 2003, 2004, 2005, 2006 Free Software Foundation, Inc.
34 Permission is granted to copy, distribute and/or modify this document
35 under the terms of the GNU Free Documentation License, Version 1.1 or
36 any later version published by the Free Software Foundation; with no
37 Invariant Sections, with the Front-Cover Texts being ``A GNU Manual,''
38 and with the Back-Cover Texts as in (a) below. A copy of the license
39 is included in the section entitled "GNU Free Documentation License".
41 (a) The FSF's Back-Cover Text is: ``You are free to copy and modify
42 this GNU Manual. Buying copies from GNU Press supports the FSF in
43 developing GNU and promoting software freedom.''
47 @dircategory Archiving
49 * Tar: (tar). Making tape (or disk) archives.
52 @dircategory Individual utilities
54 * tar: (tar)tar invocation. Invoking @GNUTAR{}.
57 @shorttitlepage @acronym{GNU} @command{tar}
60 @title @acronym{GNU} tar: an archiver tool
61 @subtitle @value{RENDITION} @value{VERSION}, @value{UPDATED}
62 @author John Gilmore, Jay Fenlason et al.
65 @vskip 0pt plus 1filll
71 @top @acronym{GNU} tar: an archiver tool
76 @cindex archiving files
78 The first part of this master menu lists the major nodes in this Info
79 document. The rest of the menu lists all the lower level nodes.
82 @c The master menu, created with texinfo-master-menu, goes here.
83 @c (However, getdate.texi's menu is interpolated by hand.)
92 * Date input formats::
99 * Configuring Help Summary::
102 * Free Software Needs Free Documentation::
103 * Copying This Manual::
104 * Index of Command Line Options::
108 --- The Detailed Node Listing ---
112 * Book Contents:: What this Book Contains
113 * Definitions:: Some Definitions
114 * What tar Does:: What @command{tar} Does
115 * Naming tar Archives:: How @command{tar} Archives are Named
116 * Authors:: @GNUTAR{} Authors
117 * Reports:: Reporting bugs or suggestions
119 Tutorial Introduction to @command{tar}
122 * stylistic conventions::
123 * basic tar options:: Basic @command{tar} Operations and Options
124 * frequent operations::
125 * Two Frequent Options::
126 * create:: How to Create Archives
127 * list:: How to List Archives
128 * extract:: How to Extract Members from an Archive
131 Two Frequently Used Options
137 How to Create Archives
139 * prepare for examples::
140 * Creating the archive::
149 How to Extract Members from an Archive
151 * extracting archives::
159 * using tar options::
167 The Three Option Styles
169 * Mnemonic Options:: Mnemonic Option Style
170 * Short Options:: Short Option Style
171 * Old Options:: Old Option Style
172 * Mixing:: Mixing Option Styles
174 All @command{tar} Options
176 * Operation Summary::
178 * Short Option Summary::
190 Advanced @GNUTAR{} Operations
199 How to Add Files to Existing Archives: @option{--append}
201 * appending files:: Appending Files to an Archive
208 Options Used by @option{--create}
210 * Ignore Failed Read::
212 Options Used by @option{--extract}
214 * Reading:: Options to Help Read Archives
215 * Writing:: Changing How @command{tar} Writes Files
216 * Scarce:: Coping with Scarce Resources
218 Options to Help Read Archives
220 * read full records::
223 Changing How @command{tar} Writes Files
225 * Dealing with Old Files::
226 * Overwrite Old Files::
231 * Data Modification Times::
232 * Setting Access Permissions::
233 * Writing to Standard Output::
236 Coping with Scarce Resources
241 Performing Backups and Restoring Files
243 * Full Dumps:: Using @command{tar} to Perform Full Dumps
244 * Incremental Dumps:: Using @command{tar} to Perform Incremental Dumps
245 * Backup Levels:: Levels of Backups
246 * Backup Parameters:: Setting Parameters for Backups and Restoration
247 * Scripted Backups:: Using the Backup Scripts
248 * Scripted Restoration:: Using the Restore Script
250 Setting Parameters for Backups and Restoration
252 * General-Purpose Variables::
253 * Magnetic Tape Control::
255 * backup-specs example:: An Example Text of @file{Backup-specs}
257 Choosing Files and Names for @command{tar}
259 * file:: Choosing the Archive's Name
260 * Selecting Archive Members::
261 * files:: Reading Names from a File
262 * exclude:: Excluding Some Files
264 * after:: Operating Only on New Files
265 * recurse:: Descending into Directories
266 * one:: Crossing File System Boundaries
268 Reading Names from a File
274 * problems with exclude::
276 Crossing File System Boundaries
278 * directory:: Changing Directory
279 * absolute:: Absolute File Names
283 * General date syntax:: Common rules.
284 * Calendar date items:: 19 Dec 1994.
285 * Time of day items:: 9:20pm.
286 * Time zone items:: @sc{est}, @sc{pdt}, @sc{gmt}, ...
287 * Day of week items:: Monday and others.
288 * Relative items in date strings:: next tuesday, 2 years ago.
289 * Pure numbers in date strings:: 19931219, 1440.
290 * Seconds since the Epoch:: @@1078100502.
291 * Authors of get_date:: Bellovin, Eggert, Salz, Berets, et al.
293 Controlling the Archive Format
295 * Portability:: Making @command{tar} Archives More Portable
296 * Compression:: Using Less Space through Compression
297 * Attributes:: Handling File Attributes
298 * Standard:: The Standard Format
299 * Extensions:: @acronym{GNU} Extensions to the Archive Format
300 * cpio:: Comparison of @command{tar} and @command{cpio}
302 Making @command{tar} Archives More Portable
304 * Portable Names:: Portable Names
305 * dereference:: Symbolic Links
306 * old:: Old V7 Archives
307 * posix:: @acronym{POSIX} archives
308 * Checksumming:: Checksumming Problems
309 * Large or Negative Values:: Large files, negative time stamps, etc.
311 Using Less Space through Compression
313 * gzip:: Creating and Reading Compressed Archives
314 * sparse:: Archiving Sparse Files
316 Tapes and Other Archive Media
318 * Device:: Device selection and switching
319 * Remote Tape Server::
320 * Common Problems and Solutions::
321 * Blocking:: Blocking
322 * Many:: Many archives on one tape
323 * Using Multiple Tapes:: Using Multiple Tapes
324 * label:: Including a Label in the Archive
330 * Format Variations:: Format Variations
331 * Blocking Factor:: The Blocking Factor of an Archive
333 Many Archives on One Tape
335 * Tape Positioning:: Tape Positions and Tape Marks
336 * mt:: The @command{mt} Utility
340 * Multi-Volume Archives:: Archives Longer than One Tape or Disk
341 * Tape Files:: Tape Files
342 * Tarcat:: Concatenate Volumes into a Single Archive
344 GNU tar internals and development
351 * Free Software Needs Free Documentation::
352 * GNU Free Documentation License:: License for copying this manual
358 @chapter Introduction
361 and manipulates @dfn{archives} which are actually collections of
362 many other files; the program provides users with an organized and
363 systematic method for controlling a large amount of data.
364 The name ``tar'' originally came from the phrase ``Tape ARchive'', but
365 archives need not (and these days, typically do not) reside on tapes.
368 * Book Contents:: What this Book Contains
369 * Definitions:: Some Definitions
370 * What tar Does:: What @command{tar} Does
371 * Naming tar Archives:: How @command{tar} Archives are Named
372 * Authors:: @GNUTAR{} Authors
373 * Reports:: Reporting bugs or suggestions
377 @section What this Book Contains
379 The first part of this chapter introduces you to various terms that will
380 recur throughout the book. It also tells you who has worked on @GNUTAR{}
381 and its documentation, and where you should send bug reports
384 The second chapter is a tutorial (@pxref{Tutorial}) which provides a
385 gentle introduction for people who are new to using @command{tar}. It is
386 meant to be self contained, not requiring any reading from subsequent
387 chapters to make sense. It moves from topic to topic in a logical,
388 progressive order, building on information already explained.
390 Although the tutorial is paced and structured to allow beginners to
391 learn how to use @command{tar}, it is not intended solely for beginners.
392 The tutorial explains how to use the three most frequently used
393 operations (@samp{create}, @samp{list}, and @samp{extract}) as well as
394 two frequently used options (@samp{file} and @samp{verbose}). The other
395 chapters do not refer to the tutorial frequently; however, if a section
396 discusses something which is a complex variant of a basic concept, there
397 may be a cross reference to that basic concept. (The entire book,
398 including the tutorial, assumes that the reader understands some basic
399 concepts of using a Unix-type operating system; @pxref{Tutorial}.)
401 The third chapter presents the remaining five operations, and
402 information about using @command{tar} options and option syntax.
404 @FIXME{this sounds more like a @acronym{GNU} Project Manuals Concept [tm] more
405 than the reality. should think about whether this makes sense to say
406 here, or not.} The other chapters are meant to be used as a
407 reference. Each chapter presents everything that needs to be said
408 about a specific topic.
410 One of the chapters (@pxref{Date input formats}) exists in its
411 entirety in other @acronym{GNU} manuals, and is mostly self-contained.
412 In addition, one section of this manual (@pxref{Standard}) contains a
413 big quote which is taken directly from @command{tar} sources.
415 In general, we give both long and short (abbreviated) option names
416 at least once in each section where the relevant option is covered, so
417 that novice readers will become familiar with both styles. (A few
418 options have no short versions, and the relevant sections will
422 @section Some Definitions
426 The @command{tar} program is used to create and manipulate @command{tar}
427 archives. An @dfn{archive} is a single file which contains the contents
428 of many files, while still identifying the names of the files, their
429 owner(s), and so forth. (In addition, archives record access
430 permissions, user and group, size in bytes, and data modification time.
431 Some archives also record the file names in each archived directory, as
432 well as other file and directory information.) You can use @command{tar}
433 to @dfn{create} a new archive in a specified directory.
436 @cindex archive member
439 The files inside an archive are called @dfn{members}. Within this
440 manual, we use the term @dfn{file} to refer only to files accessible in
441 the normal ways (by @command{ls}, @command{cat}, and so forth), and the term
442 @dfn{member} to refer only to the members of an archive. Similarly, a
443 @dfn{file name} is the name of a file, as it resides in the file system,
444 and a @dfn{member name} is the name of an archive member within the
449 The term @dfn{extraction} refers to the process of copying an archive
450 member (or multiple members) into a file in the file system. Extracting
451 all the members of an archive is often called @dfn{extracting the
452 archive}. The term @dfn{unpack} can also be used to refer to the
453 extraction of many or all the members of an archive. Extracting an
454 archive does not destroy the archive's structure, just as creating an
455 archive does not destroy the copies of the files that exist outside of
456 the archive. You may also @dfn{list} the members in a given archive
457 (this is often thought of as ``printing'' them to the standard output,
458 or the command line), or @dfn{append} members to a pre-existing archive.
459 All of these operations can be performed using @command{tar}.
462 @section What @command{tar} Does
465 The @command{tar} program provides the ability to create @command{tar}
466 archives, as well as various other kinds of manipulation. For example,
467 you can use @command{tar} on previously created archives to extract files,
468 to store additional files, or to update or list files which were already
471 Initially, @command{tar} archives were used to store files conveniently on
472 magnetic tape. The name @command{tar} comes from this use; it stands for
473 @code{t}ape @code{ar}chiver. Despite the utility's name, @command{tar} can
474 direct its output to available devices, files, or other programs (using
475 pipes). @command{tar} may even access remote devices or files (as archives).
477 @FIXME{the following table entries need a bit of work..}
479 You can use @command{tar} archives in many ways. We want to stress a few
480 of them: storage, backup, and transportation.
484 Often, @command{tar} archives are used to store related files for
485 convenient file transfer over a network. For example, the
486 @acronym{GNU} Project distributes its software bundled into
487 @command{tar} archives, so that all the files relating to a particular
488 program (or set of related programs) can be transferred as a single
491 A magnetic tape can store several files in sequence. However, the tape
492 has no names for these files; it only knows their relative position on
493 the tape. One way to store several files on one tape and retain their
494 names is by creating a @command{tar} archive. Even when the basic transfer
495 mechanism can keep track of names, as FTP can, the nuisance of handling
496 multiple files, directories, and multiple links makes @command{tar}
499 Archive files are also used for long-term storage. You can think of
500 this as transportation from the present into the future. (It is a
501 science-fiction idiom that you can move through time as well as in
502 space; the idea here is that @command{tar} can be used to move archives in
503 all dimensions, even time!)
506 Because the archive created by @command{tar} is capable of preserving
507 file information and directory structure, @command{tar} is commonly
508 used for performing full and incremental backups of disks. A backup
509 puts a collection of files (possibly pertaining to many users and
510 projects) together on a disk or a tape. This guards against
511 accidental destruction of the information in those files.
512 @GNUTAR{} has special features that allow it to be
513 used to make incremental and full dumps of all the files in a
517 You can create an archive on one system, transfer it to another system,
518 and extract the contents there. This allows you to transport a group of
519 files from one system to another.
522 @node Naming tar Archives
523 @section How @command{tar} Archives are Named
525 Conventionally, @command{tar} archives are given names ending with
526 @samp{.tar}. This is not necessary for @command{tar} to operate properly,
527 but this manual follows that convention in order to accustom readers to
528 it and to make examples more clear.
533 Often, people refer to @command{tar} archives as ``@command{tar} files,'' and
534 archive members as ``files'' or ``entries''. For people familiar with
535 the operation of @command{tar}, this causes no difficulty. However, in
536 this manual, we consistently refer to ``archives'' and ``archive
537 members'' to make learning to use @command{tar} easier for novice users.
540 @section @GNUTAR{} Authors
542 @GNUTAR{} was originally written by John Gilmore,
543 and modified by many people. The @acronym{GNU} enhancements were
544 written by Jay Fenlason, then Joy Kendall, and the whole package has
545 been further maintained by Thomas Bushnell, n/BSG, Fran@,{c}ois
546 Pinard, Paul Eggert, and finally Sergey Poznyakoff with the help of
547 numerous and kind users.
549 We wish to stress that @command{tar} is a collective work, and owes much to
550 all those people who reported problems, offered solutions and other
551 insights, or shared their thoughts and suggestions. An impressive, yet
552 partial list of those contributors can be found in the @file{THANKS}
553 file from the @GNUTAR{} distribution.
555 @FIXME{i want all of these names mentioned, Absolutely. BUT, i'm not
556 sure i want to spell out the history in this detail, at least not for
557 the printed book. i'm just not sure it needs to be said this way.
558 i'll think about it.}
560 @FIXME{History is more important, and surely more interesting, than
561 actual names. Quoting names without history would be meaningless. FP}
563 Jay Fenlason put together a draft of a @GNUTAR{}
564 manual, borrowing notes from the original man page from John Gilmore.
565 This was withdrawn in version 1.11. Thomas Bushnell, n/BSG and Amy
566 Gorin worked on a tutorial and manual for @GNUTAR{}.
567 Fran@,{c}ois Pinard put version 1.11.8 of the manual together by
568 taking information from all these sources and merging them. Melissa
569 Weisshaus finally edited and redesigned the book to create version
570 1.12. @FIXME{update version number as necessary; i'm being
571 optimistic!} @FIXME{Someone [maybe karl berry? maybe bob chassell?
572 maybe melissa? maybe julie sussman?] needs to properly index the
575 For version 1.12, Daniel Hagerty contributed a great deal of technical
576 consulting. In particular, he is the primary author of @ref{Backups}.
578 In July, 2003 @GNUTAR{} was put on CVS at savannah.gnu.org
579 (see @url{http://savannah.gnu.org/projects/tar}), and
580 active development and maintenance work has started
581 again. Currently @GNUTAR{} is being maintained by Paul Eggert, Sergey
582 Poznyakoff and Jeff Bailey.
584 Support for @acronym{POSIX} archives was added by Sergey Poznyakoff.
587 @section Reporting bugs or suggestions
590 @cindex reporting bugs
591 If you find problems or have suggestions about this program or manual,
592 please report them to @file{bug-tar@@gnu.org}.
594 When reporting a bug, please be sure to include as much detail as
595 possible, in order to reproduce it. @FIXME{Be more specific, I'd
596 like to make this node as detailed as 'Bug reporting' node in Emacs
600 @chapter Tutorial Introduction to @command{tar}
602 This chapter guides you through some basic examples of three @command{tar}
603 operations: @option{--create}, @option{--list}, and @option{--extract}. If
604 you already know how to use some other version of @command{tar}, then you
605 may not need to read this chapter. This chapter omits most complicated
606 details about how @command{tar} works.
610 * stylistic conventions::
611 * basic tar options:: Basic @command{tar} Operations and Options
612 * frequent operations::
613 * Two Frequent Options::
614 * create:: How to Create Archives
615 * list:: How to List Archives
616 * extract:: How to Extract Members from an Archive
621 @section Assumptions this Tutorial Makes
623 This chapter is paced to allow beginners to learn about @command{tar}
624 slowly. At the same time, we will try to cover all the basic aspects of
625 these three operations. In order to accomplish both of these tasks, we
626 have made certain assumptions about your knowledge before reading this
627 manual, and the hardware you will be using:
631 Before you start to work through this tutorial, you should understand
632 what the terms ``archive'' and ``archive member'' mean
633 (@pxref{Definitions}). In addition, you should understand something
634 about how Unix-type operating systems work, and you should know how to
635 use some basic utilities. For example, you should know how to create,
636 list, copy, rename, edit, and delete files and directories; how to
637 change between directories; and how to figure out where you are in the
638 file system. You should have some basic understanding of directory
639 structure and how files are named according to which directory they are
640 in. You should understand concepts such as standard output and standard
641 input, what various definitions of the term ``argument'' mean, and the
642 differences between relative and absolute path names. @FIXME{and what
646 This manual assumes that you are working from your own home directory
647 (unless we state otherwise). In this tutorial, you will create a
648 directory to practice @command{tar} commands in. When we show path names,
649 we will assume that those paths are relative to your home directory.
650 For example, my home directory path is @file{/home/fsf/melissa}. All of
651 my examples are in a subdirectory of the directory named by that path
652 name; the subdirectory is called @file{practice}.
655 In general, we show examples of archives which exist on (or can be
656 written to, or worked with from) a directory on a hard disk. In most
657 cases, you could write those archives to, or work with them on any other
658 device, such as a tape drive. However, some of the later examples in
659 the tutorial and next chapter will not work on tape drives.
660 Additionally, working with tapes is much more complicated than working
661 with hard disks. For these reasons, the tutorial does not cover working
662 with tape drives. @xref{Media}, for complete information on using
663 @command{tar} archives with tape drives.
665 @FIXME{this is a cop out. need to add some simple tape drive info.}
668 @node stylistic conventions
669 @section Stylistic Conventions
671 In the examples, @samp{$} represents a typical shell prompt. It
672 precedes lines you should type; to make this more clear, those lines are
673 shown in @kbd{this font}, as opposed to lines which represent the
674 computer's response; those lines are shown in @code{this font}, or
675 sometimes @samp{like this}.
677 @c When we have lines which are too long to be
678 @c displayed in any other way, we will show them like this:
680 @node basic tar options
681 @section Basic @command{tar} Operations and Options
683 @command{tar} can take a wide variety of arguments which specify and define
684 the actions it will have on the particular set of files or the archive.
685 The main types of arguments to @command{tar} fall into one of two classes:
686 operations, and options.
688 Some arguments fall into a class called @dfn{operations}; exactly one of
689 these is both allowed and required for any instance of using @command{tar};
690 you may @emph{not} specify more than one. People sometimes speak of
691 @dfn{operating modes}. You are in a particular operating mode when you
692 have specified the operation which specifies it; there are eight
693 operations in total, and thus there are eight operating modes.
695 The other arguments fall into the class known as @dfn{options}. You are
696 not required to specify any options, and you are allowed to specify more
697 than one at a time (depending on the way you are using @command{tar} at
698 that time). Some options are used so frequently, and are so useful for
699 helping you type commands more carefully that they are effectively
700 ``required''. We will discuss them in this chapter.
702 You can write most of the @command{tar} operations and options in any
703 of three forms: long (mnemonic) form, short form, and old style. Some
704 of the operations and options have no short or ``old'' forms; however,
705 the operations and options which we will cover in this tutorial have
706 corresponding abbreviations. @FIXME{make sure this is still the case,
707 at the end}We will indicate those abbreviations appropriately to get
708 you used to seeing them. (Note that the ``old style'' option forms
709 exist in @GNUTAR{} for compatibility with Unix
710 @command{tar}. In this book we present a full discussion of this way
711 of writing options and operations (@pxref{Old Options}), and we discuss
712 the other two styles of writing options (@xref{Mnemonic Options}, and
713 @pxref{Short Options}).
715 In the examples and in the text of this tutorial, we usually use the
716 long forms of operations and options; but the ``short'' forms produce
717 the same result and can make typing long @command{tar} commands easier.
718 For example, instead of typing
721 @kbd{tar --create --verbose --file=afiles.tar apple angst aspic}
727 @kbd{tar -c -v -f afiles.tar apple angst aspic}
733 @kbd{tar -cvf afiles.tar apple angst aspic}
737 For more information on option syntax, see @ref{Advanced tar}. In
738 discussions in the text, when we name an option by its long form, we
739 also give the corresponding short option in parentheses.
741 The term, ``option'', can be confusing at times, since ``operations''
742 are often lumped in with the actual, @emph{optional} ``options'' in certain
743 general class statements. For example, we just talked about ``short and
744 long forms of options and operations''. However, experienced @command{tar}
745 users often refer to these by shorthand terms such as, ``short and long
746 options''. This term assumes that the ``operations'' are included, also.
747 Context will help you determine which definition of ``options'' to use.
749 Similarly, the term ``command'' can be confusing, as it is often used in
750 two different ways. People sometimes refer to @command{tar} ``commands''.
751 A @command{tar} @dfn{command} is the entire command line of user input
752 which tells @command{tar} what to do --- including the operation, options,
753 and any arguments (file names, pipes, other commands, etc). However,
754 you will also sometimes hear the term ``the @command{tar} command''. When
755 the word ``command'' is used specifically like this, a person is usually
756 referring to the @command{tar} @emph{operation}, not the whole line.
757 Again, use context to figure out which of the meanings the speaker
760 @node frequent operations
761 @section The Three Most Frequently Used Operations
763 Here are the three most frequently used operations (both short and long
764 forms), as well as a brief description of their meanings. The rest of
765 this chapter will cover how to use these operations in detail. We will
766 present the rest of the operations in the next chapter.
771 Create a new @command{tar} archive.
774 List the contents of an archive.
777 Extract one or more members from an archive.
780 @node Two Frequent Options
781 @section Two Frequently Used Options
783 To understand how to run @command{tar} in the three operating modes listed
784 previously, you also need to understand how to use two of the options to
785 @command{tar}: @option{--file} (which takes an archive file as an argument)
786 and @option{--verbose}. (You are usually not @emph{required} to specify
787 either of these options when you run @command{tar}, but they can be very
788 useful in making things more clear and helping you avoid errors.)
797 @unnumberedsubsec The @option{--file} Option
800 @opindex file, tutorial
801 @item --file=@var{archive-name}
802 @itemx -f @var{archive-name}
803 Specify the name of an archive file.
806 You can specify an argument for the @option{--file=@var{archive-name}} (@option{-f @var{archive-name}}) option whenever you
807 use @command{tar}; this option determines the name of the archive file
808 that @command{tar} will work on.
811 If you don't specify this argument, then @command{tar} will examine
812 the environment variable @env{TAPE}. If it is set, its value will be
813 used as the archive name. Otherwise, @command{tar} will use the
814 default archive, determined at the compile time. Usually it is
815 standard output or some physical tape drive attached to your machine
816 (you can verify what the default is by running @kbd{tar
817 --show-defaults}, @pxref{defaults}). If there is no tape drive
818 attached, or the default is not meaningful, then @command{tar} will
819 print an error message. The error message might look roughly like one
823 tar: can't open /dev/rmt8 : No such device or address
824 tar: can't open /dev/rsmt0 : I/O error
828 To avoid confusion, we recommend that you always specify an archive file
829 name by using @option{--file=@var{archive-name}} (@option{-f @var{archive-name}}) when writing your @command{tar} commands.
830 For more information on using the @option{--file=@var{archive-name}} (@option{-f @var{archive-name}}) option, see
833 @node verbose tutorial
834 @unnumberedsubsec The @option{--verbose} Option
837 @opindex verbose, introduced
840 Show the files being worked on as @command{tar} is running.
843 @option{--verbose} (@option{-v}) shows details about the results of running
844 @command{tar}. This can be especially useful when the results might not be
845 obvious. For example, if you want to see the progress of @command{tar} as
846 it writes files into the archive, you can use the @option{--verbose}
847 option. In the beginning, you may find it useful to use
848 @option{--verbose} at all times; when you are more accustomed to
849 @command{tar}, you will likely want to use it at certain times but not at
850 others. We will use @option{--verbose} at times to help make something
851 clear, and we will give many examples both using and not using
852 @option{--verbose} to show the differences.
854 Sometimes, a single instance of @option{--verbose} on the command line
855 will show a full, @samp{ls} style listing of an archive or files,
856 giving sizes, owners, and similar information. @FIXME{Describe the
857 exact output format, e.g., how hard links are displayed.}
858 Other times, @option{--verbose} will only show files or members that the particular
859 operation is operating on at the time. In the latter case, you can
860 use @option{--verbose} twice in a command to get a listing such as that
861 in the former case. For example, instead of saying
864 @kbd{tar -cvf afiles.tar apple angst aspic}
871 @kbd{tar -cvvf afiles.tar apple angst aspic}
875 This works equally well using short or long forms of options. Using
876 long forms, you would simply write out the mnemonic form of the option
880 $ @kbd{tar --create --verbose --verbose @dots{}}
884 Note that you must double the hyphens properly each time.
886 Later in the tutorial, we will give examples using @w{@option{--verbose
890 @unnumberedsubsec Getting Help: Using the @option{--help} Option
896 The @option{--help} option to @command{tar} prints out a very brief list of
897 all operations and option available for the current version of
898 @command{tar} available on your system.
902 @section How to Create Archives
905 @cindex Creation of the archive
906 @cindex Archive, creation of
907 One of the basic operations of @command{tar} is @option{--create} (@option{-c}), which
908 you use to create a @command{tar} archive. We will explain
909 @option{--create} first because, in order to learn about the other
910 operations, you will find it useful to have an archive available to
913 To make this easier, in this section you will first create a directory
914 containing three files. Then, we will show you how to create an
915 @emph{archive} (inside the new directory). Both the directory, and
916 the archive are specifically for you to practice on. The rest of this
917 chapter and the next chapter will show many examples using this
918 directory and the files you will create: some of those files may be
919 other directories and other archives.
921 The three files you will archive in this example are called
922 @file{blues}, @file{folk}, and @file{jazz}. The archive is called
923 @file{collection.tar}.
925 This section will proceed slowly, detailing how to use @option{--create}
926 in @code{verbose} mode, and showing examples using both short and long
927 forms. In the rest of the tutorial, and in the examples in the next
928 chapter, we will proceed at a slightly quicker pace. This section
929 moves more slowly to allow beginning users to understand how
933 * prepare for examples::
934 * Creating the archive::
940 @node prepare for examples
941 @subsection Preparing a Practice Directory for Examples
943 To follow along with this and future examples, create a new directory
944 called @file{practice} containing files called @file{blues}, @file{folk}
945 and @file{jazz}. The files can contain any information you like:
946 ideally, they should contain information which relates to their names,
947 and be of different lengths. Our examples assume that @file{practice}
948 is a subdirectory of your home directory.
950 Now @command{cd} to the directory named @file{practice}; @file{practice}
951 is now your @dfn{working directory}. (@emph{Please note}: Although
952 the full path name of this directory is
953 @file{/@var{homedir}/practice}, in our examples we will refer to
954 this directory as @file{practice}; the @var{homedir} is presumed.
956 In general, you should check that the files to be archived exist where
957 you think they do (in the working directory) by running @command{ls}.
958 Because you just created the directory and the files and have changed to
959 that directory, you probably don't need to do that this time.
961 It is very important to make sure there isn't already a file in the
962 working directory with the archive name you intend to use (in this case,
963 @samp{collection.tar}), or that you don't care about its contents.
964 Whenever you use @samp{create}, @command{tar} will erase the current
965 contents of the file named by @option{--file=@var{archive-name}} (@option{-f @var{archive-name}}) if it exists. @command{tar}
966 will not tell you if you are about to overwrite an archive unless you
967 specify an option which does this (@pxref{backup}, for the
968 information on how to do so). To add files to an existing archive,
969 you need to use a different option, such as @option{--append} (@option{-r}); see
970 @ref{append} for information on how to do this.
972 @node Creating the archive
973 @subsection Creating the Archive
975 @opindex create, introduced
976 To place the files @file{blues}, @file{folk}, and @file{jazz} into an
977 archive named @file{collection.tar}, use the following command:
980 $ @kbd{tar --create --file=collection.tar blues folk jazz}
983 The order of the arguments is not very important, @emph{when using long
984 option forms}. You could also say:
987 $ @kbd{tar blues --create folk --file=collection.tar jazz}
991 However, you can see that this order is harder to understand; this is
992 why we will list the arguments in the order that makes the commands
993 easiest to understand (and we encourage you to do the same when you use
994 @command{tar}, to avoid errors).
996 Note that the part of the command which says,
997 @w{@option{--file=collection.tar}} is considered to be @emph{one} argument.
998 If you substituted any other string of characters for
999 @kbd{collection.tar}, then that string would become the name of the
1000 archive file you create.
1002 The order of the options becomes more important when you begin to use
1003 short forms. With short forms, if you type commands in the wrong order
1004 (even if you type them correctly in all other ways), you may end up with
1005 results you don't expect. For this reason, it is a good idea to get
1006 into the habit of typing options in the order that makes inherent sense.
1007 @xref{short create}, for more information on this.
1009 In this example, you type the command as shown above: @option{--create}
1010 is the operation which creates the new archive
1011 (@file{collection.tar}), and @option{--file} is the option which lets
1012 you give it the name you chose. The files, @file{blues}, @file{folk},
1013 and @file{jazz}, are now members of the archive, @file{collection.tar}
1014 (they are @dfn{file name arguments} to the @option{--create} operation).
1015 @FIXME{xref here to the discussion of file name args?}Now that they are
1016 in the archive, they are called @emph{archive members}, not files.
1017 (@pxref{Definitions,members}).
1019 When you create an archive, you @emph{must} specify which files you
1020 want placed in the archive. If you do not specify any archive
1021 members, @GNUTAR{} will complain.
1023 If you now list the contents of the working directory (@command{ls}), you will
1024 find the archive file listed as well as the files you saw previously:
1027 blues folk jazz collection.tar
1031 Creating the archive @samp{collection.tar} did not destroy the copies of
1032 the files in the directory.
1034 Keep in mind that if you don't indicate an operation, @command{tar} will not
1035 run and will prompt you for one. If you don't name any files, @command{tar}
1036 will complain. You must have write access to the working directory,
1037 or else you will not be able to create an archive in that directory.
1039 @emph{Caution}: Do not attempt to use @option{--create} (@option{-c}) to add files to
1040 an existing archive; it will delete the archive and write a new one.
1041 Use @option{--append} (@option{-r}) instead. @xref{append}.
1043 @node create verbose
1044 @subsection Running @option{--create} with @option{--verbose}
1046 @opindex create, using with @option{--verbose}
1047 @opindex verbose, using with @option{--create}
1048 If you include the @option{--verbose} (@option{-v}) option on the command line,
1049 @command{tar} will list the files it is acting on as it is working. In
1050 verbose mode, the @code{create} example above would appear as:
1053 $ @kbd{tar --create --verbose --file=collection.tar blues folk jazz}
1059 This example is just like the example we showed which did not use
1060 @option{--verbose}, except that @command{tar} generated the remaining lines
1062 (note the different font styles).
1068 In the rest of the examples in this chapter, we will frequently use
1069 @code{verbose} mode so we can show actions or @command{tar} responses that
1070 you would otherwise not see, and which are important for you to
1074 @subsection Short Forms with @samp{create}
1076 As we said before, the @option{--create} (@option{-c}) operation is one of the most
1077 basic uses of @command{tar}, and you will use it countless times.
1078 Eventually, you will probably want to use abbreviated (or ``short'')
1079 forms of options. A full discussion of the three different forms that
1080 options can take appears in @ref{Styles}; for now, here is what the
1081 previous example (including the @option{--verbose} (@option{-v}) option) looks like
1082 using short option forms:
1085 $ @kbd{tar -cvf collection.tar blues folk jazz}
1092 As you can see, the system responds the same no matter whether you use
1093 long or short option forms.
1095 @FIXME{i don't like how this is worded:} One difference between using
1096 short and long option forms is that, although the exact placement of
1097 arguments following options is no more specific when using short forms,
1098 it is easier to become confused and make a mistake when using short
1099 forms. For example, suppose you attempted the above example in the
1103 $ @kbd{tar -cfv collection.tar blues folk jazz}
1107 In this case, @command{tar} will make an archive file called @file{v},
1108 containing the files @file{blues}, @file{folk}, and @file{jazz}, because
1109 the @samp{v} is the closest ``file name'' to the @option{-f} option, and
1110 is thus taken to be the chosen archive file name. @command{tar} will try
1111 to add a file called @file{collection.tar} to the @file{v} archive file;
1112 if the file @file{collection.tar} did not already exist, @command{tar} will
1113 report an error indicating that this file does not exist. If the file
1114 @file{collection.tar} does already exist (e.g., from a previous command
1115 you may have run), then @command{tar} will add this file to the archive.
1116 Because the @option{-v} option did not get registered, @command{tar} will not
1117 run under @samp{verbose} mode, and will not report its progress.
1119 The end result is that you may be quite confused about what happened,
1120 and possibly overwrite a file. To illustrate this further, we will show
1121 you how an example we showed previously would look using short forms.
1126 $ @kbd{tar blues --create folk --file=collection.tar jazz}
1130 is confusing as it is. When shown using short forms, however, it
1131 becomes much more so:
1134 $ @kbd{tar blues -c folk -f collection.tar jazz}
1138 It would be very easy to put the wrong string of characters
1139 immediately following the @option{-f}, but doing that could sacrifice
1142 For this reason, we recommend that you pay very careful attention to
1143 the order of options and placement of file and archive names,
1144 especially when using short option forms. Not having the option name
1145 written out mnemonically can affect how well you remember which option
1146 does what, and therefore where different names have to be placed.
1149 @subsection Archiving Directories
1151 @cindex Archiving Directories
1152 @cindex Directories, Archiving
1153 You can archive a directory by specifying its directory name as a
1154 file name argument to @command{tar}. The files in the directory will be
1155 archived relative to the working directory, and the directory will be
1156 re-created along with its contents when the archive is extracted.
1158 To archive a directory, first move to its superior directory. If you
1159 have followed the previous instructions in this tutorial, you should
1168 This will put you into the directory which contains @file{practice},
1169 i.e., your home directory. Once in the superior directory, you can
1170 specify the subdirectory, @file{practice}, as a file name argument. To
1171 store @file{practice} in the new archive file @file{music.tar}, type:
1174 $ @kbd{tar --create --verbose --file=music.tar practice}
1178 @command{tar} should output:
1185 practice/collection.tar
1188 Note that the archive thus created is not in the subdirectory
1189 @file{practice}, but rather in the current working directory---the
1190 directory from which @command{tar} was invoked. Before trying to archive a
1191 directory from its superior directory, you should make sure you have
1192 write access to the superior directory itself, not only the directory
1193 you are trying archive with @command{tar}. For example, you will probably
1194 not be able to store your home directory in an archive by invoking
1195 @command{tar} from the root directory; @xref{absolute}. (Note
1196 also that @file{collection.tar}, the original archive file, has itself
1197 been archived. @command{tar} will accept any file as a file to be
1198 archived, regardless of its content. When @file{music.tar} is
1199 extracted, the archive file @file{collection.tar} will be re-written
1200 into the file system).
1202 If you give @command{tar} a command such as
1205 $ @kbd{tar --create --file=foo.tar .}
1209 @command{tar} will report @samp{tar: ./foo.tar is the archive; not
1210 dumped}. This happens because @command{tar} creates the archive
1211 @file{foo.tar} in the current directory before putting any files into
1212 it. Then, when @command{tar} attempts to add all the files in the
1213 directory @file{.} to the archive, it notices that the file
1214 @file{./foo.tar} is the same as the archive @file{foo.tar}, and skips
1215 it. (It makes no sense to put an archive into itself.) @GNUTAR{}
1216 will continue in this case, and create the archive
1217 normally, except for the exclusion of that one file. (@emph{Please
1218 note:} Other versions of @command{tar} are not so clever; they will
1219 enter an infinite loop when this happens, so you should not depend on
1220 this behavior unless you are certain you are running @GNUTAR{}.)
1221 @FIXME{bob doesn't like this sentence, since he does
1222 it all the time, and we've been doing it in the editing passes for
1223 this manual: In general, make sure that the archive is not inside a
1224 directory being dumped.}
1227 @section How to List Archives
1230 Frequently, you will find yourself wanting to determine exactly what a
1231 particular archive contains. You can use the @option{--list} (@option{-t}) operation
1232 to get the member names as they currently appear in the archive, as well
1233 as various attributes of the files at the time they were archived. For
1234 example, you can examine the archive @file{collection.tar} that you
1235 created in the last section with the command,
1238 $ @kbd{tar --list --file=collection.tar}
1242 The output of @command{tar} would then be:
1250 @FIXME{we hope this will change. if it doesn't, need to show the
1251 creation of bfiles somewhere above!!! : }
1254 The archive @file{bfiles.tar} would list as follows:
1263 Be sure to use a @option{--file=@var{archive-name}} (@option{-f @var{archive-name}}) option just as with @option{--create} (@option{-c})
1264 to specify the name of the archive.
1266 @opindex list, using with @option{--verbose}
1267 @opindex verbose, using with @option{--list}
1268 If you use the @option{--verbose} (@option{-v}) option with @option{--list}, then
1269 @command{tar} will print out a listing reminiscent of @w{@samp{ls -l}},
1270 showing owner, file size, and so forth.
1272 If you had used @option{--verbose} (@option{-v}) mode, the example above would look
1276 $ @kbd{tar --list --verbose --file=collection.tar folk}
1277 -rw-r--r-- myself user 62 1990-05-23 10:55 folk
1280 @cindex listing member and file names
1281 @anchor{listing member and file names}
1282 It is important to notice that the output of @kbd{tar --list
1283 --verbose} does not necessarily match that produced by @kbd{tar
1284 --create --verbose} while creating the archive. It is because
1285 @GNUTAR{}, unless told explicitly not to do so, removes some directory
1286 prefixes from file names before storing them in the archive
1287 (@xref{absolute}, for more information). In other
1288 words, in verbose mode @GNUTAR{} shows @dfn{file names} when creating
1289 an archive and @dfn{member names} when listing it. Consider this
1294 $ @kbd{tar cfv archive /etc/mail}
1295 tar: Removing leading `/' from member names
1297 /etc/mail/sendmail.cf
1299 $ @kbd{tar tf archive}
1301 etc/mail/sendmail.cf
1306 @opindex show-stored-names
1307 This default behavior can sometimes be inconvenient. You can force
1308 @GNUTAR{} show member names when creating archive by supplying
1309 @option{--show-stored-names} option.
1312 @item --show-stored-names
1313 Print member (as opposed to @emph{file}) names when creating the archive.
1316 @cindex File name arguments, using @option{--list} with
1317 @opindex list, using with file name arguments
1318 You can specify one or more individual member names as arguments when
1319 using @samp{list}. In this case, @command{tar} will only list the
1320 names of members you identify. For example, @w{@kbd{tar --list
1321 --file=afiles.tar apple}} would only print @file{apple}.
1323 @FIXME{we hope the relevant aspects of this will change:}Because
1324 @command{tar} preserves paths, file names must be specified as they appear
1325 in the archive (ie., relative to the directory from which the archive
1326 was created). Therefore, it is essential when specifying member names
1327 to @command{tar} that you give the exact member names. For example,
1328 @w{@kbd{tar --list --file=bfiles birds}} would produce an error message
1329 something like @samp{tar: birds: Not found in archive}, because there is
1330 no member named @file{birds}, only one named @file{./birds}. While the
1331 names @file{birds} and @file{./birds} name the same file, @emph{member}
1332 names are compared using a simplistic name comparison, in which an exact
1333 match is necessary. @xref{absolute}.
1335 However, @w{@kbd{tar --list --file=collection.tar folk}} would respond
1336 with @file{folk}, because @file{folk} is in the archive file
1337 @file{collection.tar}. If you are not sure of the exact file name, try
1338 listing all the files in the archive and searching for the one you
1339 expect to find; remember that if you use @option{--list} with no file
1340 names as arguments, @command{tar} will print the names of all the members
1341 stored in the specified archive.
1348 @unnumberedsubsec Listing the Contents of a Stored Directory
1350 To get information about the contents of an archived directory,
1351 use the directory name as a file name argument in conjunction with
1352 @option{--list} (@option{-t}). To find out file attributes, include the
1353 @option{--verbose} (@option{-v}) option.
1355 For example, to find out about files in the directory @file{practice}, in
1356 the archive file @file{music.tar}, type:
1359 $ @kbd{tar --list --verbose --file=music.tar practice}
1362 @command{tar} responds:
1365 drwxrwxrwx myself user 0 1990-05-31 21:49 practice/
1366 -rw-r--r-- myself user 42 1990-05-21 13:29 practice/blues
1367 -rw-r--r-- myself user 62 1990-05-23 10:55 practice/folk
1368 -rw-r--r-- myself user 40 1990-05-21 13:30 practice/jazz
1369 -rw-r--r-- myself user 10240 1990-05-31 21:49 practice/collection.tar
1372 When you use a directory name as a file name argument, @command{tar} acts on
1373 all the files (including sub-directories) in that directory.
1376 @section How to Extract Members from an Archive
1379 @cindex Retrieving files from an archive
1380 @cindex Resurrecting files from an archive
1383 Creating an archive is only half the job---there is no point in storing
1384 files in an archive if you can't retrieve them. The act of retrieving
1385 members from an archive so they can be used and manipulated as
1386 unarchived files again is called @dfn{extraction}. To extract files
1387 from an archive, use the @option{--extract} (@option{--get} or
1388 @option{-x}) operation. As with @option{--create}, specify the name
1389 of the archive with @option{--file} (@option{-f}) option. Extracting
1390 an archive does not modify the archive in any way; you can extract it
1391 multiple times if you want or need to.
1393 Using @option{--extract}, you can extract an entire archive, or specific
1394 files. The files can be directories containing other files, or not. As
1395 with @option{--create} (@option{-c}) and @option{--list} (@option{-t}), you may use the short or the
1396 long form of the operation without affecting the performance.
1399 * extracting archives::
1400 * extracting files::
1402 * extracting untrusted archives::
1403 * failing commands::
1406 @node extracting archives
1407 @subsection Extracting an Entire Archive
1409 To extract an entire archive, specify the archive file name only, with
1410 no individual file names as arguments. For example,
1413 $ @kbd{tar -xvf collection.tar}
1420 -rw-r--r-- me user 28 1996-10-18 16:31 jazz
1421 -rw-r--r-- me user 21 1996-09-23 16:44 blues
1422 -rw-r--r-- me user 20 1996-09-23 16:44 folk
1425 @node extracting files
1426 @subsection Extracting Specific Files
1428 To extract specific archive members, give their exact member names as
1429 arguments, as printed by @option{--list} (@option{-t}). If you had mistakenly deleted
1430 one of the files you had placed in the archive @file{collection.tar}
1431 earlier (say, @file{blues}), you can extract it from the archive without
1432 changing the archive's structure. Its contents will be identical to the
1433 original file @file{blues} that you deleted.
1435 First, make sure you are in the @file{practice} directory, and list the
1436 files in the directory. Now, delete the file, @samp{blues}, and list
1437 the files in the directory again.
1439 You can now extract the member @file{blues} from the archive file
1440 @file{collection.tar} like this:
1443 $ @kbd{tar --extract --file=collection.tar blues}
1447 If you list the files in the directory again, you will see that the file
1448 @file{blues} has been restored, with its original permissions, data modification
1449 times, and owner.@FIXME{This is only accidentally true, but not in
1450 general. In most cases, one has to be root for restoring the owner, and
1451 use a special option for restoring permissions. Here, it just happens
1452 that the restoring user is also the owner of the archived members, and
1453 that the current @code{umask} is compatible with original permissions.}
1454 (These parameters will be identical to those which
1455 the file had when you originally placed it in the archive; any changes
1456 you may have made before deleting the file from the file system,
1457 however, will @emph{not} have been made to the archive member.) The
1458 archive file, @samp{collection.tar}, is the same as it was before you
1459 extracted @samp{blues}. You can confirm this by running @command{tar} with
1460 @option{--list} (@option{-t}).
1462 @FIXME{we hope this will change:}Remember that as with other operations,
1463 specifying the exact member name is important. @w{@kbd{tar --extract
1464 --file=bfiles.tar birds}} will fail, because there is no member named
1465 @file{birds}. To extract the member named @file{./birds}, you must
1466 specify @w{@kbd{tar --extract --file=bfiles.tar ./birds}}. To find the
1467 exact member names of the members of an archive, use @option{--list} (@option{-t})
1470 You can extract a file to standard output by combining the above options
1471 with the @option{--to-stdout} (@option{-O}) option (@pxref{Writing to Standard
1474 If you give the @option{--verbose} option, then @option{--extract}
1475 will print the names of the archive members as it extracts them.
1478 @subsection Extracting Files that are Directories
1480 Extracting directories which are members of an archive is similar to
1481 extracting other files. The main difference to be aware of is that if
1482 the extracted directory has the same name as any directory already in
1483 the working directory, then files in the extracted directory will be
1484 placed into the directory of the same name. Likewise, if there are
1485 files in the pre-existing directory with the same names as the members
1486 which you extract, the files from the extracted archive will replace
1487 the files already in the working directory (and possible
1488 subdirectories). This will happen regardless of whether or not the
1489 files in the working directory were more recent than those extracted
1490 (there exist, however, special options that alter this behavior
1493 However, if a file was stored with a directory name as part of its file
1494 name, and that directory does not exist under the working directory when
1495 the file is extracted, @command{tar} will create the directory.
1497 We can demonstrate how to use @option{--extract} to extract a directory
1498 file with an example. Change to the @file{practice} directory if you
1499 weren't there, and remove the files @file{folk} and @file{jazz}. Then,
1500 go back to the parent directory and extract the archive
1501 @file{music.tar}. You may either extract the entire archive, or you may
1502 extract only the files you just deleted. To extract the entire archive,
1503 don't give any file names as arguments after the archive name
1504 @file{music.tar}. To extract only the files you deleted, use the
1508 $ @kbd{tar -xvf music.tar practice/folk practice/jazz}
1514 If you were to specify two @option{--verbose} (@option{-v}) options, @command{tar}
1515 would have displayed more detail about the extracted files, as shown
1516 in the example below:
1519 $ @kbd{tar -xvvf music.tar practice/folk practice/jazz}
1520 -rw-r--r-- me user 28 1996-10-18 16:31 practice/jazz
1521 -rw-r--r-- me user 20 1996-09-23 16:44 practice/folk
1525 Because you created the directory with @file{practice} as part of the
1526 file names of each of the files by archiving the @file{practice}
1527 directory as @file{practice}, you must give @file{practice} as part
1528 of the file names when you extract those files from the archive.
1530 @FIXME{IMPORTANT! show the final structure, here. figure out what it
1533 @node extracting untrusted archives
1534 @subsection Extracting Archives from Untrusted Sources
1536 Extracting files from archives can overwrite files that already exist.
1537 If you receive an archive from an untrusted source, you should make a
1538 new directory and extract into that directory, so that you don't have
1539 to worry about the extraction overwriting one of your existing files.
1540 For example, if @file{untrusted.tar} came from somewhere else on the
1541 Internet, and you don't necessarily trust its contents, you can
1542 extract it as follows:
1545 $ @kbd{mkdir newdir}
1547 $ @kbd{tar -xvf ../untrusted.tar}
1550 It is also a good practice to examine contents of the archive
1551 before extracting it, using @option{--list} (@option{-t}) option, possibly combined
1552 with @option{--verbose} (@option{-v}).
1554 @node failing commands
1555 @subsection Commands That Will Fail
1557 Here are some sample commands you might try which will not work, and why
1560 If you try to use this command,
1563 $ @kbd{tar -xvf music.tar folk jazz}
1567 you will get the following response:
1570 tar: folk: Not found in archive
1571 tar: jazz: Not found in archive
1576 This is because these files were not originally @emph{in} the parent
1577 directory @file{..}, where the archive is located; they were in the
1578 @file{practice} directory, and their file names reflect this:
1581 $ @kbd{tar -tvf music.tar}
1587 @FIXME{make sure the above works when going through the examples in
1591 Likewise, if you try to use this command,
1594 $ @kbd{tar -tvf music.tar folk jazz}
1598 you would get a similar response. Members with those names are not in the
1599 archive. You must use the correct member names in order to extract the
1600 files from the archive.
1602 If you have forgotten the correct names of the files in the archive,
1603 use @w{@kbd{tar --list --verbose}} to list them correctly.
1605 @FIXME{more examples, here? hag thinks it's a good idea.}
1608 @section Going Further Ahead in this Manual
1610 @FIXME{need to write up a node here about the things that are going to
1611 be in the rest of the manual.}
1613 @node tar invocation
1614 @chapter Invoking @GNUTAR{}
1617 This chapter is about how one invokes the @GNUTAR{}
1618 command, from the command synopsis (@pxref{Synopsis}). There are
1619 numerous options, and many styles for writing them. One mandatory
1620 option specifies the operation @command{tar} should perform
1621 (@pxref{Operation Summary}), other options are meant to detail how
1622 this operation should be performed (@pxref{Option Summary}).
1623 Non-option arguments are not always interpreted the same way,
1624 depending on what the operation is.
1626 You will find in this chapter everything about option styles and rules for
1627 writing them (@pxref{Styles}). On the other hand, operations and options
1628 are fully described elsewhere, in other chapters. Here, you will find
1629 only synthetic descriptions for operations and options, together with
1630 pointers to other parts of the @command{tar} manual.
1632 Some options are so special they are fully described right in this
1633 chapter. They have the effect of inhibiting the normal operation of
1634 @command{tar} or else, they globally alter the amount of feedback the user
1635 receives about what is going on. These are the @option{--help} and
1636 @option{--version} (@pxref{help}), @option{--verbose} (@pxref{verbose})
1637 and @option{--interactive} options (@pxref{interactive}).
1641 * using tar options::
1651 @section General Synopsis of @command{tar}
1653 The @GNUTAR{} program is invoked as either one of:
1656 @kbd{tar @var{option}@dots{} [@var{name}]@dots{}}
1657 @kbd{tar @var{letter}@dots{} [@var{argument}]@dots{} [@var{option}]@dots{} [@var{name}]@dots{}}
1660 The second form is for when old options are being used.
1662 You can use @command{tar} to store files in an archive, to extract them from
1663 an archive, and to do other types of archive manipulation. The primary
1664 argument to @command{tar}, which is called the @dfn{operation}, specifies
1665 which action to take. The other arguments to @command{tar} are either
1666 @dfn{options}, which change the way @command{tar} performs an operation,
1667 or file names or archive members, which specify the files or members
1668 @command{tar} is to act on.
1670 You can actually type in arguments in any order, even if in this manual
1671 the options always precede the other arguments, to make examples easier
1672 to understand. Further, the option stating the main operation mode
1673 (the @command{tar} main command) is usually given first.
1675 Each @var{name} in the synopsis above is interpreted as an archive member
1676 name when the main command is one of @option{--compare}
1677 (@option{--diff}, @option{-d}), @option{--delete}, @option{--extract}
1678 (@option{--get}, @option{-x}), @option{--list} (@option{-t}) or
1679 @option{--update} (@option{-u}). When naming archive members, you
1680 must give the exact name of the member in the archive, as it is
1681 printed by @option{--list}. For @option{--append} (@option{-r}) and
1682 @option{--create} (@option{-c}), these @var{name} arguments specify
1683 the names of either files or directory hierarchies to place in the archive.
1684 These files or hierarchies should already exist in the file system,
1685 prior to the execution of the @command{tar} command.
1687 @command{tar} interprets relative file names as being relative to the
1688 working directory. @command{tar} will make all file names relative
1689 (by removing leading slashes when archiving or restoring files),
1690 unless you specify otherwise (using the @option{--absolute-names}
1691 option). @xref{absolute}, for more information about
1692 @option{--absolute-names}.
1694 If you give the name of a directory as either a file name or a member
1695 name, then @command{tar} acts recursively on all the files and directories
1696 beneath that directory. For example, the name @file{/} identifies all
1697 the files in the file system to @command{tar}.
1699 The distinction between file names and archive member names is especially
1700 important when shell globbing is used, and sometimes a source of confusion
1701 for newcomers. @xref{Wildcards}, for more information about globbing.
1702 The problem is that shells may only glob using existing files in the
1703 file system. Only @command{tar} itself may glob on archive members, so when
1704 needed, you must ensure that wildcard characters reach @command{tar} without
1705 being interpreted by the shell first. Using a backslash before @samp{*}
1706 or @samp{?}, or putting the whole argument between quotes, is usually
1707 sufficient for this.
1709 Even if @var{name}s are often specified on the command line, they
1710 can also be read from a text file in the file system, using the
1711 @option{--files-from=@var{file-of-names}} (@option{-T @var{file-of-names}}) option.
1713 If you don't use any file name arguments, @option{--append} (@option{-r}),
1714 @option{--delete} and @option{--concatenate} (@option{--catenate},
1715 @option{-A}) will do nothing, while @option{--create} (@option{-c})
1716 will usually yield a diagnostic and inhibit @command{tar} execution.
1717 The other operations of @command{tar} (@option{--list},
1718 @option{--extract}, @option{--compare}, and @option{--update})
1719 will act on the entire contents of the archive.
1722 @cindex return status
1723 Besides successful exits, @GNUTAR{} may fail for
1724 many reasons. Some reasons correspond to bad usage, that is, when the
1725 @command{tar} command is improperly written. Errors may be
1726 encountered later, while encountering an error processing the archive
1727 or the files. Some errors are recoverable, in which case the failure
1728 is delayed until @command{tar} has completed all its work. Some
1729 errors are such that it would not meaningful, or at least risky, to
1730 continue processing: @command{tar} then aborts processing immediately.
1731 All abnormal exits, whether immediate or delayed, should always be
1732 clearly diagnosed on @code{stderr}, after a line stating the nature of
1735 @GNUTAR{} returns only a few exit statuses. I'm really
1736 aiming simplicity in that area, for now. If you are not using the
1737 @option{--compare} @option{--diff}, @option{-d}) option, zero means
1738 that everything went well, besides maybe innocuous warnings. Nonzero
1739 means that something went wrong. Right now, as of today, ``nonzero''
1740 is almost always 2, except for remote operations, where it may be
1743 @node using tar options
1744 @section Using @command{tar} Options
1746 @GNUTAR{} has a total of eight operating modes which
1747 allow you to perform a variety of tasks. You are required to choose
1748 one operating mode each time you employ the @command{tar} program by
1749 specifying one, and only one operation as an argument to the
1750 @command{tar} command (two lists of four operations each may be found
1751 at @ref{frequent operations} and @ref{Operations}). Depending on
1752 circumstances, you may also wish to customize how the chosen operating
1753 mode behaves. For example, you may wish to change the way the output
1754 looks, or the format of the files that you wish to archive may require
1755 you to do something special in order to make the archive look right.
1757 You can customize and control @command{tar}'s performance by running
1758 @command{tar} with one or more options (such as @option{--verbose}
1759 (@option{-v}), which we used in the tutorial). As we said in the
1760 tutorial, @dfn{options} are arguments to @command{tar} which are (as
1761 their name suggests) optional. Depending on the operating mode, you
1762 may specify one or more options. Different options will have different
1763 effects, but in general they all change details of the operation, such
1764 as archive format, archive name, or level of user interaction. Some
1765 options make sense with all operating modes, while others are
1766 meaningful only with particular modes. You will likely use some
1767 options frequently, while you will only use others infrequently, or
1768 not at all. (A full list of options is available in @pxref{All Options}.)
1770 @vrindex TAR_OPTIONS, environment variable
1771 @anchor{TAR_OPTIONS}
1772 The @env{TAR_OPTIONS} environment variable specifies default options to
1773 be placed in front of any explicit options. For example, if
1774 @code{TAR_OPTIONS} is @samp{-v --unlink-first}, @command{tar} behaves as
1775 if the two options @option{-v} and @option{--unlink-first} had been
1776 specified before any explicit options. Option specifications are
1777 separated by whitespace. A backslash escapes the next character, so it
1778 can be used to specify an option containing whitespace or a backslash.
1780 Note that @command{tar} options are case sensitive. For example, the
1781 options @option{-T} and @option{-t} are different; the first requires an
1782 argument for stating the name of a file providing a list of @var{name}s,
1783 while the second does not require an argument and is another way to
1784 write @option{--list} (@option{-t}).
1786 In addition to the eight operations, there are many options to
1787 @command{tar}, and three different styles for writing both: long (mnemonic)
1788 form, short form, and old style. These styles are discussed below.
1789 Both the options and the operations can be written in any of these three
1792 @FIXME{menu at end of this node. need to think of an actual outline
1793 for this chapter; probably do that after stuff from chapter 4 is
1797 @section The Three Option Styles
1799 There are three styles for writing operations and options to the command
1800 line invoking @command{tar}. The different styles were developed at
1801 different times during the history of @command{tar}. These styles will be
1802 presented below, from the most recent to the oldest.
1804 Some options must take an argument. (For example, @option{--file}
1805 (@option{-f})) takes the name of an archive file as an argument. If
1806 you do not supply an archive file name, @command{tar} will use a
1807 default, but this can be confusing; thus, we recommend that you always
1808 supply a specific archive file name.) Where you @emph{place} the
1809 arguments generally depends on which style of options you choose. We
1810 will detail specific information relevant to each option style in the
1811 sections on the different option styles, below. The differences are
1812 subtle, yet can often be very important; incorrect option placement
1813 can cause you to overwrite a number of important files. We urge you
1814 to note these differences, and only use the option style(s) which
1815 makes the most sense to you until you feel comfortable with the others.
1817 Some options @emph{may} take an argument (currently, there are
1818 two such options: @option{--backup} and @option{--occurrence}). Such
1819 options may have at most long and short forms, they do not have old style
1820 equivalent. The rules for specifying an argument for such options
1821 are stricter than those for specifying mandatory arguments. Please,
1822 pay special attention to them.
1825 * Mnemonic Options:: Mnemonic Option Style
1826 * Short Options:: Short Option Style
1827 * Old Options:: Old Option Style
1828 * Mixing:: Mixing Option Styles
1831 @node Mnemonic Options
1832 @subsection Mnemonic Option Style
1834 @FIXME{have to decide whether or not to replace other occurrences of
1835 "mnemonic" with "long", or *ugh* vice versa.}
1837 Each option has at least one long (or mnemonic) name starting with two
1838 dashes in a row, e.g., @option{--list}. The long names are more clear than
1839 their corresponding short or old names. It sometimes happens that a
1840 single mnemonic option has many different different names which are
1841 synonymous, such as @option{--compare} and @option{--diff}. In addition,
1842 long option names can be given unique abbreviations. For example,
1843 @option{--cre} can be used in place of @option{--create} because there is no
1844 other mnemonic option which begins with @samp{cre}. (One way to find
1845 this out is by trying it and seeing what happens; if a particular
1846 abbreviation could represent more than one option, @command{tar} will tell
1847 you that that abbreviation is ambiguous and you'll know that that
1848 abbreviation won't work. You may also choose to run @samp{tar --help}
1849 to see a list of options. Be aware that if you run @command{tar} with a
1850 unique abbreviation for the long name of an option you didn't want to
1851 use, you are stuck; @command{tar} will perform the command as ordered.)
1853 Mnemonic options are meant to be obvious and easy to remember, and their
1854 meanings are generally easier to discern than those of their
1855 corresponding short options (see below). For example:
1858 $ @kbd{tar --create --verbose --blocking-factor=20 --file=/dev/rmt0}
1862 gives a fairly good set of hints about what the command does, even
1863 for those not fully acquainted with @command{tar}.
1865 Mnemonic options which require arguments take those arguments
1866 immediately following the option name. There are two ways of
1867 specifying a mandatory argument. It can be separated from the
1868 option name either by an equal sign, or by any amount of
1869 white space characters. For example, the @option{--file} option (which
1870 tells the name of the @command{tar} archive) is given a file such as
1871 @file{archive.tar} as argument by using any of the following notations:
1872 @option{--file=archive.tar} or @option{--file archive.tar}.
1874 In contrast, optional arguments must always be introduced using
1875 an equal sign. For example, the @option{--backup} option takes
1876 an optional argument specifying backup type. It must be used
1877 as @option{--backup=@var{backup-type}}.
1880 @subsection Short Option Style
1882 Most options also have a short option name. Short options start with
1883 a single dash, and are followed by a single character, e.g., @option{-t}
1884 (which is equivalent to @option{--list}). The forms are absolutely
1885 identical in function; they are interchangeable.
1887 The short option names are faster to type than long option names.
1889 Short options which require arguments take their arguments immediately
1890 following the option, usually separated by white space. It is also
1891 possible to stick the argument right after the short option name, using
1892 no intervening space. For example, you might write @w{@option{-f
1893 archive.tar}} or @option{-farchive.tar} instead of using
1894 @option{--file=archive.tar}. Both @option{--file=@var{archive-name}} and
1895 @w{@option{-f @var{archive-name}}} denote the option which indicates a
1896 specific archive, here named @file{archive.tar}.
1898 Short options which take optional arguments take their arguments
1899 immediately following the option letter, @emph{without any intervening
1900 white space characters}.
1902 Short options' letters may be clumped together, but you are not
1903 required to do this (as compared to old options; see below). When
1904 short options are clumped as a set, use one (single) dash for them
1905 all, e.g., @w{@samp{@command{tar} -cvf}}. Only the last option in
1906 such a set is allowed to have an argument@footnote{Clustering many
1907 options, the last of which has an argument, is a rather opaque way to
1908 write options. Some wonder if @acronym{GNU} @code{getopt} should not
1909 even be made helpful enough for considering such usages as invalid.}.
1911 When the options are separated, the argument for each option which requires
1912 an argument directly follows that option, as is usual for Unix programs.
1916 $ @kbd{tar -c -v -b 20 -f /dev/rmt0}
1919 If you reorder short options' locations, be sure to move any arguments
1920 that belong to them. If you do not move the arguments properly, you may
1921 end up overwriting files.
1924 @subsection Old Option Style
1927 Like short options, old options are single letters. However, old options
1928 must be written together as a single clumped set, without spaces separating
1929 them or dashes preceding them@footnote{Beware that if you precede options
1930 with a dash, you are announcing the short option style instead of the
1931 old option style; short options are decoded differently.}. This set
1932 of letters must be the first to appear on the command line, after the
1933 @command{tar} program name and some white space; old options cannot appear
1934 anywhere else. The letter of an old option is exactly the same letter as
1935 the corresponding short option. For example, the old option @samp{t} is
1936 the same as the short option @option{-t}, and consequently, the same as the
1937 mnemonic option @option{--list}. So for example, the command @w{@samp{tar
1938 cv}} specifies the option @option{-v} in addition to the operation @option{-c}.
1940 @FIXME{bob suggests having an uglier example. :-) }
1942 When options that need arguments are given together with the command,
1943 all the associated arguments follow, in the same order as the options.
1944 Thus, the example given previously could also be written in the old
1948 $ @kbd{tar cvbf 20 /dev/rmt0}
1952 Here, @samp{20} is the argument of @option{-b} and @samp{/dev/rmt0} is
1953 the argument of @option{-f}.
1955 On the other hand, this old style syntax makes it difficult to match
1956 option letters with their corresponding arguments, and is often
1957 confusing. In the command @w{@samp{tar cvbf 20 /dev/rmt0}}, for example,
1958 @samp{20} is the argument for @option{-b}, @samp{/dev/rmt0} is the
1959 argument for @option{-f}, and @option{-v} does not have a corresponding
1960 argument. Even using short options like in @w{@samp{tar -c -v -b 20 -f
1961 /dev/rmt0}} is clearer, putting all arguments next to the option they
1964 If you want to reorder the letters in the old option argument, be
1965 sure to reorder any corresponding argument appropriately.
1967 This old way of writing @command{tar} options can surprise even experienced
1968 users. For example, the two commands:
1971 @kbd{tar cfz archive.tar.gz file}
1972 @kbd{tar -cfz archive.tar.gz file}
1976 are quite different. The first example uses @file{archive.tar.gz} as
1977 the value for option @samp{f} and recognizes the option @samp{z}. The
1978 second example, however, uses @file{z} as the value for option
1979 @samp{f} --- probably not what was intended.
1981 Old options are kept for compatibility with old versions of @command{tar}.
1983 This second example could be corrected in many ways, among which the
1984 following are equivalent:
1987 @kbd{tar -czf archive.tar.gz file}
1988 @kbd{tar -cf archive.tar.gz -z file}
1989 @kbd{tar cf archive.tar.gz -z file}
1992 @FIXME{still could explain this better; it's redundant:}
1994 @cindex option syntax, traditional
1995 As far as we know, all @command{tar} programs, @acronym{GNU} and
1996 non-@acronym{GNU}, support old options. @GNUTAR{}
1997 supports them not only for historical reasons, but also because many
1998 people are used to them. For compatibility with Unix @command{tar},
1999 the first argument is always treated as containing command and option
2000 letters even if it doesn't start with @samp{-}. Thus, @samp{tar c} is
2001 equivalent to @w{@samp{tar -c}:} both of them specify the
2002 @option{--create} (@option{-c}) command to create an archive.
2005 @subsection Mixing Option Styles
2007 All three styles may be intermixed in a single @command{tar} command,
2008 so long as the rules for each style are fully
2009 respected@footnote{Before @GNUTAR{} version 1.11.6,
2010 a bug prevented intermixing old style options with mnemonic options in
2011 some cases.}. Old style options and either of the modern styles of
2012 options may be mixed within a single @command{tar} command. However,
2013 old style options must be introduced as the first arguments only,
2014 following the rule for old options (old options must appear directly
2015 after the @command{tar} command and some white space). Modern options
2016 may be given only after all arguments to the old options have been
2017 collected. If this rule is not respected, a modern option might be
2018 falsely interpreted as the value of the argument to one of the old
2021 For example, all the following commands are wholly equivalent, and
2022 illustrate the many combinations and orderings of option styles.
2025 @kbd{tar --create --file=archive.tar}
2026 @kbd{tar --create -f archive.tar}
2027 @kbd{tar --create -farchive.tar}
2028 @kbd{tar --file=archive.tar --create}
2029 @kbd{tar --file=archive.tar -c}
2030 @kbd{tar -c --file=archive.tar}
2031 @kbd{tar -c -f archive.tar}
2032 @kbd{tar -c -farchive.tar}
2033 @kbd{tar -cf archive.tar}
2034 @kbd{tar -cfarchive.tar}
2035 @kbd{tar -f archive.tar --create}
2036 @kbd{tar -f archive.tar -c}
2037 @kbd{tar -farchive.tar --create}
2038 @kbd{tar -farchive.tar -c}
2039 @kbd{tar c --file=archive.tar}
2040 @kbd{tar c -f archive.tar}
2041 @kbd{tar c -farchive.tar}
2042 @kbd{tar cf archive.tar}
2043 @kbd{tar f archive.tar --create}
2044 @kbd{tar f archive.tar -c}
2045 @kbd{tar fc archive.tar}
2048 On the other hand, the following commands are @emph{not} equivalent to
2052 @kbd{tar -f -c archive.tar}
2053 @kbd{tar -fc archive.tar}
2054 @kbd{tar -fcarchive.tar}
2055 @kbd{tar -farchive.tarc}
2056 @kbd{tar cfarchive.tar}
2060 These last examples mean something completely different from what the
2061 user intended (judging based on the example in the previous set which
2062 uses long options, whose intent is therefore very clear). The first
2063 four specify that the @command{tar} archive would be a file named
2064 @option{-c}, @samp{c}, @samp{carchive.tar} or @samp{archive.tarc},
2065 respectively. The first two examples also specify a single non-option,
2066 @var{name} argument having the value @samp{archive.tar}. The last
2067 example contains only old style option letters (repeating option
2068 @samp{c} twice), not all of which are meaningful (eg., @samp{.},
2069 @samp{h}, or @samp{i}), with no argument value. @FIXME{not sure i liked
2070 the first sentence of this paragraph..}
2073 @section All @command{tar} Options
2075 The coming manual sections contain an alphabetical listing of all
2076 @command{tar} operations and options, with brief descriptions and cross
2077 references to more in-depth explanations in the body of the manual.
2078 They also contain an alphabetically arranged table of the short option
2079 forms with their corresponding long option. You can use this table as
2080 a reference for deciphering @command{tar} commands in scripts.
2083 * Operation Summary::
2085 * Short Option Summary::
2088 @node Operation Summary
2089 @subsection Operations
2093 @opindex append, summary
2097 Appends files to the end of the archive. @xref{append}.
2099 @opindex catenate, summary
2103 Same as @option{--concatenate}. @xref{concatenate}.
2105 @opindex compare, summary
2109 Compares archive members with their counterparts in the file
2110 system, and reports differences in file size, mode, owner,
2111 modification date and contents. @xref{compare}.
2113 @opindex concatenate, summary
2117 Appends other @command{tar} archives to the end of the archive.
2120 @opindex create, summary
2124 Creates a new @command{tar} archive. @xref{create}.
2126 @opindex delete, summary
2129 Deletes members from the archive. Don't try this on a archive on a
2130 tape! @xref{delete}.
2132 @opindex diff, summary
2136 Same @option{--compare}. @xref{compare}.
2138 @opindex extract, summary
2142 Extracts members from the archive into the file system. @xref{extract}.
2144 @opindex get, summary
2148 Same as @option{--extract}. @xref{extract}.
2150 @opindex list, summary
2154 Lists the members in an archive. @xref{list}.
2156 @opindex update, summary
2160 @FIXME{It was: A combination of the @option{--compare} and
2161 @option{--append} operations. This is not true and rather misleading,
2162 as @option{--compare} (@option{--diff}, @option{-d}) does a lot more than @option{--update} (@option{-u}) for
2163 ensuring files are identical.} Adds files to the end of the archive,
2164 but only if they are newer than their counterparts already in the
2165 archive, or if they do not already exist in the archive.
2170 @node Option Summary
2171 @subsection @command{tar} Options
2175 @opindex absolute-names, summary
2176 @item --absolute-names
2179 Normally when creating an archive, @command{tar} strips an initial
2180 @samp{/} from member names. This option disables that behavior.
2183 @opindex after-date, summary
2186 (See @option{--newer}, @pxref{after})
2188 @opindex anchored, summary
2191 A pattern must match an initial subsequence of the name's components.
2192 @xref{controlling pattern-matching}.
2194 @opindex atime-preserve, summary
2195 @item --atime-preserve
2196 @itemx --atime-preserve=replace
2197 @itemx --atime-preserve=system
2199 Attempt to preserve the access time of files when reading them. This
2200 option currently is effective only on files that you own, unless you
2201 have superuser privileges.
2203 @option{--atime-preserve=replace} remembers the access time of a file
2204 before reading it, and then restores the access time afterwards. This
2205 may cause problems if other programs are reading the file at the same
2206 time, as the times of their accesses will be lost. On most platforms
2207 restoring the access time also requires @command{tar} to restore the
2208 data modification time too, so this option may also cause problems if
2209 other programs are writing the file at the same time. (Tar attempts
2210 to detect this situation, but cannot do so reliably due to race
2211 conditions.) Worse, on most platforms restoring the access time also
2212 updates the status change time, which means that this option is
2213 incompatible with incremental backups.
2215 @option{--atime-preserve=system} avoids changing time stamps on files,
2216 without interfering with time stamp updates
2217 caused by other programs, so it works better with incremental backups.
2218 However, it requires a special @code{O_NOATIME} option from the
2219 underlying operating and file system implementation, and it also requires
2220 that searching directories does not update their access times. As of
2221 this writing (November 2005) this works only with Linux, and only with
2222 Linux kernels 2.6.8 and later. Worse, there is currently no reliable
2223 way to know whether this feature actually works. Sometimes
2224 @command{tar} knows that it does not work, and if you use
2225 @option{--atime-preserve=system} then @command{tar} complains and
2226 exits right away. But other times @command{tar} might think that the
2227 option works when it actually does not.
2229 Currently @option{--atime-preserve} with no operand defaults to
2230 @option{--atime-preserve=replace}, but this may change in the future
2231 as support for @option{--atime-preserve=system} improves.
2233 If your operating system does not support
2234 @option{--atime-preserve=system}, you might be able to preserve access
2235 times reliably by by using the @command{mount} command. For example,
2236 you can mount the file system read-only, or access the file system via
2237 a read-only loopback mount, or use the @samp{noatime} mount option
2238 available on some systems. However, mounting typically requires
2239 superuser privileges and can be a pain to manage.
2241 @opindex backup, summary
2242 @item --backup=@var{backup-type}
2244 Rather than deleting files from the file system, @command{tar} will
2245 back them up using simple or numbered backups, depending upon
2246 @var{backup-type}. @xref{backup}.
2248 @opindex block-number, summary
2249 @item --block-number
2252 With this option present, @command{tar} prints error messages for read errors
2253 with the block number in the archive file. @xref{block-number}.
2255 @opindex blocking-factor, summary
2256 @item --blocking-factor=@var{blocking}
2257 @itemx -b @var{blocking}
2259 Sets the blocking factor @command{tar} uses to @var{blocking} x 512 bytes per
2260 record. @xref{Blocking Factor}.
2262 @opindex bzip2, summary
2266 This option tells @command{tar} to read or write archives through
2267 @code{bzip2}. @xref{gzip}.
2269 @opindex checkpoint, summary
2272 This option directs @command{tar} to print periodic checkpoint messages as it
2273 reads through the archive. Its intended for when you want a visual
2274 indication that @command{tar} is still running, but don't want to see
2275 @option{--verbose} output. @FIXME-xref{}
2277 @opindex check-links, summary
2280 If this option was given, @command{tar} will check the number of links
2281 dumped for each processed file. If this number does not match the
2282 total number of hard links for the file, a warning message will be
2283 output @footnote{Earlier versions of @GNUTAR{} understood @option{-l} as a
2284 synonym for @option{--one-file-system}. The current semantics, wich
2285 complies to UNIX98, was introduced with version
2286 1.15.91. @xref{Changes}, for more information.}.
2288 @opindex compress, summary
2289 @opindex uncompress, summary
2294 @command{tar} will use the @command{compress} program when reading or
2295 writing the archive. This allows you to directly act on archives
2296 while saving space. @xref{gzip}.
2298 @opindex confirmation, summary
2299 @item --confirmation
2301 (See @option{--interactive}.) @xref{interactive}.
2303 @opindex delay-directory-restore, summary
2304 @item --delay-directory-restore
2306 Delay setting modification times and permissions of extracted
2307 directories until the end of extraction. @xref{Directory Modification Times and Permissions}.
2309 @opindex dereference, summary
2313 When creating a @command{tar} archive, @command{tar} will archive the
2314 file that a symbolic link points to, rather than archiving the
2315 symlink. @xref{dereference}.
2317 @opindex directory, summary
2318 @item --directory=@var{dir}
2321 When this option is specified, @command{tar} will change its current directory
2322 to @var{dir} before performing any operations. When this option is used
2323 during archive creation, it is order sensitive. @xref{directory}.
2325 @opindex exclude, summary
2326 @item --exclude=@var{pattern}
2328 When performing operations, @command{tar} will skip files that match
2329 @var{pattern}. @xref{exclude}.
2331 @opindex exclude-from, summary
2332 @item --exclude-from=@var{file}
2333 @itemx -X @var{file}
2335 Similar to @option{--exclude}, except @command{tar} will use the list of
2336 patterns in the file @var{file}. @xref{exclude}.
2338 @opindex exclude-caches, summary
2339 @item --exclude-caches
2341 Automatically excludes all directories
2342 containing a cache directory tag. @xref{exclude}.
2344 @opindex file, summary
2345 @item --file=@var{archive}
2346 @itemx -f @var{archive}
2348 @command{tar} will use the file @var{archive} as the @command{tar} archive it
2349 performs operations on, rather than @command{tar}'s compilation dependent
2350 default. @xref{file tutorial}.
2352 @opindex files-from, summary
2353 @item --files-from=@var{file}
2354 @itemx -T @var{file}
2356 @command{tar} will use the contents of @var{file} as a list of archive members
2357 or files to operate on, in addition to those specified on the
2358 command-line. @xref{files}.
2360 @opindex force-local, summary
2363 Forces @command{tar} to interpret the filename given to @option{--file}
2364 as a local file, even if it looks like a remote tape drive name.
2365 @xref{local and remote archives}.
2367 @opindex format, summary
2368 @item --format=@var{format}
2370 Selects output archive format. @var{Format} may be one of the
2375 Creates an archive that is compatible with Unix V7 @command{tar}.
2378 Creates an archive that is compatible with GNU @command{tar} version
2382 Creates archive in GNU tar 1.13 format. Basically it is the same as
2383 @samp{oldgnu} with the only difference in the way it handles long
2387 Creates a @acronym{POSIX.1-1988} compatible archive.
2390 Creates a @acronym{POSIX.1-2001 archive}.
2394 @xref{Formats}, for a detailed discussion of these formats.
2396 @opindex group, summary
2397 @item --group=@var{group}
2399 Files added to the @command{tar} archive will have a group id of @var{group},
2400 rather than the group from the source file. @var{group} is first decoded
2401 as a group symbolic name, but if this interpretation fails, it has to be
2402 a decimal numeric group ID. @FIXME-xref{}
2404 Also see the comments for the @option{--owner=@var{user}} option.
2406 @opindex gzip, summary
2407 @opindex gunzip, summary
2408 @opindex ungzip, summary
2414 This option tells @command{tar} to read or write archives through
2415 @command{gzip}, allowing @command{tar} to directly operate on several
2416 kinds of compressed archives transparently. @xref{gzip}.
2418 @opindex help, summary
2421 @command{tar} will print out a short message summarizing the operations and
2422 options to @command{tar} and exit. @xref{help}.
2424 @opindex ignore-case, summary
2427 Ignore case when matching member or file names with patterns. @xref{controlling pattern-matching}.
2429 @opindex ignore-command-error, summary
2430 @item --ignore-command-error
2431 Ignore exit codes of subprocesses. @xref{Writing to an External Program}.
2433 @opindex ignore-failed-read, summary
2434 @item --ignore-failed-read
2436 Do not exit unsuccessfully merely because an unreadable file was encountered.
2439 @opindex ignore-zeros, summary
2440 @item --ignore-zeros
2443 With this option, @command{tar} will ignore zeroed blocks in the
2444 archive, which normally signals EOF. @xref{Reading}.
2446 @opindex incremental, summary
2450 Used to inform @command{tar} that it is working with an old
2451 @acronym{GNU}-format incremental backup archive. It is intended
2452 primarily for backwards compatibility only. @FIXME{incremental and
2453 listed-incremental}.
2455 @opindex index-file, summary
2456 @item --index-file=@var{file}
2458 Send verbose output to @var{file} instead of to standard output.
2460 @opindex info-script, summary
2461 @opindex new-volume-script, summary
2462 @item --info-script=@var{script-file}
2463 @itemx --new-volume-script=@var{script-file}
2464 @itemx -F @var{script-file}
2466 When @command{tar} is performing multi-tape backups, @var{script-file} is run
2467 at the end of each tape. If @var{script-file} exits with nonzero status,
2468 @command{tar} fails immediately. @xref{info-script}, for a detailed
2469 discussion of @var{script-file}.
2471 @opindex interactive, summary
2473 @itemx --confirmation
2476 Specifies that @command{tar} should ask the user for confirmation before
2477 performing potentially destructive options, such as overwriting files.
2480 @opindex keep-newer-files, summary
2481 @item --keep-newer-files
2483 Do not replace existing files that are newer than their archive copies
2484 when extracting files from an archive.
2486 @opindex keep-old-files, summary
2487 @item --keep-old-files
2490 Do not overwrite existing files when extracting files from an archive.
2491 @xref{Keep Old Files}.
2493 @opindex label, summary
2494 @item --label=@var{name}
2495 @itemx -V @var{name}
2497 When creating an archive, instructs @command{tar} to write @var{name}
2498 as a name record in the archive. When extracting or listing archives,
2499 @command{tar} will only operate on archives that have a label matching
2500 the pattern specified in @var{name}. @xref{Tape Files}.
2502 @opindex listed-incremental, summary
2503 @item --listed-incremental=@var{snapshot-file}
2504 @itemx -g @var{snapshot-file}
2506 During a @option{--create} operation, specifies that the archive that
2507 @command{tar} creates is a new @acronym{GNU}-format incremental
2508 backup, using @var{snapshot-file} to determine which files to backup.
2509 With other operations, informs @command{tar} that the archive is in
2510 incremental format. @FIXME{incremental and listed-incremental}.
2512 @opindex mode, summary
2513 @item --mode=@var{permissions}
2515 When adding files to an archive, @command{tar} will use
2516 @var{permissions} for the archive members, rather than the permissions
2517 from the files. The program @command{chmod} and this @command{tar}
2518 option share the same syntax for what @var{permissions} might be.
2519 @xref{File permissions, Permissions, File permissions, fileutils,
2520 @acronym{GNU} file utilities}. This reference also has useful
2521 information for those not being overly familiar with the Unix
2524 Of course, @var{permissions} might be plainly specified as an octal number.
2525 However, by using generic symbolic modifications to mode bits, this allows
2526 more flexibility. For example, the value @samp{a+rw} adds read and write
2527 permissions for everybody, while retaining executable bits on directories
2528 or on any other file already marked as executable.
2530 @opindex multi-volume, summary
2531 @item --multi-volume
2534 Informs @command{tar} that it should create or otherwise operate on a
2535 multi-volume @command{tar} archive. @xref{Using Multiple Tapes}.
2537 @opindex new-volume-script, summary
2538 @item --new-volume-script
2542 @opindex seek, summary
2546 Assume that the archive media supports seeks to arbitrary
2547 locations. Usually @command{tar} determines automatically whether
2548 the archive can be seeked or not. This option is intended for use
2549 in cases when such recognition fails.
2551 @opindex newer, summary
2552 @item --newer=@var{date}
2553 @itemx --after-date=@var{date}
2556 When creating an archive, @command{tar} will only add files that have changed
2557 since @var{date}. If @var{date} begins with @samp{/} or @samp{.}, it
2558 is taken to be the name of a file whose data modification time specifies
2559 the date. @xref{after}.
2561 @opindex newer-mtime, summary
2562 @item --newer-mtime=@var{date}
2564 Like @option{--newer}, but add only files whose
2565 contents have changed (as opposed to just @option{--newer}, which will
2566 also back up files for which any status information has changed).
2568 @opindex no-anchored, summary
2571 An exclude pattern can match any subsequence of the name's components.
2572 @xref{controlling pattern-matching}.
2574 @opindex no-delay-directory-restore, summary
2575 @item --no-delay-directory-restore
2577 Setting modification times and permissions of extracted
2578 directories when all files from this directory has been
2579 extracted. This is the default. @xref{Directory Modification Times and Permissions}.
2581 @opindex no-ignore-case, summary
2582 @item --no-ignore-case
2584 Use case-sensitive matching.
2585 @xref{controlling pattern-matching}.
2587 @opindex no-ignore-command-error, summary
2588 @item --no-ignore-command-error
2589 Print warnings about subprocesses terminated with a non-zero exit
2590 code. @xref{Writing to an External Program}.
2592 @opindex no-quote-chars, summary
2593 @item --no-quote-chars=@var{string}
2594 Do not quote characters from @var{string}, even if the selected
2595 quoting style implies they should be quoted (@FIXME-pxref{Quoting Styles}).
2597 @opindex no-recursion, summary
2598 @item --no-recursion
2600 With this option, @command{tar} will not recurse into directories.
2603 @opindex no-same-owner, summary
2604 @item --no-same-owner
2607 When extracting an archive, do not attempt to preserve the owner
2608 specified in the @command{tar} archive. This the default behavior
2611 @opindex no-same-permissions, summary
2612 @item --no-same-permissions
2614 When extracting an archive, subtract the user's umask from files from
2615 the permissions specified in the archive. This is the default behavior
2618 @opindex no-wildcards, summary
2619 @item --no-wildcards
2621 Do not use wildcards.
2622 @xref{controlling pattern-matching}.
2624 @opindex no-wildcards-match-slash, summary
2625 @item --no-wildcards-match-slash
2627 Wildcards do not match @samp{/}.
2628 @xref{controlling pattern-matching}.
2630 @opindex null, summary
2633 When @command{tar} is using the @option{--files-from} option, this option
2634 instructs @command{tar} to expect filenames terminated with @option{NUL}, so
2635 @command{tar} can correctly work with file names that contain newlines.
2638 @opindex numeric-owner, summary
2639 @item --numeric-owner
2641 This option will notify @command{tar} that it should use numeric user
2642 and group IDs when creating a @command{tar} file, rather than names.
2646 When extracting files, this option is a synonym for
2647 @option{--no-same-owner}, i.e. it prevents @command{tar} from
2648 restoring ownership of files being extracted.
2650 When creating an archive, @option{-o} is a synonym for
2651 @option{--old-archive}. This behavior is for compatibility
2652 with previous versions of @GNUTAR{}, and will be
2653 removed in the future releases.
2655 @xref{Changes}, for more information.
2657 @opindex occurrence, summary
2658 @item --occurrence[=@var{number}]
2660 This option can be used in conjunction with one of the subcommands
2661 @option{--delete}, @option{--diff}, @option{--extract} or
2662 @option{--list} when a list of files is given either on the command
2663 line or via @option{-T} option.
2665 This option instructs @command{tar} to process only the @var{number}th
2666 occurrence of each named file. @var{Number} defaults to 1, so
2669 tar -x -f archive.tar --occurrence filename
2673 will extract the first occurrence of @file{filename} from @file{archive.tar}
2674 and will terminate without scanning to the end of the archive.
2676 @opindex old-archive, summary
2678 Synonym for @option{--format=v7}.
2680 @opindex one-file-system, summary
2681 @item --one-file-system
2682 Used when creating an archive. Prevents @command{tar} from recursing into
2683 directories that are on different file systems from the current
2684 directory @footnote{Earlier versions of @GNUTAR{} understood @option{-l} as a
2685 synonym for @option{--one-file-system}. This has changed in version
2686 1.15.91. @xref{Changes}, for more information.}.
2688 @opindex overwrite, summary
2691 Overwrite existing files and directory metadata when extracting files
2692 from an archive. @xref{Overwrite Old Files}.
2694 @opindex overwrite-dir, summary
2695 @item --overwrite-dir
2697 Overwrite the metadata of existing directories when extracting files
2698 from an archive. @xref{Overwrite Old Files}.
2700 @opindex owner, summary
2701 @item --owner=@var{user}
2703 Specifies that @command{tar} should use @var{user} as the owner of members
2704 when creating archives, instead of the user associated with the source
2705 file. @var{user} is first decoded as a user symbolic name, but if
2706 this interpretation fails, it has to be a decimal numeric user ID.
2709 There is no value indicating a missing number, and @samp{0} usually means
2710 @code{root}. Some people like to force @samp{0} as the value to offer in
2711 their distributions for the owner of files, because the @code{root} user is
2712 anonymous anyway, so that might as well be the owner of anonymous archives.
2714 This option does not affect extraction from archives.
2716 @opindex quote-chars, summary
2717 @item --quote-chars=@var{string}
2718 Always quote characters from @var{string}, even if the selected
2719 quoting style would not quote them (@FIXME-pxref{Quoting Styles}).
2721 @opindex quoting-style, summary
2722 @item --quoting-style=@var{style}
2723 Set quoting style to use when printing member and file names
2724 (@FIXME-pxref{Quoting Styles}). Valid @var{style} values are:
2725 @code{literal}, @code{shell}, @code{shell-always}, @code{c},
2726 @code{escape}, @code{locale}, and @code{clocale}. Default quoting
2727 style is @code{escape}, unless overridden while configuring the
2730 @opindex pax-option, summary
2731 @item --pax-option=@var{keyword-list}
2732 @FIXME{Such a detailed description does not belong there, move it elsewhere.}
2733 This option is meaningful only with @acronym{POSIX.1-2001} archives
2734 (@pxref{posix}). It modifies the way @command{tar} handles the
2735 extended header keywords. @var{Keyword-list} is a comma-separated
2736 list of keyword options, each keyword option taking one of
2737 the following forms:
2740 @item delete=@var{pattern}
2741 When used with one of archive-creation commands,
2742 this option instructs @command{tar} to omit from extended header records
2743 that it produces any keywords matching the string @var{pattern}.
2745 When used in extract or list mode, this option instructs tar
2746 to ignore any keywords matching the given @var{pattern} in the extended
2747 header records. In both cases, matching is performed using the pattern
2748 matching notation described in @acronym{POSIX 1003.2}, 3.13
2749 (See @cite{glob(7)}). For example:
2752 --pax-option delete=security.*
2755 would suppress security-related information.
2757 @item exthdr.name=@var{string}
2759 This keyword allows user control over the name that is written into the
2760 ustar header blocks for the extended headers. The name is obtained
2761 from @var{string} after making the following substitutions:
2763 @multitable @columnfractions .30 .70
2764 @headitem Meta-character @tab Replaced By
2765 @item %d @tab The directory name of the file, equivalent to the
2766 result of the @command{dirname} utility on the translated pathname.
2767 @item %f @tab The filename of the file, equivalent to the result
2768 of the @command{basename} utility on the translated pathname.
2769 @item %p @tab The process ID of the @command{tar} process.
2770 @item %% @tab A @samp{%} character.
2773 Any other @samp{%} characters in @var{string} produce undefined
2776 If no option @samp{exthdr.name=string} is specified, @command{tar}
2777 will use the following default value:
2783 @item globexthdr.name=@var{string}
2784 This keyword allows user control over the name that is written into
2785 the ustar header blocks for global extended header records. The name
2786 is obtained from the contents of @var{string}, after making
2787 the following substitutions:
2789 @multitable @columnfractions .30 .70
2790 @headitem Meta-character @tab Replaced By
2791 @item %n @tab An integer that represents the
2792 sequence number of the global extended header record in the archive,
2794 @item %p @tab The process ID of the @command{tar} process.
2795 @item %% @tab A @samp{%} character.
2798 Any other @samp{%} characters in @var{string} produce undefined results.
2800 If no option @samp{globexthdr.name=string} is specified, @command{tar}
2801 will use the following default value:
2804 $TMPDIR/GlobalHead.%p.%n
2808 where @samp{$TMPDIR} represents the value of the @var{TMPDIR}
2809 environment variable. If @var{TMPDIR} is not set, @command{tar}
2812 @item @var{keyword}=@var{value}
2813 When used with one of archive-creation commands, these keyword/value pairs
2814 will be included at the beginning of the archive in a global extended
2815 header record. When used with one of archive-reading commands,
2816 @command{tar} will behave as if it has encountered these keyword/value
2817 pairs at the beginning of the archive in a global extended header
2820 @item @var{keyword}:=@var{value}
2821 When used with one of archive-creation commands, these keyword/value pairs
2822 will be included as records at the beginning of an extended header for
2823 each file. This is effectively equivalent to @var{keyword}=@var{value}
2824 form except that it creates no global extended header records.
2826 When used with one of archive-reading commands, @command{tar} will
2827 behave as if these keyword/value pairs were included as records at the
2828 end of each extended header; thus, they will override any global or
2829 file-specific extended header record keywords of the same names.
2830 For example, in the command:
2833 tar --format=posix --create \
2834 --file archive --pax-option gname:=user .
2837 the group name will be forced to a new value for all files
2838 stored in the archive.
2841 @opindex portability, summary
2843 @itemx --old-archive
2844 Synonym for @option{--format=v7}.
2846 @opindex posix, summary
2848 Same as @option{--format=posix}.
2850 @opindex preserve, summary
2853 Synonymous with specifying both @option{--preserve-permissions} and
2854 @option{--same-order}. @xref{Setting Access Permissions}.
2856 @opindex preserve-order, summary
2857 @item --preserve-order
2859 (See @option{--same-order}; @pxref{Reading}.)
2861 @opindex preserve-permissions, summary
2862 @opindex same-permissions, summary
2863 @item --preserve-permissions
2864 @itemx --same-permissions
2867 When @command{tar} is extracting an archive, it normally subtracts the
2868 users' umask from the permissions specified in the archive and uses
2869 that number as the permissions to create the destination file.
2870 Specifying this option instructs @command{tar} that it should use the
2871 permissions directly from the archive. @xref{Setting Access Permissions}.
2873 @opindex read-full-records, summary
2874 @item --read-full-records
2877 Specifies that @command{tar} should reblock its input, for reading
2878 from pipes on systems with buggy implementations. @xref{Reading}.
2880 @opindex record-size, summary
2881 @item --record-size=@var{size}
2883 Instructs @command{tar} to use @var{size} bytes per record when accessing the
2884 archive. @xref{Blocking Factor}.
2886 @opindex recursion, summary
2889 With this option, @command{tar} recurses into directories.
2892 @opindex recursive-unlink, summary
2893 @item --recursive-unlink
2896 directory hierarchies before extracting directories of the same name
2897 from the archive. @xref{Recursive Unlink}.
2899 @opindex remove-files, summary
2900 @item --remove-files
2902 Directs @command{tar} to remove the source file from the file system after
2903 appending it to an archive. @xref{remove files}.
2905 @opindex restrict, summary
2908 Disable use of some potentially harmful @command{tar} options.
2909 Currently this option disables shell invocaton from multi-volume menu
2910 (@pxref{Using Multiple Tapes}).
2912 @opindex rmt-command, summary
2913 @item --rmt-command=@var{cmd}
2915 Notifies @command{tar} that it should use @var{cmd} instead of
2916 the default @file{/usr/libexec/rmt} (@pxref{Remote Tape Server}).
2918 @opindex rsh-command, summary
2919 @item --rsh-command=@var{cmd}
2921 Notifies @command{tar} that is should use @var{cmd} to communicate with remote
2922 devices. @xref{Device}.
2924 @opindex same-order, summary
2926 @itemx --preserve-order
2929 This option is an optimization for @command{tar} when running on machines with
2930 small amounts of memory. It informs @command{tar} that the list of file
2931 arguments has already been sorted to match the order of files in the
2932 archive. @xref{Reading}.
2934 @opindex same-owner, summary
2937 When extracting an archive, @command{tar} will attempt to preserve the owner
2938 specified in the @command{tar} archive with this option present.
2939 This is the default behavior for the superuser; this option has an
2940 effect only for ordinary users. @xref{Attributes}.
2942 @opindex same-permissions, summary
2943 @item --same-permissions
2945 (See @option{--preserve-permissions}; @pxref{Setting Access Permissions}.)
2947 @opindex show-defaults, summary
2948 @item --show-defaults
2950 Displays the default options used by @command{tar} and exits
2951 successfully. This option is intended for use in shell scripts.
2952 Here is an example of what you can see using this option:
2955 $ tar --show-defaults
2956 --format=gnu -f- -b20 --quoting-style=escape \
2957 --rmt-command=/usr/libexec/rmt --rsh-command=/usr/bin/rsh
2960 @opindex show-omitted-dirs, summary
2961 @item --show-omitted-dirs
2963 Instructs @command{tar} to mention directories its skipping over when
2964 operating on a @command{tar} archive. @xref{show-omitted-dirs}.
2966 @opindex show-stored-names, summary
2967 @item --show-stored-names
2969 This option has effect only when used in conjunction with one of
2970 archive creation operations. It instructs tar to list the member names
2971 stored in the archive, as opposed to the actual file
2972 names. @xref{listing member and file names}.
2974 @opindex sparse, summary
2978 Invokes a @acronym{GNU} extension when adding files to an archive that handles
2979 sparse files efficiently. @xref{sparse}.
2981 @opindex starting-file, summary
2982 @item --starting-file=@var{name}
2983 @itemx -K @var{name}
2985 This option affects extraction only; @command{tar} will skip extracting
2986 files in the archive until it finds one that matches @var{name}.
2989 @opindex strip-components, summary
2990 @item --strip-components=@var{number}
2991 Strip given @var{number} of leading components from file names before
2992 extraction.@footnote{This option was called @option{--strip-path} in
2993 version 1.14.} For example, if archive @file{archive.tar} contained
2994 @file{/some/file/name}, then running
2997 tar --extract --file archive.tar --strip-components=2
3001 would extracted this file to file @file{name}.
3003 @opindex suffix, summary
3004 @item --suffix=@var{suffix}
3006 Alters the suffix @command{tar} uses when backing up files from the default
3007 @samp{~}. @xref{backup}.
3009 @opindex tape-length, summary
3010 @item --tape-length=@var{num}
3013 Specifies the length of tapes that @command{tar} is writing as being
3014 @w{@var{num} x 1024} bytes long. @xref{Using Multiple Tapes}.
3016 @opindex test-label, summary
3019 Reads the volume label. If an argument is specified, test whether it
3020 matches the volume label. @xref{--test-label option}.
3022 @opindex to-command, summary
3023 @item --to-command=@var{command}
3025 During extraction @command{tar} will pipe extracted files to the
3026 standard input of @var{command}. @xref{Writing to an External Program}.
3028 @opindex to-stdout, summary
3032 During extraction, @command{tar} will extract files to stdout rather
3033 than to the file system. @xref{Writing to Standard Output}.
3035 @opindex totals, summary
3038 Displays the total number of bytes written after creating an archive.
3041 @opindex touch, summary
3045 Sets the data modification time of extracted files to the extraction time,
3046 rather than the data modification time stored in the archive.
3047 @xref{Data Modification Times}.
3049 @opindex uncompress, summary
3052 (See @option{--compress}. @pxref{gzip})
3054 @opindex ungzip, summary
3057 (See @option{--gzip}. @pxref{gzip})
3059 @opindex unlink-first, summary
3060 @item --unlink-first
3063 Directs @command{tar} to remove the corresponding file from the file
3064 system before extracting it from the archive. @xref{Unlink First}.
3066 @opindex use-compress-program, summary
3067 @item --use-compress-program=@var{prog}
3069 Instructs @command{tar} to access the archive through @var{prog}, which is
3070 presumed to be a compression program of some sort. @xref{gzip}.
3072 @opindex utc, summary
3075 Display file modification dates in @acronym{UTC}. This option implies
3078 @opindex verbose, summary
3082 Specifies that @command{tar} should be more verbose about the operations its
3083 performing. This option can be specified multiple times for some
3084 operations to increase the amount of information displayed.
3087 @opindex verify, summary
3091 Verifies that the archive was correctly written when creating an
3092 archive. @xref{verify}.
3094 @opindex version, summary
3097 Print information about the program's name, version, origin and legal
3098 status, all on standard output, and then exit successfully.
3101 @opindex volno-file, summary
3102 @item --volno-file=@var{file}
3104 Used in conjunction with @option{--multi-volume}. @command{tar} will keep track
3105 of which volume of a multi-volume archive its working in @var{file}.
3108 @opindex wildcards, summary
3111 Use wildcards when excluding files.
3112 @xref{controlling pattern-matching}.
3114 @opindex wildcards-match-slash, summary
3115 @item --wildcards-match-slash
3117 Wildcards match @samp{/}.
3118 @xref{controlling pattern-matching}.
3121 @node Short Option Summary
3122 @subsection Short Options Cross Reference
3124 Here is an alphabetized list of all of the short option forms, matching
3125 them with the equivalent long option.
3131 @option{--concatenate}
3135 @option{--read-full-records}
3139 @option{--directory}
3143 @option{--info-script}
3147 @option{--incremental}
3151 @option{--starting-file}
3155 @option{--tape-length}
3159 @option{--multi-volume}
3167 @option{--to-stdout}
3171 @option{--absolute-names}
3175 @option{--block-number}
3183 @option{--files-from}
3187 @option{--unlink-first}
3199 @option{--exclude-from}
3207 @option{--blocking-factor}
3223 @option{--listed-incremental}
3227 @option{--dereference}
3231 @option{--ignore-zeros}
3239 @option{--keep-old-files}
3243 @option{--one-file-system}. Use of this short option is deprecated. It
3244 is retained for compatibility with the earlier versions of GNU
3245 @command{tar}, and will be changed in future releases.
3247 @xref{Changes}, for more information.
3255 When creating --- @option{--no-same-owner}, when extracting ---
3256 @option{--portability}.
3258 The later usage is deprecated. It is retained for compatibility with
3259 the earlier versions of @GNUTAR{}. In the future releases
3260 @option{-o} will be equivalent to @option{--no-same-owner} only.
3264 @option{--preserve-permissions}
3272 @option{--same-order}
3288 @option{--interactive}
3301 @section @GNUTAR{} documentation
3303 @cindex Getting program version number
3305 @cindex Version of the @command{tar} program
3306 Being careful, the first thing is really checking that you are using
3307 @GNUTAR{}, indeed. The @option{--version} option
3308 causes @command{tar} to print information about its name, version,
3309 origin and legal status, all on standard output, and then exit
3310 successfully. For example, @w{@samp{tar --version}} might print:
3313 tar (GNU tar) 1.15.2
3314 Copyright (C) 2006 Free Software Foundation, Inc.
3315 This is free software. You may redistribute copies of it under the terms of
3316 the GNU General Public License <http://www.gnu.org/licenses/gpl.html>.
3317 There is NO WARRANTY, to the extent permitted by law.
3319 Written by John Gilmore and Jay Fenlason.
3323 The first occurrence of @samp{tar} in the result above is the program
3324 name in the package (for example, @command{rmt} is another program),
3325 while the second occurrence of @samp{tar} is the name of the package
3326 itself, containing possibly many programs. The package is currently
3327 named @samp{tar}, after the name of the main program it
3328 contains@footnote{There are plans to merge the @command{cpio} and
3329 @command{tar} packages into a single one which would be called
3330 @code{paxutils}. So, who knows if, one of this days, the
3331 @option{--version} would not output @w{@samp{tar (@acronym{GNU}
3334 @cindex Obtaining help
3335 @cindex Listing all @command{tar} options
3336 @opindex help, introduction
3337 Another thing you might want to do is checking the spelling or meaning
3338 of some particular @command{tar} option, without resorting to this
3339 manual, for once you have carefully read it. @GNUTAR{}
3340 has a short help feature, triggerable through the
3341 @option{--help} option. By using this option, @command{tar} will
3342 print a usage message listing all available options on standard
3343 output, then exit successfully, without doing anything else and
3344 ignoring all other options. Even if this is only a brief summary, it
3345 may be several screens long. So, if you are not using some kind of
3346 scrollable window, you might prefer to use something like:
3349 $ @kbd{tar --help | less}
3353 presuming, here, that you like using @command{less} for a pager. Other
3354 popular pagers are @command{more} and @command{pg}. If you know about some
3355 @var{keyword} which interests you and do not want to read all the
3356 @option{--help} output, another common idiom is doing:
3359 tar --help | grep @var{keyword}
3363 for getting only the pertinent lines. Notice, however, that some
3364 @command{tar} options have long description lines and the above
3365 command will list only the first of them.
3367 The exact look of the option summary displayed by @kbd{tar --help} is
3368 configurable. @xref{Configuring Help Summary}, for a detailed description.
3371 If you only wish to check the spelling of an option, running @kbd{tar
3372 --usage} may be a better choice. This will display a terse list of
3373 @command{tar} option without accompanying explanations.
3375 The short help output is quite succinct, and you might have to get
3376 back to the full documentation for precise points. If you are reading
3377 this paragraph, you already have the @command{tar} manual in some
3378 form. This manual is available in a variety of forms from
3379 @url{http://www.gnu.org/software/tar/manual}. It may printed out of the @GNUTAR{}
3380 distribution, provided you have @TeX{} already installed somewhere,
3381 and a laser printer around. Just configure the distribution, execute
3382 the command @w{@samp{make dvi}}, then print @file{doc/tar.dvi} the
3383 usual way (contact your local guru to know how). If @GNUTAR{}
3384 has been conveniently installed at your place, this
3385 manual is also available in interactive, hypertextual form as an Info
3386 file. Just call @w{@samp{info tar}} or, if you do not have the
3387 @command{info} program handy, use the Info reader provided within
3388 @acronym{GNU} Emacs, calling @samp{tar} from the main Info menu.
3390 There is currently no @code{man} page for @GNUTAR{}.
3391 If you observe such a @code{man} page on the system you are running,
3392 either it does not belong to @GNUTAR{}, or it has not
3393 been produced by @acronym{GNU}. Some package maintainers convert
3394 @kbd{tar --help} output to a man page, using @command{help2man}. In
3395 any case, please bear in mind that the authoritative source of
3396 information about @GNUTAR{} is this Texinfo documentation.
3399 @section Obtaining @GNUTAR{} default values
3401 @opindex show-defaults
3402 @GNUTAR{} has some predefined defaults that are used when you do not
3403 explicitely specify another values. To obtain a list of such
3404 defaults, use @option{--show-defaults} option. This will output the
3405 values in the form of @command{tar} command line options:
3409 @kbd{tar --show-defaults}
3410 --format=gnu -f- -b20 --rmt-command=/etc/rmt --rsh-command=/usr/bin/rsh
3415 The above output shows that this version of @GNUTAR{} defaults to
3416 using @samp{gnu} archive format (@pxref{Formats}), it uses standard
3417 output as the archive, if no @option{--file} option has been given
3418 (@pxref{file tutorial}), the default blocking factor is 20
3419 (@pxref{Blocking Factor}). It also shows the default locations where
3420 @command{tar} will look for @command{rmt} and @command{rsh} binaries.
3423 @section Checking @command{tar} progress
3425 Typically, @command{tar} performs most operations without reporting any
3426 information to the user except error messages. When using @command{tar}
3427 with many options, particularly ones with complicated or
3428 difficult-to-predict behavior, it is possible to make serious mistakes.
3429 @command{tar} provides several options that make observing @command{tar}
3430 easier. These options cause @command{tar} to print information as it
3431 progresses in its job, and you might want to use them just for being
3432 more careful about what is going on, or merely for entertaining
3433 yourself. If you have encountered a problem when operating on an
3434 archive, however, you may need more information than just an error
3435 message in order to solve the problem. The following options can be
3436 helpful diagnostic tools.
3438 @cindex Verbose operation
3440 Normally, the @option{--list} (@option{-t}) command to list an archive
3441 prints just the file names (one per line) and the other commands are
3442 silent. When used with most operations, the @option{--verbose}
3443 (@option{-v}) option causes @command{tar} to print the name of each
3444 file or archive member as it is processed. This and the other options
3445 which make @command{tar} print status information can be useful in
3446 monitoring @command{tar}.
3448 With @option{--create} or @option{--extract}, @option{--verbose} used
3449 once just prints the names of the files or members as they are processed.
3450 Using it twice causes @command{tar} to print a longer listing
3451 (reminiscent of @samp{ls -l}) for each member. Since @option{--list}
3452 already prints the names of the members, @option{--verbose} used once
3453 with @option{--list} causes @command{tar} to print an @samp{ls -l}
3454 type listing of the files in the archive. The following examples both
3455 extract members with long list output:
3458 $ @kbd{tar --extract --file=archive.tar --verbose --verbose}
3459 $ @kbd{tar xvvf archive.tar}
3462 Verbose output appears on the standard output except when an archive is
3463 being written to the standard output, as with @samp{tar --create
3464 --file=- --verbose} (@samp{tar cfv -}, or even @samp{tar cv}---if the
3465 installer let standard output be the default archive). In that case
3466 @command{tar} writes verbose output to the standard error stream.
3468 If @option{--index-file=@var{file}} is specified, @command{tar} sends
3469 verbose output to @var{file} rather than to standard output or standard
3472 @cindex Obtaining total status information
3474 The @option{--totals} option---which is only meaningful when used with
3475 @option{--create} (@option{-c})---causes @command{tar} to print the total
3476 amount written to the archive, after it has been fully created.
3478 @cindex Progress information
3480 The @option{--checkpoint} option prints an occasional message
3481 as @command{tar} reads or writes the archive. In fact, it prints
3482 a message each 10 records read or written. It is designed for
3483 those who don't need the more detailed (and voluminous) output of
3484 @option{--block-number} (@option{-R}), but do want visual confirmation that @command{tar}
3485 is actually making forward progress.
3487 @FIXME{There is some confusion here. It seems that -R once wrote a
3488 message at @samp{every} record read or written.}
3490 @opindex show-omitted-dirs
3491 @anchor{show-omitted-dirs}
3492 The @option{--show-omitted-dirs} option, when reading an archive---with
3493 @option{--list} or @option{--extract}, for example---causes a message
3494 to be printed for each directory in the archive which is skipped.
3495 This happens regardless of the reason for skipping: the directory might
3496 not have been named on the command line (implicitly or explicitly),
3497 it might be excluded by the use of the @option{--exclude=@var{pattern}} option, or
3500 @opindex block-number
3501 @cindex Block number where error occurred
3502 @anchor{block-number}
3503 If @option{--block-number} (@option{-R}) is used, @command{tar} prints, along with
3504 every message it would normally produce, the block number within the
3505 archive where the message was triggered. Also, supplementary messages
3506 are triggered when reading blocks full of NULs, or when hitting end of
3507 file on the archive. As of now, if the archive if properly terminated
3508 with a NUL block, the reading of the file may stop before end of file
3509 is met, so the position of end of file will not usually show when
3510 @option{--block-number} (@option{-R}) is used. Note that @GNUTAR{}
3511 drains the archive before exiting when reading the
3512 archive from a pipe.
3514 @cindex Error message, block number of
3515 This option is especially useful when reading damaged archives, since
3516 it helps pinpoint the damaged sections. It can also be used with
3517 @option{--list} (@option{-t}) when listing a file-system backup tape, allowing you to
3518 choose among several backup tapes when retrieving a file later, in
3519 favor of the tape where the file appears earliest (closest to the
3520 front of the tape). @xref{backup}.
3523 @section Asking for Confirmation During Operations
3524 @cindex Interactive operation
3526 Typically, @command{tar} carries out a command without stopping for
3527 further instructions. In some situations however, you may want to
3528 exclude some files and archive members from the operation (for instance
3529 if disk or storage space is tight). You can do this by excluding
3530 certain files automatically (@pxref{Choosing}), or by performing
3531 an operation interactively, using the @option{--interactive} (@option{-w}) option.
3532 @command{tar} also accepts @option{--confirmation} for this option.
3534 @opindex interactive
3535 When the @option{--interactive} (@option{-w}) option is specified, before
3536 reading, writing, or deleting files, @command{tar} first prints a message
3537 for each such file, telling what operation it intends to take, then asks
3538 for confirmation on the terminal. The actions which require
3539 confirmation include adding a file to the archive, extracting a file
3540 from the archive, deleting a file from the archive, and deleting a file
3541 from disk. To confirm the action, you must type a line of input
3542 beginning with @samp{y}. If your input line begins with anything other
3543 than @samp{y}, @command{tar} skips that file.
3545 If @command{tar} is reading the archive from the standard input,
3546 @command{tar} opens the file @file{/dev/tty} to support the interactive
3549 Verbose output is normally sent to standard output, separate from
3550 other error messages. However, if the archive is produced directly
3551 on standard output, then verbose output is mixed with errors on
3552 @code{stderr}. Producing the archive on standard output may be used
3553 as a way to avoid using disk space, when the archive is soon to be
3554 consumed by another process reading it, say. Some people felt the need
3555 of producing an archive on stdout, still willing to segregate between
3556 verbose output and error output. A possible approach would be using a
3557 named pipe to receive the archive, and having the consumer process to
3558 read from that named pipe. This has the advantage of letting standard
3559 output free to receive verbose output, all separate from errors.
3562 @chapter @GNUTAR{} Operations
3575 @section Basic @GNUTAR{} Operations
3577 The basic @command{tar} operations, @option{--create} (@option{-c}),
3578 @option{--list} (@option{-t}) and @option{--extract} (@option{--get},
3579 @option{-x}), are currently presented and described in the tutorial
3580 chapter of this manual. This section provides some complementary notes
3581 for these operations.
3584 @opindex create, complementary notes
3588 Creating an empty archive would have some kind of elegance. One can
3589 initialize an empty archive and later use @option{--append}
3590 (@option{-r}) for adding all members. Some applications would not
3591 welcome making an exception in the way of adding the first archive
3592 member. On the other hand, many people reported that it is
3593 dangerously too easy for @command{tar} to destroy a magnetic tape with
3594 an empty archive@footnote{This is well described in @cite{Unix-haters
3595 Handbook}, by Simson Garfinkel, Daniel Weise & Steven Strassmann, IDG
3596 Books, ISBN 1-56884-203-1.}. The two most common errors are:
3600 Mistakingly using @code{create} instead of @code{extract}, when the
3601 intent was to extract the full contents of an archive. This error
3602 is likely: keys @kbd{c} and @kbd{x} are right next to each other on
3603 the QWERTY keyboard. Instead of being unpacked, the archive then
3604 gets wholly destroyed. When users speak about @dfn{exploding} an
3605 archive, they usually mean something else :-).
3608 Forgetting the argument to @code{file}, when the intent was to create
3609 an archive with a single file in it. This error is likely because a
3610 tired user can easily add the @kbd{f} key to the cluster of option
3611 letters, by the mere force of habit, without realizing the full
3612 consequence of doing so. The usual consequence is that the single
3613 file, which was meant to be saved, is rather destroyed.
3616 So, recognizing the likelihood and the catastrophical nature of these
3617 errors, @GNUTAR{} now takes some distance from elegance, and
3618 cowardly refuses to create an archive when @option{--create} option is
3619 given, there are no arguments besides options, and
3620 @option{--files-from} (@option{-T}) option is @emph{not} used. To get
3621 around the cautiousness of @GNUTAR{} and nevertheless create an
3622 archive with nothing in it, one may still use, as the value for the
3623 @option{--files-from} option, a file with no names in it, as shown in
3624 the following commands:
3627 @kbd{tar --create --file=empty-archive.tar --files-from=/dev/null}
3628 @kbd{tar cfT empty-archive.tar /dev/null}
3631 @opindex extract, complementary notes
3636 A socket is stored, within a @GNUTAR{} archive, as a pipe.
3638 @item @option{--list} (@option{-t})
3640 @GNUTAR{} now shows dates as @samp{1996-08-30},
3641 while it used to show them as @samp{Aug 30 1996}. Preferably,
3642 people should get used to ISO 8601 dates. Local American dates should
3643 be made available again with full date localization support, once
3644 ready. In the meantime, programs not being localizable for dates
3645 should prefer international dates, that's really the way to go.
3647 Look up @url{http://www.ft.uni-erlangen.de/~mskuhn/iso-time.html} if you
3648 are curious, it contains a detailed explanation of the ISO 8601 standard.
3653 @section Advanced @GNUTAR{} Operations
3655 Now that you have learned the basics of using @GNUTAR{}, you may want
3656 to learn about further ways in which @command{tar} can help you.
3658 This chapter presents five, more advanced operations which you probably
3659 won't use on a daily basis, but which serve more specialized functions.
3660 We also explain the different styles of options and why you might want
3661 to use one or another, or a combination of them in your @command{tar}
3662 commands. Additionally, this chapter includes options which allow you to
3663 define the output from @command{tar} more carefully, and provide help and
3664 error correction in special circumstances.
3666 @FIXME{check this after the chapter is actually revised to make sure
3667 it still introduces the info in the chapter correctly : ).}
3679 @subsection The Five Advanced @command{tar} Operations
3682 In the last chapter, you learned about the first three operations to
3683 @command{tar}. This chapter presents the remaining five operations to
3684 @command{tar}: @option{--append}, @option{--update}, @option{--concatenate},
3685 @option{--delete}, and @option{--compare}.
3687 You are not likely to use these operations as frequently as those
3688 covered in the last chapter; however, since they perform specialized
3689 functions, they are quite useful when you do need to use them. We
3690 will give examples using the same directory and files that you created
3691 in the last chapter. As you may recall, the directory is called
3692 @file{practice}, the files are @samp{jazz}, @samp{blues}, @samp{folk},
3693 @samp{rock}, and the two archive files you created are
3694 @samp{collection.tar} and @samp{music.tar}.
3696 We will also use the archive files @samp{afiles.tar} and
3697 @samp{bfiles.tar}. @samp{afiles.tar} contains the members @samp{apple},
3698 @samp{angst}, and @samp{aspic}. @samp{bfiles.tar} contains the members
3699 @samp{./birds}, @samp{baboon}, and @samp{./box}.
3701 Unless we state otherwise, all practicing you do and examples you follow
3702 in this chapter will take place in the @file{practice} directory that
3703 you created in the previous chapter; see @ref{prepare for examples}.
3704 (Below in this section, we will remind you of the state of the examples
3705 where the last chapter left them.)
3707 The five operations that we will cover in this chapter are:
3712 Add new entries to an archive that already exists.
3715 Add more recent copies of archive members to the end of an archive, if
3720 Add one or more pre-existing archives to the end of another archive.
3722 Delete items from an archive (does not work on tapes).
3726 Compare archive members to their counterparts in the file system.
3730 @subsection How to Add Files to Existing Archives: @option{--append}
3734 If you want to add files to an existing archive, you don't need to
3735 create a new archive; you can use @option{--append} (@option{-r}).
3736 The archive must already exist in order to use @option{--append}. (A
3737 related operation is the @option{--update} operation; you can use this
3738 to add newer versions of archive members to an existing archive. To learn how to
3739 do this with @option{--update}, @pxref{update}.)
3741 If you use @option{--append} to add a file that has the same name as an
3742 archive member to an archive containing that archive member, then the
3743 old member is not deleted. What does happen, however, is somewhat
3744 complex. @command{tar} @emph{allows} you to have infinite number of files
3745 with the same name. Some operations treat these same-named members no
3746 differently than any other set of archive members: for example, if you
3747 view an archive with @option{--list} (@option{-t}), you will see all
3748 of those members listed, with their data modification times, owners, etc.
3750 Other operations don't deal with these members as perfectly as you might
3751 prefer; if you were to use @option{--extract} to extract the archive,
3752 only the most recently added copy of a member with the same name as four
3753 other members would end up in the working directory. This is because
3754 @option{--extract} extracts an archive in the order the members appeared
3755 in the archive; the most recently archived members will be extracted
3756 last. Additionally, an extracted member will @emph{replace} a file of
3757 the same name which existed in the directory already, and @command{tar}
3758 will not prompt you about this@footnote{Unless you give it
3759 @option{--keep-old-files} option, or the disk copy is newer than the
3760 the one in the archive and you invoke @command{tar} with
3761 @option{--keep-newer-files} option}. Thus, only the most recently archived
3762 member will end up being extracted, as it will replace the one
3763 extracted before it, and so on.
3765 There exists a special option that allows you to get around this
3766 behavior and extract (or list) only a particular copy of the file.
3767 This is @option{--occurrence} option. If you run @command{tar} with
3768 this option, it will extract only the first copy of the file. You
3769 may also give this option an argument specifying the number of
3770 copy to be extracted. Thus, for example if the archive
3771 @file{archive.tar} contained three copies of file @file{myfile}, then
3775 tar --extract --file archive.tar --occurrence=2 myfile
3779 would extract only the second copy. @xref{Option
3780 Summary,---occurrence}, for the description of @option{--occurrence}
3783 @FIXME{ hag -- you might want to incorporate some of the above into the
3784 MMwtSN node; not sure. i didn't know how to make it simpler...
3786 There are a few ways to get around this. (maybe xref Multiple Members
3787 with the Same Name.}
3789 @cindex Members, replacing with other members
3790 @cindex Replacing members with other members
3791 If you want to replace an archive member, use @option{--delete} to
3792 delete the member you want to remove from the archive, , and then use
3793 @option{--append} to add the member you want to be in the archive. Note
3794 that you can not change the order of the archive; the most recently
3795 added member will still appear last. In this sense, you cannot truly
3796 ``replace'' one member with another. (Replacing one member with another
3797 will not work on certain types of media, such as tapes; see @ref{delete}
3798 and @ref{Media}, for more information.)
3801 * appending files:: Appending Files to an Archive
3805 @node appending files
3806 @subsubsection Appending Files to an Archive
3808 @cindex Adding files to an Archive
3809 @cindex Appending files to an Archive
3810 @cindex Archives, Appending files to
3812 The simplest way to add a file to an already existing archive is the
3813 @option{--append} (@option{-r}) operation, which writes specified files into the
3814 archive whether or not they are already among the archived files.
3815 When you use @option{--append}, you @emph{must} specify file name
3816 arguments, as there is no default. If you specify a file that already
3817 exists in the archive, another copy of the file will be added to the
3818 end of the archive. As with other operations, the member names of the
3819 newly added files will be exactly the same as their names given on the
3820 command line. The @option{--verbose} (@option{-v}) option will print
3821 out the names of the files as they are written into the archive.
3823 @option{--append} cannot be performed on some tape drives, unfortunately,
3824 due to deficiencies in the formats those tape drives use. The archive
3825 must be a valid @command{tar} archive, or else the results of using this
3826 operation will be unpredictable. @xref{Media}.
3828 To demonstrate using @option{--append} to add a file to an archive,
3829 create a file called @file{rock} in the @file{practice} directory.
3830 Make sure you are in the @file{practice} directory. Then, run the
3831 following @command{tar} command to add @file{rock} to
3832 @file{collection.tar}:
3835 $ @kbd{tar --append --file=collection.tar rock}
3839 If you now use the @option{--list} (@option{-t}) operation, you will see that
3840 @file{rock} has been added to the archive:
3843 $ @kbd{tar --list --file=collection.tar}
3844 -rw-r--r-- me user 28 1996-10-18 16:31 jazz
3845 -rw-r--r-- me user 21 1996-09-23 16:44 blues
3846 -rw-r--r-- me user 20 1996-09-23 16:44 folk
3847 -rw-r--r-- me user 20 1996-09-23 16:44 rock
3850 @FIXME{in theory, dan will (soon) try to turn this node into what it's
3851 title claims it will become...}
3854 @subsubsection Multiple Files with the Same Name
3856 You can use @option{--append} (@option{-r}) to add copies of files which have been
3857 updated since the archive was created. (However, we do not recommend
3858 doing this since there is another @command{tar} option called
3859 @option{--update}; @pxref{update} for more information. We describe this
3860 use of @option{--append} here for the sake of completeness.) @FIXME{is
3861 this really a good idea, to give this whole description for something
3862 which i believe is basically a Stupid way of doing something? certain
3863 aspects of it show ways in which tar is more broken than i'd personally
3864 like to admit to, specifically the last sentence. On the other hand, i
3865 don't think it's a good idea to be saying that we explicitly don't
3866 recommend using something, but i can't see any better way to deal with
3867 the situation.}When you extract the archive, the older version will be
3868 effectively lost. This works because files are extracted from an
3869 archive in the order in which they were archived. Thus, when the
3870 archive is extracted, a file archived later in time will replace a
3871 file of the same name which was archived earlier, even though the older
3872 version of the file will remain in the archive unless you delete all
3873 versions of the file.
3875 Supposing you change the file @file{blues} and then append the changed
3876 version to @file{collection.tar}. As you saw above, the original
3877 @file{blues} is in the archive @file{collection.tar}. If you change the
3878 file and append the new version of the file to the archive, there will
3879 be two copies in the archive. When you extract the archive, the older
3880 version of the file will be extracted first, and then replaced by the
3881 newer version when it is extracted.
3883 You can append the new, changed copy of the file @file{blues} to the
3884 archive in this way:
3887 $ @kbd{tar --append --verbose --file=collection.tar blues}
3892 Because you specified the @option{--verbose} option, @command{tar} has
3893 printed the name of the file being appended as it was acted on. Now
3894 list the contents of the archive:
3897 $ @kbd{tar --list --verbose --file=collection.tar}
3898 -rw-r--r-- me user 28 1996-10-18 16:31 jazz
3899 -rw-r--r-- me user 21 1996-09-23 16:44 blues
3900 -rw-r--r-- me user 20 1996-09-23 16:44 folk
3901 -rw-r--r-- me user 20 1996-09-23 16:44 rock
3902 -rw-r--r-- me user 58 1996-10-24 18:30 blues
3906 The newest version of @file{blues} is now at the end of the archive
3907 (note the different creation dates and file sizes). If you extract
3908 the archive, the older version of the file @file{blues} will be
3909 replaced by the newer version. You can confirm this by extracting
3910 the archive and running @samp{ls} on the directory.
3912 If you wish to extract the first occurrence of the file @file{blues}
3913 from the archive, use @option{--occurrence} option, as shown in
3914 the following example:
3917 $ @kbd{tar --extract -vv --occurrence --file=collection.tar blues}
3918 -rw-r--r-- me user 21 1996-09-23 16:44 blues
3921 @xref{Writing}, for more information on @option{--extract} and
3922 @xref{Option Summary, --occurrence}, for the description of
3923 @option{--occurrence} option.
3926 @subsection Updating an Archive
3928 @cindex Updating an archive
3931 In the previous section, you learned how to use @option{--append} to
3932 add a file to an existing archive. A related operation is
3933 @option{--update} (@option{-u}). The @option{--update} operation
3934 updates a @command{tar} archive by comparing the date of the specified
3935 archive members against the date of the file with the same name. If
3936 the file has been modified more recently than the archive member, then
3937 the newer version of the file is added to the archive (as with
3940 Unfortunately, you cannot use @option{--update} with magnetic tape drives.
3941 The operation will fail.
3943 @FIXME{other examples of media on which --update will fail? need to ask
3944 charles and/or mib/thomas/dave shevett..}
3946 Both @option{--update} and @option{--append} work by adding to the end
3947 of the archive. When you extract a file from the archive, only the
3948 version stored last will wind up in the file system, unless you use
3949 the @option{--backup} option. @FIXME-ref{Multiple Members with the
3957 @subsubsection How to Update an Archive Using @option{--update}
3959 You must use file name arguments with the @option{--update} (@option{-u}) operation.
3960 If you don't specify any files, @command{tar} won't act on any files and
3961 won't tell you that it didn't do anything (which may end up confusing
3964 @FIXME{note: the above parenthetical added because in fact, this
3965 behavior just confused the author. :-) }
3967 To see the @option{--update} option at work, create a new file,
3968 @file{classical}, in your practice directory, and some extra text to the
3969 file @file{blues}, using any text editor. Then invoke @command{tar} with
3970 the @samp{update} operation and the @option{--verbose} (@option{-v}) option specified,
3971 using the names of all the files in the practice directory as file name
3975 $ @kbd{tar --update -v -f collection.tar blues folk rock classical}
3982 Because we have specified verbose mode, @command{tar} prints out the names
3983 of the files it is working on, which in this case are the names of the
3984 files that needed to be updated. If you run @samp{tar --list} and look
3985 at the archive, you will see @file{blues} and @file{classical} at its
3986 end. There will be a total of two versions of the member @samp{blues};
3987 the one at the end will be newer and larger, since you added text before
3990 (The reason @command{tar} does not overwrite the older file when updating
3991 it is because writing to the middle of a section of tape is a difficult
3992 process. Tapes are not designed to go backward. @xref{Media}, for more
3993 information about tapes.
3995 @option{--update} (@option{-u}) is not suitable for performing backups for two
3996 reasons: it does not change directory content entries, and it
3997 lengthens the archive every time it is used. The @GNUTAR{}
3998 options intended specifically for backups are more
3999 efficient. If you need to run backups, please consult @ref{Backups}.
4002 @subsection Combining Archives with @option{--concatenate}
4004 @cindex Adding archives to an archive
4005 @cindex Concatenating Archives
4006 @opindex concatenate
4008 @c @cindex @option{-A} described
4009 Sometimes it may be convenient to add a second archive onto the end of
4010 an archive rather than adding individual files to the archive. To add
4011 one or more archives to the end of another archive, you should use the
4012 @option{--concatenate} (@option{--catenate}, @option{-A}) operation.
4014 To use @option{--concatenate}, give the first archive with
4015 @option{--file} option and name the rest of archives to be
4016 concatenated on the command line. The members, and their member
4017 names, will be copied verbatim from those archives to the first one.
4018 @FIXME-ref{This can cause multiple members to have the same name, for
4019 information on how this affects reading the archive, Multiple
4020 Members with the Same Name.}
4021 The new, concatenated archive will be called by the same name as the
4022 one given with the @option{--file} option. As usual, if you omit
4023 @option{--file}, @command{tar} will use the value of the environment
4024 variable @env{TAPE}, or, if this has not been set, the default archive name.
4026 @FIXME{There is no way to specify a new name...}
4028 To demonstrate how @option{--concatenate} works, create two small archives
4029 called @file{bluesrock.tar} and @file{folkjazz.tar}, using the relevant
4030 files from @file{practice}:
4033 $ @kbd{tar -cvf bluesrock.tar blues rock}
4036 $ @kbd{tar -cvf folkjazz.tar folk jazz}
4042 If you like, You can run @samp{tar --list} to make sure the archives
4043 contain what they are supposed to:
4046 $ @kbd{tar -tvf bluesrock.tar}
4047 -rw-r--r-- melissa user 105 1997-01-21 19:42 blues
4048 -rw-r--r-- melissa user 33 1997-01-20 15:34 rock
4049 $ @kbd{tar -tvf jazzfolk.tar}
4050 -rw-r--r-- melissa user 20 1996-09-23 16:44 folk
4051 -rw-r--r-- melissa user 65 1997-01-30 14:15 jazz
4054 We can concatenate these two archives with @command{tar}:
4058 $ @kbd{tar --concatenate --file=bluesrock.tar jazzfolk.tar}
4061 If you now list the contents of the @file{bluesrock.tar}, you will see
4062 that now it also contains the archive members of @file{jazzfolk.tar}:
4065 $ @kbd{tar --list --file=bluesrock.tar}
4072 When you use @option{--concatenate}, the source and target archives must
4073 already exist and must have been created using compatible format
4074 parameters. Notice, that @command{tar} does not check whether the
4075 archives it concatenates have compatible formats, it does not
4076 even check if the files are really tar archives.
4078 Like @option{--append} (@option{-r}), this operation cannot be performed on some
4079 tape drives, due to deficiencies in the formats those tape drives use.
4081 @cindex @code{concatenate} vs @command{cat}
4082 @cindex @command{cat} vs @code{concatenate}
4083 It may seem more intuitive to you to want or try to use @command{cat} to
4084 concatenate two archives instead of using the @option{--concatenate}
4085 operation; after all, @command{cat} is the utility for combining files.
4087 However, @command{tar} archives incorporate an end-of-file marker which
4088 must be removed if the concatenated archives are to be read properly as
4089 one archive. @option{--concatenate} removes the end-of-archive marker
4090 from the target archive before each new archive is appended. If you use
4091 @command{cat} to combine the archives, the result will not be a valid
4092 @command{tar} format archive. If you need to retrieve files from an
4093 archive that was added to using the @command{cat} utility, use the
4094 @option{--ignore-zeros} (@option{-i}) option. @xref{Ignore Zeros}, for further
4095 information on dealing with archives improperly combined using the
4096 @command{cat} shell utility.
4099 @subsection Removing Archive Members Using @option{--delete}
4101 @cindex Deleting files from an archive
4102 @cindex Removing files from an archive
4105 You can remove members from an archive by using the @option{--delete}
4106 option. Specify the name of the archive with @option{--file}
4107 (@option{-f}) and then specify the names of the members to be deleted;
4108 if you list no member names, nothing will be deleted. The
4109 @option{--verbose} option will cause @command{tar} to print the names
4110 of the members as they are deleted. As with @option{--extract}, you
4111 must give the exact member names when using @samp{tar --delete}.
4112 @option{--delete} will remove all versions of the named file from the
4113 archive. The @option{--delete} operation can run very slowly.
4115 Unlike other operations, @option{--delete} has no short form.
4117 @cindex Tapes, using @option{--delete} and
4118 @cindex Deleting from tape archives
4119 This operation will rewrite the archive. You can only use
4120 @option{--delete} on an archive if the archive device allows you to
4121 write to any point on the media, such as a disk; because of this, it
4122 does not work on magnetic tapes. Do not try to delete an archive member
4123 from a magnetic tape; the action will not succeed, and you will be
4124 likely to scramble the archive and damage your tape. There is no safe
4125 way (except by completely re-writing the archive) to delete files from
4126 most kinds of magnetic tape. @xref{Media}.
4128 To delete all versions of the file @file{blues} from the archive
4129 @file{collection.tar} in the @file{practice} directory, make sure you
4130 are in that directory, and then,
4133 $ @kbd{tar --list --file=collection.tar}
4143 $ @kbd{tar --delete --file=collection.tar blues}
4144 $ @kbd{tar --list --file=collection.tar}
4151 @FIXME{I changed the order of these nodes around and haven't had a chance
4152 to fix the above example's results, yet. I have to play with this and
4153 follow it and see what it actually does!}
4155 The @option{--delete} option has been reported to work properly when
4156 @command{tar} acts as a filter from @code{stdin} to @code{stdout}.
4159 @subsection Comparing Archive Members with the File System
4160 @cindex Verifying the currency of an archive
4164 The @option{--compare} (@option{-d}), or @option{--diff} operation compares
4165 specified archive members against files with the same names, and then
4166 reports differences in file size, mode, owner, modification date and
4167 contents. You should @emph{only} specify archive member names, not file
4168 names. If you do not name any members, then @command{tar} will compare the
4169 entire archive. If a file is represented in the archive but does not
4170 exist in the file system, @command{tar} reports a difference.
4172 You have to specify the record size of the archive when modifying an
4173 archive with a non-default record size.
4175 @command{tar} ignores files in the file system that do not have
4176 corresponding members in the archive.
4178 The following example compares the archive members @file{rock},
4179 @file{blues} and @file{funk} in the archive @file{bluesrock.tar} with
4180 files of the same name in the file system. (Note that there is no file,
4181 @file{funk}; @command{tar} will report an error message.)
4184 $ @kbd{tar --compare --file=bluesrock.tar rock blues funk}
4187 tar: funk not found in archive
4190 The spirit behind the @option{--compare} (@option{--diff}, @option{-d}) option is to check whether the
4191 archive represents the current state of files on disk, more than validating
4192 the integrity of the archive media. For this later goal, @xref{verify}.
4194 @node create options
4195 @section Options Used by @option{--create}
4197 @opindex create, additional options
4198 The previous chapter described the basics of how to use
4199 @option{--create} (@option{-c}) to create an archive from a set of files.
4200 @xref{create}. This section described advanced options to be used with
4204 * Ignore Failed Read::
4207 @node Ignore Failed Read
4208 @subsection Ignore Fail Read
4211 @item --ignore-failed-read
4212 Do not exit with nonzero on unreadable files or directories.
4215 @node extract options
4216 @section Options Used by @option{--extract}
4219 @FIXME{i need to get dan to go over these options with me and see if
4220 there's a better way of organizing them.}
4222 @opindex extract, additional options
4223 The previous chapter showed how to use @option{--extract} to extract
4224 an archive into the file system. Various options cause @command{tar} to
4225 extract more information than just file contents, such as the owner,
4226 the permissions, the modification date, and so forth. This section
4227 presents options to be used with @option{--extract} when certain special
4228 considerations arise. You may review the information presented in
4229 @ref{extract} for more basic information about the
4230 @option{--extract} operation.
4233 * Reading:: Options to Help Read Archives
4234 * Writing:: Changing How @command{tar} Writes Files
4235 * Scarce:: Coping with Scarce Resources
4239 @subsection Options to Help Read Archives
4240 @cindex Options when reading archives
4243 @cindex Reading incomplete records
4244 @cindex Records, incomplete
4245 @opindex read-full-records
4246 Normally, @command{tar} will request data in full record increments from
4247 an archive storage device. If the device cannot return a full record,
4248 @command{tar} will report an error. However, some devices do not always
4249 return full records, or do not require the last record of an archive to
4250 be padded out to the next record boundary. To keep reading until you
4251 obtain a full record, or to accept an incomplete record if it contains
4252 an end-of-archive marker, specify the @option{--read-full-records} (@option{-B}) option
4253 in conjunction with the @option{--extract} or @option{--list} operations.
4256 The @option{--read-full-records} (@option{-B}) option is turned on by default when
4257 @command{tar} reads an archive from standard input, or from a remote
4258 machine. This is because on BSD Unix systems, attempting to read a
4259 pipe returns however much happens to be in the pipe, even if it is
4260 less than was requested. If this option were not enabled, @command{tar}
4261 would fail as soon as it read an incomplete record from the pipe.
4263 If you're not sure of the blocking factor of an archive, you can
4264 read the archive by specifying @option{--read-full-records} (@option{-B}) and
4265 @option{--blocking-factor=@var{512-size}} (@option{-b
4266 @var{512-size}}), using a blocking factor larger than what the archive
4267 uses. This lets you avoid having to determine the blocking factor
4268 of an archive. @xref{Blocking Factor}.
4271 * read full records::
4275 @node read full records
4276 @unnumberedsubsubsec Reading Full Records
4278 @FIXME{need sentence or so of intro here}
4281 @opindex read-full-records
4282 @item --read-full-records
4284 Use in conjunction with @option{--extract} (@option{--get},
4285 @option{-x}) to read an archive which contains incomplete records, or
4286 one which has a blocking factor less than the one specified.
4290 @unnumberedsubsubsec Ignoring Blocks of Zeros
4292 @cindex End-of-archive blocks, ignoring
4293 @cindex Ignoring end-of-archive blocks
4294 @opindex ignore-zeros
4295 Normally, @command{tar} stops reading when it encounters a block of zeros
4296 between file entries (which usually indicates the end of the archive).
4297 @option{--ignore-zeros} (@option{-i}) allows @command{tar} to
4298 completely read an archive which contains a block of zeros before the
4299 end (i.e., a damaged archive, or one that was created by concatenating
4300 several archives together).
4302 The @option{--ignore-zeros} (@option{-i}) option is turned off by default because many
4303 versions of @command{tar} write garbage after the end-of-archive entry,
4304 since that part of the media is never supposed to be read. @GNUTAR{}
4305 does not write after the end of an archive, but seeks to
4306 maintain compatiblity among archiving utilities.
4309 @item --ignore-zeros
4311 To ignore blocks of zeros (i.e., end-of-archive entries) which may be
4312 encountered while reading an archive. Use in conjunction with
4313 @option{--extract} or @option{--list}.
4317 @subsection Changing How @command{tar} Writes Files
4320 @FIXME{Introductory paragraph}
4323 * Dealing with Old Files::
4324 * Overwrite Old Files::
4326 * Keep Newer Files::
4328 * Recursive Unlink::
4329 * Data Modification Times::
4330 * Setting Access Permissions::
4331 * Directory Modification Times and Permissions::
4332 * Writing to Standard Output::
4333 * Writing to an External Program::
4337 @node Dealing with Old Files
4338 @unnumberedsubsubsec Options Controlling the Overwriting of Existing Files
4340 @opindex overwrite-dir, introduced
4341 When extracting files, if @command{tar} discovers that the extracted
4342 file already exists, it normally replaces the file by removing it before
4343 extracting it, to prevent confusion in the presence of hard or symbolic
4344 links. (If the existing file is a symbolic link, it is removed, not
4345 followed.) However, if a directory cannot be removed because it is
4346 nonempty, @command{tar} normally overwrites its metadata (ownership,
4347 permission, etc.). The @option{--overwrite-dir} option enables this
4348 default behavior. To be more cautious and preserve the metadata of
4349 such a directory, use the @option{--no-overwrite-dir} option.
4351 @cindex Overwriting old files, prevention
4352 @opindex keep-old-files, introduced
4353 To be even more cautious and prevent existing files from being replaced, use
4354 the @option{--keep-old-files} (@option{-k}) option. It causes @command{tar} to refuse
4355 to replace or update a file that already exists, i.e., a file with the
4356 same name as an archive member prevents extraction of that archive
4357 member. Instead, it reports an error.
4359 @opindex overwrite, introduced
4360 To be more aggressive about altering existing files, use the
4361 @option{--overwrite} option. It causes @command{tar} to overwrite
4362 existing files and to follow existing symbolic links when extracting.
4364 @cindex Protecting old files
4365 Some people argue that @GNUTAR{} should not hesitate
4366 to overwrite files with other files when extracting. When extracting
4367 a @command{tar} archive, they expect to see a faithful copy of the
4368 state of the file system when the archive was created. It is debatable
4369 that this would always be a proper behavior. For example, suppose one
4370 has an archive in which @file{usr/local} is a link to
4371 @file{usr/local2}. Since then, maybe the site removed the link and
4372 renamed the whole hierarchy from @file{/usr/local2} to
4373 @file{/usr/local}. Such things happen all the time. I guess it would
4374 not be welcome at all that @GNUTAR{} removes the
4375 whole hierarchy just to make room for the link to be reinstated
4376 (unless it @emph{also} simultaneously restores the full
4377 @file{/usr/local2}, of course!) @GNUTAR{} is indeed
4378 able to remove a whole hierarchy to reestablish a symbolic link, for
4379 example, but @emph{only if} @option{--recursive-unlink} is specified
4380 to allow this behavior. In any case, single files are silently
4383 @opindex unlink-first, introduced
4384 Finally, the @option{--unlink-first} (@option{-U}) option can improve performance in
4385 some cases by causing @command{tar} to remove files unconditionally
4386 before extracting them.
4388 @node Overwrite Old Files
4389 @unnumberedsubsubsec Overwrite Old Files
4394 Overwrite existing files and directory metadata when extracting files
4397 This causes @command{tar} to write extracted files into the file system without
4398 regard to the files already on the system; i.e., files with the same
4399 names as archive members are overwritten when the archive is extracted.
4400 It also causes @command{tar} to extract the ownership, permissions,
4401 and time stamps onto any preexisting files or directories.
4402 If the name of a corresponding file name is a symbolic link, the file
4403 pointed to by the symbolic link will be overwritten instead of the
4404 symbolic link itself (if this is possible). Moreover, special devices,
4405 empty directories and even symbolic links are automatically removed if
4406 they are in the way of extraction.
4408 Be careful when using the @option{--overwrite} option, particularly when
4409 combined with the @option{--absolute-names} (@option{-P}) option, as this combination
4410 can change the contents, ownership or permissions of any file on your
4411 system. Also, many systems do not take kindly to overwriting files that
4412 are currently being executed.
4414 @opindex overwrite-dir
4415 @item --overwrite-dir
4416 Overwrite the metadata of directories when extracting files from an
4417 archive, but remove other files before extracting.
4420 @node Keep Old Files
4421 @unnumberedsubsubsec Keep Old Files
4424 @opindex keep-old-files
4425 @item --keep-old-files
4427 Do not replace existing files from archive. The
4428 @option{--keep-old-files} (@option{-k}) option prevents @command{tar}
4429 from replacing existing files with files with the same name from the
4430 archive. The @option{--keep-old-files} option is meaningless with
4431 @option{--list} (@option{-t}). Prevents @command{tar} from replacing
4432 files in the file system during extraction.
4435 @node Keep Newer Files
4436 @unnumberedsubsubsec Keep Newer Files
4439 @opindex keep-newer-files
4440 @item --keep-newer-files
4441 Do not replace existing files that are newer than their archive
4442 copies. This option is meaningless with @option{--list} (@option{-t}).
4446 @unnumberedsubsubsec Unlink First
4449 @opindex unlink-first
4450 @item --unlink-first
4452 Remove files before extracting over them.
4453 This can make @command{tar} run a bit faster if you know in advance
4454 that the extracted files all need to be removed. Normally this option
4455 slows @command{tar} down slightly, so it is disabled by default.
4458 @node Recursive Unlink
4459 @unnumberedsubsubsec Recursive Unlink
4462 @opindex recursive-unlink
4463 @item --recursive-unlink
4464 When this option is specified, try removing files and directory hierarchies
4465 before extracting over them. @emph{This is a dangerous option!}
4468 If you specify the @option{--recursive-unlink} option,
4469 @command{tar} removes @emph{anything} that keeps you from extracting a file
4470 as far as current permissions will allow it. This could include removal
4471 of the contents of a full directory hierarchy.
4473 @node Data Modification Times
4474 @unnumberedsubsubsec Setting Data Modification Times
4476 @cindex Data modification times of extracted files
4477 @cindex Modification times of extracted files
4478 Normally, @command{tar} sets the data modification times of extracted
4479 files to the corresponding times recorded for the files in the archive, but
4480 limits the permissions of extracted files by the current @code{umask}
4483 To set the data modification times of extracted files to the time when
4484 the files were extracted, use the @option{--touch} (@option{-m}) option in
4485 conjunction with @option{--extract} (@option{--get}, @option{-x}).
4491 Sets the data modification time of extracted archive members to the time
4492 they were extracted, not the time recorded for them in the archive.
4493 Use in conjunction with @option{--extract} (@option{--get}, @option{-x}).
4496 @node Setting Access Permissions
4497 @unnumberedsubsubsec Setting Access Permissions
4499 @cindex Permissions of extracted files
4500 @cindex Modes of extracted files
4501 To set the modes (access permissions) of extracted files to those
4502 recorded for those files in the archive, use @option{--same-permissions}
4503 in conjunction with the @option{--extract} (@option{--get},
4504 @option{-x}) operation. @FIXME{Should be aliased to ignore-umask.}
4507 @opindex preserve-permission
4508 @opindex same-permission
4509 @item --preserve-permission
4510 @itemx --same-permission
4511 @c @itemx --ignore-umask
4513 Set modes of extracted archive members to those recorded in the
4514 archive, instead of current umask settings. Use in conjunction with
4515 @option{--extract} (@option{--get}, @option{-x}).
4518 @node Directory Modification Times and Permissions
4519 @unnumberedsubsubsec Directory Modification Times and Permissions
4521 After sucessfully extracting a file member, @GNUTAR{} normally
4522 restores its permissions and modification times, as described in the
4523 previous sections. This cannot be done for directories, because
4524 after extracting a directory @command{tar} will almost certainly
4525 extract files into that directory and this will cause the directory
4526 modification time to be updated. Moreover, restoring that directory
4527 permissions may not permit file creation within it. Thus, restoring
4528 directory permissions and modification times must be delayed at least
4529 until all files have been extracted into that directory. @GNUTAR{}
4530 restores directories using the following approach.
4532 The extracted directories are created with the mode specified in the
4533 archive, as modified by the umask of the user, which gives sufficient
4534 permissions to allow file creation. The meta-information about the
4535 directory is recorded in the temporary list of directories. When
4536 preparing to extract next archive member, @GNUTAR{} checks if the
4537 directory prefix of this file contains the remembered directory. If
4538 it does not, the program assumes that all files have been extracted
4539 into that directory, restores its modification time and permissions
4540 and removes its entry from the internal list. This approach allows
4541 to correctly restore directory meta-information in the majority of
4542 cases, while keeping memory requirements sufficiently small. It is
4543 based on the fact, that most @command{tar} archives use the predefined
4544 order of members: first the directory, then all the files and
4545 subdirectories in that directory.
4547 However, this is not always true. The most important exception are
4548 incremental archives (@pxref{Incremental Dumps}). The member order in
4549 an incremental archive is reversed: first all directory members are
4550 stored, followed by other (non-directory) members. So, when extracting
4551 from incremental archives, @GNUTAR{} alters the above procedure. It
4552 remebers all restored directories, and restores their meta-data
4553 only after the entire archive has been processed. Notice, that you do
4554 not need to specity any special options for that, as @GNUTAR{}
4555 automatically detects archives in incremental format.
4557 There may be cases, when such processing is required for normal archives
4558 too. Consider the following example:
4562 $ @kbd{tar --no-recursion -cvf archive \
4563 foo foo/file1 bar bar/file foo/file2}
4572 During the normal operation, after encountering @file{bar}
4573 @GNUTAR{} will assume that all files from the directory @file{foo}
4574 were already extracted and will therefore restore its timestamp and
4575 permission bits. However, after extracting @file{foo/file2} the
4576 directory timestamp will be offset again.
4578 To correctly restore directory meta-information in such cases, use
4579 @option{delay-directory-restore} command line option:
4582 @opindex delay-directory-restore
4583 @item --delay-directory-restore
4584 Delays restoring of the modification times and permissions of extracted
4585 directories until the end of extraction. This way, correct
4586 meta-information is restored even if the archive has unusual member
4589 @opindex no-delay-directory-restore
4590 @item --no-delay-directory-restore
4591 Cancel the effect of the previous @option{--delay-directory-restore}.
4592 Use this option if you have used @option{--delay-directory-restore} in
4593 @env{TAR_OPTIONS} variable (@pxref{TAR_OPTIONS}) and wish to
4594 temporarily disable it.
4597 @node Writing to Standard Output
4598 @unnumberedsubsubsec Writing to Standard Output
4600 @cindex Writing extracted files to standard output
4601 @cindex Standard output, writing extracted files to
4602 To write the extracted files to the standard output, instead of
4603 creating the files on the file system, use @option{--to-stdout} (@option{-O}) in
4604 conjunction with @option{--extract} (@option{--get}, @option{-x}). This option is useful if you are
4605 extracting files to send them through a pipe, and do not need to
4606 preserve them in the file system. If you extract multiple members,
4607 they appear on standard output concatenated, in the order they are
4608 found in the archive.
4614 Writes files to the standard output. Use only in conjunction with
4615 @option{--extract} (@option{--get}, @option{-x}). When this option is
4616 used, instead of creating the files specified, @command{tar} writes
4617 the contents of the files extracted to its standard output. This may
4618 be useful if you are only extracting the files in order to send them
4619 through a pipe. This option is meaningless with @option{--list}
4623 This can be useful, for example, if you have a tar archive containing
4624 a big file and don't want to store the file on disk before processing
4625 it. You can use a command like this:
4628 tar -xOzf foo.tgz bigfile | process
4631 or even like this if you want to process the concatenation of the files:
4634 tar -xOzf foo.tgz bigfile1 bigfile2 | process
4637 Hovewer, @option{--to-command} may be more convenient for use with
4638 multiple files. See the next section.
4640 @node Writing to an External Program
4641 @unnumberedsubsubsec Writing to an External Program
4643 You can instruct @command{tar} to send the contents of each extracted
4644 file to the standard input of an external program:
4648 @item --to-command=@var{command}
4649 Extract files and pipe their contents to the standard input of
4650 @var{command}. When this option is used, instead of creating the
4651 files specified, @command{tar} invokes @var{command} and pipes the
4652 contents of the files to its standard output. @var{Command} may
4653 contain command line arguments. The program is executed via
4654 @code{sh -c}. Notice, that @var{command} is executed once for each regular file
4655 extracted. Non-regular files (directories, etc.) are ignored when this
4659 The command can obtain the information about the file it processes
4660 from the following environment variables:
4663 @vrindex TAR_FILETYPE, to-command environment
4665 Type of the file. It is a single letter with the following meaning:
4667 @multitable @columnfractions 0.10 0.90
4668 @item f @tab Regular file
4669 @item d @tab Directory
4670 @item l @tab Symbolic link
4671 @item h @tab Hard link
4672 @item b @tab Block device
4673 @item c @tab Character device
4676 Currently only regular files are supported.
4678 @vrindex TAR_MODE, to-command environment
4680 File mode, an octal number.
4682 @vrindex TAR_FILENAME, to-command environment
4684 The name of the file.
4686 @vrindex TAR_REALNAME, to-command environment
4688 Name of the file as stored in the archive.
4690 @vrindex TAR_UNAME, to-command environment
4692 Name of the file owner.
4694 @vrindex TAR_GNAME, to-command environment
4696 Name of the file owner group.
4698 @vrindex TAR_ATIME, to-command environment
4700 Time of last access. It is a decimal number, representing seconds
4701 since the epoch. If the archive provides times with nanosecond
4702 precision, the nanoseconds are appended to the timestamp after a
4705 @vrindex TAR_MTIME, to-command environment
4707 Time of last modification.
4709 @vrindex TAR_CTIME, to-command environment
4711 Time of last status change.
4713 @vrindex TAR_SIZE, to-command environment
4717 @vrindex TAR_UID, to-command environment
4719 UID of the file owner.
4721 @vrindex TAR_GID, to-command environment
4723 GID of the file owner.
4726 In addition to these variables, @env{TAR_VERSION} contains the
4727 @GNUTAR{} version number.
4729 If @var{command} exits with a non-0 status, @command{tar} will print
4730 an error message similar to the following:
4733 tar: 2345: Child returned status 1
4736 Here, @samp{2345} is the PID of the finished process.
4738 If this behavior is not wanted, use @option{--ignore-command-error}:
4741 @opindex ignore-command-error
4742 @item --ignore-command-error
4743 Ignore exit codes of subprocesses. Notice that if the program
4744 exits on signal or otherwise terminates abnormally, the error message
4745 will be printed even if this option is used.
4747 @opindex no-ignore-command-error
4748 @item --no-ignore-command-error
4749 Cancel the effect of any previous @option{--ignore-command-error}
4750 option. This option is useful if you have set
4751 @option{--ignore-command-error} in @env{TAR_OPTIONS}
4752 (@pxref{TAR_OPTIONS}) and wish to temporarily cancel it.
4756 @unnumberedsubsubsec Removing Files
4758 @FIXME{the various macros in the front of the manual think that this
4759 option goes in this section. i have no idea; i only know it's nowhere
4760 else in the book...}
4763 @opindex remove-files
4764 @item --remove-files
4765 Remove files after adding them to the archive.
4769 @subsection Coping with Scarce Resources
4772 @cindex Small memory
4773 @cindex Running out of space
4781 @unnumberedsubsubsec Starting File
4784 @opindex starting-file
4785 @item --starting-file=@var{name}
4786 @itemx -K @var{name}
4787 Starts an operation in the middle of an archive. Use in conjunction
4788 with @option{--extract} (@option{--get}, @option{-x}) or @option{--list} (@option{-t}).
4791 @cindex Middle of the archive, starting in the
4792 If a previous attempt to extract files failed due to lack of disk
4793 space, you can use @option{--starting-file=@var{name}} (@option{-K
4794 @var{name}}) to start extracting only after member @var{name} of the
4795 archive. This assumes, of course, that there is now free space, or
4796 that you are now extracting into a different file system. (You could
4797 also choose to suspend @command{tar}, remove unnecessary files from
4798 the file system, and then restart the same @command{tar} operation.
4799 In this case, @option{--starting-file} is not necessary.
4800 @xref{Incremental Dumps}, @xref{interactive}, and @ref{exclude}.)
4803 @unnumberedsubsubsec Same Order
4806 @cindex Large lists of file names on small machines
4808 @opindex preserve-order
4810 @itemx --preserve-order
4812 To process large lists of file names on machines with small amounts of
4813 memory. Use in conjunction with @option{--compare} (@option{--diff},
4814 @option{-d}), @option{--list} (@option{-t}) or @option{--extract}
4815 (@option{--get}, @option{-x}).
4818 The @option{--same-order} (@option{--preserve-order}, @option{-s}) option tells @command{tar} that the list of file
4819 names to be listed or extracted is sorted in the same order as the
4820 files in the archive. This allows a large list of names to be used,
4821 even on a small machine that would not otherwise be able to hold all
4822 the names in memory at the same time. Such a sorted list can easily be
4823 created by running @samp{tar -t} on the archive and editing its output.
4825 This option is probably never needed on modern computer systems.
4828 @section Backup options
4830 @cindex backup options
4832 @GNUTAR{} offers options for making backups of files
4833 before writing new versions. These options control the details of
4834 these backups. They may apply to the archive itself before it is
4835 created or rewritten, as well as individual extracted members. Other
4836 @acronym{GNU} programs (@command{cp}, @command{install}, @command{ln},
4837 and @command{mv}, for example) offer similar options.
4839 Backup options may prove unexpectedly useful when extracting archives
4840 containing many members having identical name, or when extracting archives
4841 on systems having file name limitations, making different members appear
4842 has having similar names through the side-effect of name truncation.
4843 (This is true only if we have a good scheme for truncated backup names,
4844 which I'm not sure at all: I suspect work is needed in this area.)
4845 When any existing file is backed up before being overwritten by extraction,
4846 then clashing files are automatically be renamed to be unique, and the
4847 true name is kept for only the last file of a series of clashing files.
4848 By using verbose mode, users may track exactly what happens.
4850 At the detail level, some decisions are still experimental, and may
4851 change in the future, we are waiting comments from our users. So, please
4852 do not learn to depend blindly on the details of the backup features.
4853 For example, currently, directories themselves are never renamed through
4854 using these options, so, extracting a file over a directory still has
4855 good chances to fail. Also, backup options apply to created archives,
4856 not only to extracted members. For created archives, backups will not
4857 be attempted when the archive is a block or character device, or when it
4858 refers to a remote file.
4860 For the sake of simplicity and efficiency, backups are made by renaming old
4861 files prior to creation or extraction, and not by copying. The original
4862 name is restored if the file creation fails. If a failure occurs after a
4863 partial extraction of a file, both the backup and the partially extracted
4867 @item --backup[=@var{method}]
4869 @vindex VERSION_CONTROL
4871 Back up files that are about to be overwritten or removed.
4872 Without this option, the original versions are destroyed.
4874 Use @var{method} to determine the type of backups made.
4875 If @var{method} is not specified, use the value of the @env{VERSION_CONTROL}
4876 environment variable. And if @env{VERSION_CONTROL} is not set,
4877 use the @samp{existing} method.
4879 @vindex version-control @r{Emacs variable}
4880 This option corresponds to the Emacs variable @samp{version-control};
4881 the same values for @var{method} are accepted as in Emacs. This option
4882 also allows more descriptive names. The valid @var{method}s are:
4887 @cindex numbered @r{backup method}
4888 Always make numbered backups.
4892 @cindex existing @r{backup method}
4893 Make numbered backups of files that already have them, simple backups
4898 @cindex simple @r{backup method}
4899 Always make simple backups.
4903 @item --suffix=@var{suffix}
4905 @cindex backup suffix
4906 @vindex SIMPLE_BACKUP_SUFFIX
4907 Append @var{suffix} to each backup file made with @option{--backup}. If this
4908 option is not specified, the value of the @env{SIMPLE_BACKUP_SUFFIX}
4909 environment variable is used. And if @env{SIMPLE_BACKUP_SUFFIX} is not
4910 set, the default is @samp{~}, just as in Emacs.
4914 Some people express the desire to @emph{always} use the @option{--backup}
4915 option, by defining some kind of alias or script. This is not as easy
4916 as one may think, due to the fact that old style options should appear first
4917 and consume arguments a bit unpredictably for an alias or script. But,
4918 if you are ready to give up using old style options, you may resort to
4919 using something like (a Bourne shell function here):
4922 tar () @{ /usr/local/bin/tar --backup $*; @}
4926 @section Notable @command{tar} Usages
4929 @FIXME{Using Unix file linking capability to recreate directory
4930 structures---linking files into one subdirectory and then
4931 @command{tar}ring that directory.}
4933 @FIXME{Nice hairy example using absolute-names, newer, etc.}
4936 You can easily use archive files to transport a group of files from
4937 one system to another: put all relevant files into an archive on one
4938 computer system, transfer the archive to another system, and extract
4939 the contents there. The basic transfer medium might be magnetic tape,
4940 Internet FTP, or even electronic mail (though you must encode the
4941 archive with @command{uuencode} in order to transport it properly by
4942 mail). Both machines do not have to use the same operating system, as
4943 long as they both support the @command{tar} program.
4945 For example, here is how you might copy a directory's contents from
4946 one disk to another, while preserving the dates, modes, owners and
4947 link-structure of all the files therein. In this case, the transfer
4948 medium is a @dfn{pipe}, which is one a Unix redirection mechanism:
4951 $ @kbd{cd sourcedir; tar -cf - . | (cd targetdir; tar -xf -)}
4955 The command also works using short option forms:
4958 $ @w{@kbd{cd sourcedir; tar --create --file=- . | (cd targetdir; tar --extract --file=-)}}
4962 This is one of the easiest methods to transfer a @command{tar} archive.
4965 @section Looking Ahead: The Rest of this Manual
4967 You have now seen how to use all eight of the operations available to
4968 @command{tar}, and a number of the possible options. The next chapter
4969 explains how to choose and change file and archive names, how to use
4970 files to store names of other files which you can then call as
4971 arguments to @command{tar} (this can help you save time if you expect to
4972 archive the same list of files a number of times), and so forth.
4973 @FIXME{in case it's not obvious, i'm making this up in some sense
4974 based on my limited memory of what the next chapter *really* does. i
4975 just wanted to flesh out this final section a little bit so i'd
4976 remember to stick it in here. :-)}
4978 If there are too many files to conveniently list on the command line,
4979 you can list the names in a file, and @command{tar} will read that file.
4982 There are various ways of causing @command{tar} to skip over some files,
4983 and not archive them. @xref{Choosing}.
4986 @chapter Performing Backups and Restoring Files
4989 @GNUTAR{} is distributed along with the scripts
4990 which the Free Software Foundation uses for performing backups. There
4991 is no corresponding scripts available yet for doing restoration of
4992 files. Even if there is a good chance those scripts may be satisfying
4993 to you, they are not the only scripts or methods available for doing
4994 backups and restore. You may well create your own, or use more
4995 sophisticated packages dedicated to that purpose.
4997 Some users are enthusiastic about @code{Amanda} (The Advanced Maryland
4998 Automatic Network Disk Archiver), a backup system developed by James
4999 da Silva @file{jds@@cs.umd.edu} and available on many Unix systems.
5000 This is free software, and it is available at these places:
5003 http://www.cs.umd.edu/projects/amanda/amanda.html
5004 ftp://ftp.cs.umd.edu/pub/amanda
5009 Here is a possible plan for a future documentation about the backuping
5010 scripts which are provided within the @GNUTAR{}
5016 @item what are dumps
5017 @item different levels of dumps
5019 @item full dump = dump everything
5020 @item level 1, level 2 dumps etc
5021 A level @var{n} dump dumps everything changed since the last level
5024 @item how to use scripts for dumps (ie, the concept)
5026 @item scripts to run after editing backup specs (details)
5028 @item Backup Specs, what is it.
5030 @item how to customize
5031 @item actual text of script [/sp/dump/backup-specs]
5035 @item rsh doesn't work
5036 @item rtape isn't installed
5039 @item the @option{--incremental} option of tar
5042 @item write protection
5043 @item types of media, different sizes and types, useful for different things
5044 @item files and tape marks
5045 one tape mark between files, two at end.
5046 @item positioning the tape
5047 MT writes two at end of write,
5048 backspaces over one when writing again.
5054 This chapter documents both the provided shell scripts and @command{tar}
5055 options which are more specific to usage as a backup tool.
5057 To @dfn{back up} a file system means to create archives that contain
5058 all the files in that file system. Those archives can then be used to
5059 restore any or all of those files (for instance if a disk crashes or a
5060 file is accidentally deleted). File system @dfn{backups} are also
5064 * Full Dumps:: Using @command{tar} to Perform Full Dumps
5065 * Incremental Dumps:: Using @command{tar} to Perform Incremental Dumps
5066 * Backup Levels:: Levels of Backups
5067 * Backup Parameters:: Setting Parameters for Backups and Restoration
5068 * Scripted Backups:: Using the Backup Scripts
5069 * Scripted Restoration:: Using the Restore Script
5073 @section Using @command{tar} to Perform Full Dumps
5079 @cindex corrupted archives
5080 Full dumps should only be made when no other people or programs
5081 are modifying files in the file system. If files are modified while
5082 @command{tar} is making the backup, they may not be stored properly in
5083 the archive, in which case you won't be able to restore them if you
5084 have to. (Files not being modified are written with no trouble, and do
5085 not corrupt the entire archive.)
5087 You will want to use the @option{--label=@var{archive-label}}
5088 (@option{-V @var{archive-label}}) option to give the archive a
5089 volume label, so you can tell what this archive is even if the label
5090 falls off the tape, or anything like that.
5092 Unless the file system you are dumping is guaranteed to fit on
5093 one volume, you will need to use the @option{--multi-volume} (@option{-M}) option.
5094 Make sure you have enough tapes on hand to complete the backup.
5096 If you want to dump each file system separately you will need to use
5097 the @option{--one-file-system} option to prevent
5098 @command{tar} from crossing file system boundaries when storing
5101 The @option{--incremental} (@option{-G}) (@pxref{Incremental Dumps})
5102 option is not needed, since this is a complete copy of everything in
5103 the file system, and a full restore from this backup would only be
5104 done onto a completely
5107 Unless you are in a hurry, and trust the @command{tar} program (and your
5108 tapes), it is a good idea to use the @option{--verify} (@option{-W})
5109 option, to make sure your files really made it onto the dump properly.
5110 This will also detect cases where the file was modified while (or just
5111 after) it was being archived. Not all media (notably cartridge tapes)
5112 are capable of being verified, unfortunately.
5114 @node Incremental Dumps
5115 @section Using @command{tar} to Perform Incremental Dumps
5117 @dfn{Incremental backup} is a special form of @GNUTAR{} archive that
5118 stores additional metadata so that exact state of the file system
5119 can be restored when extracting the archive.
5121 @GNUTAR{} currently offers two options for handling incremental
5122 backups: @option{--listed-incremental=@var{snapshot-file}} (@option{-g
5123 @var{snapshot-file}}) and @option{--incremental} (@option{-G}).
5125 @opindex listed-incremental
5126 The option @option{--listed-incremental} instructs tar to operate on
5127 an incremental archive with additional metadata stored in a standalone
5128 file, called a @dfn{snapshot file}. The purpose of this file is to help
5129 determine which files have been changed, added or deleted since the
5130 last backup, so that the next incremental backup will contain only
5131 modified files. The name of the snapshot file is given as an argument
5135 @item --listed-incremental=@var{file}
5136 @itemx -g @var{file}
5137 Handle incremental backups with snapshot data in @var{file}.
5140 To create an incremental backup, you would use
5141 @option{--listed-incremental} together with @option{--create}
5142 (@pxref{create}). For example:
5145 $ @kbd{tar --create \
5146 --file=archive.1.tar \
5147 --listed-incremental=/var/log/usr.snar \
5151 This will create in @file{archive.1.tar} an incremental backup of
5152 the @file{/usr} file system, storing additional metadata in the file
5153 @file{/var/log/usr.snar}. If this file does not exist, it will be
5154 created. The created archive will then be a @dfn{level 0 backup};
5155 please see the next section for more on backup levels.
5157 Otherwise, if the file @file{/var/log/usr.snar} exists, it
5158 determines which files are modified. In this case only these files will be
5159 stored in the archive. Suppose, for example, that after running the
5160 above command, you delete file @file{/usr/doc/old} and create
5161 directory @file{/usr/local/db} with the following contents:
5164 $ @kbd{ls /usr/local/db}
5169 Some time later you create another incremental backup. You will
5173 $ @kbd{tar --create \
5174 --file=archive.2.tar \
5175 --listed-incremental=/var/log/usr.snar \
5177 tar: usr/local/db: Directory is new
5184 The created archive @file{archive.2.tar} will contain only these
5185 three members. This archive is called a @dfn{level 1 backup}. Notice
5186 that @file{/var/log/usr.snar} will be updated with the new data, so if
5187 you plan to create more @samp{level 1} backups, it is necessary to
5188 create a working copy of the snapshot file before running
5189 @command{tar}. The above example will then be modified as follows:
5192 $ @kbd{cp /var/log/usr.snar /var/log/usr.snar-1}
5193 $ @kbd{tar --create \
5194 --file=archive.2.tar \
5195 --listed-incremental=/var/log/usr.snar-1 \
5199 Incremental dumps depend crucially on time stamps, so the results are
5200 unreliable if you modify a file's time stamps during dumping (e.g.,
5201 with the @option{--atime-preserve=replace} option), or if you set the clock
5204 Metadata stored in snapshot files include device numbers, which,
5205 obviously is supposed to be a non-volatile value. However, it turns
5206 out that NFS devices have undependable values when an automounter
5207 gets in the picture. This can lead to a great deal of spurious
5208 redumping in incremental dumps, so it is somewhat useless to compare
5209 two NFS devices numbers over time. The solution implemented currently
5210 is to considers all NFS devices as being equal when it comes to
5211 comparing directories; this is fairly gross, but there does not seem
5212 to be a better way to go.
5214 Note that incremental archives use @command{tar} extensions and may
5215 not be readable by non-@acronym{GNU} versions of the @command{tar} program.
5217 @opindex listed-incremental, using with @option{--extract}
5218 @opindex extract, using with @option{--listed-incremental}
5219 To extract from the incremental dumps, use
5220 @option{--listed-incremental} together with @option{--extract}
5221 option (@pxref{extracting files}). In this case, @command{tar} does
5222 not need to access snapshot file, since all the data necessary for
5223 extraction are stored in the archive itself. So, when extracting, you
5224 can give whatever argument to @option{--listed-incremental}, the usual
5225 practice is to use @option{--listed-incremental=/dev/null}.
5226 Alternatively, you can use @option{--incremental}, which needs no
5227 arguments. In general, @option{--incremental} (@option{-G}) can be
5228 used as a shortcut for @option{--listed-incremental} when listing or
5229 extracting incremental backups (for more information, regarding this
5230 option, @pxref{incremental-op}).
5232 When extracting from the incremental backup @GNUTAR{} attempts to
5233 restore the exact state the file system had when the archive was
5234 created. In particular, it will @emph{delete} those files in the file
5235 system that did not exist in their directories when the archive was
5236 created. If you have created several levels of incremental files,
5237 then in order to restore the exact contents the file system had when
5238 the last level was created, you will need to restore from all backups
5239 in turn. Continuing our example, to restore the state of @file{/usr}
5240 file system, one would do@footnote{Notice, that since both archives
5241 were created withouth @option{-P} option (@pxref{absolute}), these
5242 commands should be run from the root file system.}:
5245 $ @kbd{tar --extract \
5246 --listed-incremental=/dev/null \
5247 --file archive.1.tar}
5248 $ @kbd{tar --extract \
5249 --listed-incremental=/dev/null \
5250 --file archive.2.tar}
5253 To list the contents of an incremental archive, use @option{--list}
5254 (@pxref{list}), as usual. To obtain more information about the
5255 archive, use @option{--listed-incremental} or @option{--incremental}
5256 combined with two @option{--verbose} options@footnote{Two
5257 @option{--verbose} options were selected to avoid breaking usual
5258 verbose listing output (@option{--list --verbose}) when using in
5261 @opindex incremental, using with @option{--list}
5262 @opindex listed-incremental, using with @option{--list}
5263 @opindex list, using with @option{--incremental}
5264 @opindex list, using with @option{--listed-incremental}
5265 Versions of @GNUTAR{} up to 1.15.1 used to dump verbatim binary
5266 contents of the DUMPDIR header (with terminating nulls) when
5267 @option{--incremental} or @option{--listed-incremental} option was
5268 given, no matter what the verbosity level. This behavior, and,
5269 especially, the binary output it produced were considered incovenient
5270 and were changed in version 1.16}:
5273 @kbd{tar --list --incremental --verbose --verbose archive.tar}
5276 This command will print, for each directory in the archive, the list
5277 of files in that directory at the time the archive was created. This
5278 information is put out in a format which is both human-readable and
5279 unambiguous for a program: each file name is printed as
5286 where @var{x} is a letter describing the status of the file: @samp{Y}
5287 if the file is present in the archive, @samp{N} if the file is not
5288 included in the archive, or a @samp{D} if the file is a directory (and
5289 is included in the archive).@FIXME-xref{dumpdir format}. Each such
5290 line is terminated by a newline character. The last line is followed
5291 by an additional newline to indicate the end of the data.
5293 @anchor{incremental-op}The option @option{--incremental} (@option{-G})
5294 gives the same behavior as @option{--listed-incremental} when used
5295 with @option{--list} and @option{--extract} options. When used with
5296 @option{--create} option, it creates an incremental archive without
5297 creating snapshot file. Thus, it is impossible to create several
5298 levels of incremental backups with @option{--incremental} option.
5301 @section Levels of Backups
5303 An archive containing all the files in the file system is called a
5304 @dfn{full backup} or @dfn{full dump}. You could insure your data by
5305 creating a full dump every day. This strategy, however, would waste a
5306 substantial amount of archive media and user time, as unchanged files
5307 are daily re-archived.
5309 It is more efficient to do a full dump only occasionally. To back up
5310 files between full dumps, you can use @dfn{incremental dumps}. A @dfn{level
5311 one} dump archives all the files that have changed since the last full
5314 A typical dump strategy would be to perform a full dump once a week,
5315 and a level one dump once a day. This means some versions of files
5316 will in fact be archived more than once, but this dump strategy makes
5317 it possible to restore a file system to within one day of accuracy by
5318 only extracting two archives---the last weekly (full) dump and the
5319 last daily (level one) dump. The only information lost would be in
5320 files changed or created since the last daily backup. (Doing dumps
5321 more than once a day is usually not worth the trouble).
5323 @GNUTAR{} comes with scripts you can use to do full
5324 and level-one (actually, even level-two and so on) dumps. Using
5325 scripts (shell programs) to perform backups and restoration is a
5326 convenient and reliable alternative to typing out file name lists
5327 and @command{tar} commands by hand.
5329 Before you use these scripts, you need to edit the file
5330 @file{backup-specs}, which specifies parameters used by the backup
5331 scripts and by the restore script. This file is usually located
5332 in @file{/etc/backup} directory. @xref{Backup Parameters}, for its
5333 detailed description. Once the backup parameters are set, you can
5334 perform backups or restoration by running the appropriate script.
5336 The name of the backup script is @code{backup}. The name of the
5337 restore script is @code{restore}. The following sections describe
5338 their use in detail.
5340 @emph{Please Note:} The backup and restoration scripts are
5341 designed to be used together. While it is possible to restore files by
5342 hand from an archive which was created using a backup script, and to create
5343 an archive by hand which could then be extracted using the restore script,
5344 it is easier to use the scripts. @xref{Incremental Dumps}, before
5345 making such an attempt.
5347 @node Backup Parameters
5348 @section Setting Parameters for Backups and Restoration
5350 The file @file{backup-specs} specifies backup parameters for the
5351 backup and restoration scripts provided with @command{tar}. You must
5352 edit @file{backup-specs} to fit your system configuration and schedule
5353 before using these scripts.
5355 Syntactically, @file{backup-specs} is a shell script, containing
5356 mainly variable assignments. However, any valid shell construct
5357 is allowed in this file. Particularly, you may wish to define
5358 functions within that script (e.g., see @code{RESTORE_BEGIN} below).
5359 For more information about shell script syntax, please refer to
5360 @url{http://www.opengroup.org/onlinepubs/009695399/utilities/xcu_chap02.html#ta
5361 g_02, the definition of the Shell Command Language}. See also
5362 @ref{Top,,Bash Features,bashref,Bash Reference Manual}.
5364 The shell variables controlling behavior of @code{backup} and
5365 @code{restore} are described in the following subsections.
5368 * General-Purpose Variables::
5369 * Magnetic Tape Control::
5371 * backup-specs example:: An Example Text of @file{Backup-specs}
5374 @node General-Purpose Variables
5375 @subsection General-Purpose Variables
5377 @defvr {Backup variable} ADMINISTRATOR
5378 The user name of the backup administrator. @code{Backup} scripts
5379 sends a backup report to this address.
5382 @defvr {Backup variable} BACKUP_HOUR
5383 The hour at which the backups are done. This can be a number from 0
5384 to 23, or the time specification in form @var{hours}:@var{minutes},
5385 or the string @samp{now}.
5387 This variable is used by @code{backup}. Its value may be overridden
5388 using @option{--time} option (@pxref{Scripted Backups}).
5391 @defvr {Backup variable} TAPE_FILE
5393 The device @command{tar} writes the archive to. If @var{TAPE_FILE}
5394 is a remote archive (@pxref{remote-dev}), backup script will suppose
5395 that your @command{mt} is able to access remote devices. If @var{RSH}
5396 (@pxref{RSH}) is set, @option{--rsh-command} option will be added to
5397 invocations of @command{mt}.
5400 @defvr {Backup variable} BLOCKING
5402 The blocking factor @command{tar} will use when writing the dump archive.
5403 @xref{Blocking Factor}.
5406 @defvr {Backup variable} BACKUP_DIRS
5408 A list of file systems to be dumped (for @code{backup}), or restored
5409 (for @code{restore}). You can include any directory
5410 name in the list --- subdirectories on that file system will be
5411 included, regardless of how they may look to other networked machines.
5412 Subdirectories on other file systems will be ignored.
5414 The host name specifies which host to run @command{tar} on, and should
5415 normally be the host that actually contains the file system. However,
5416 the host machine must have @GNUTAR{} installed, and
5417 must be able to access the directory containing the backup scripts and
5418 their support files using the same file name that is used on the
5419 machine where the scripts are run (ie. what @command{pwd} will print
5420 when in that directory on that machine). If the host that contains
5421 the file system does not have this capability, you can specify another
5422 host as long as it can access the file system through NFS.
5424 If the list of file systems is very long you may wish to put it
5425 in a separate file. This file is usually named
5426 @file{/etc/backup/dirs}, but this name may be overridden in
5427 @file{backup-specs} using @code{DIRLIST} variable.
5430 @defvr {Backup variable} DIRLIST
5432 A path to the file containing the list of the file systems to backup
5433 or restore. By default it is @file{/etc/backup/dirs}.
5436 @defvr {Backup variable} BACKUP_FILES
5438 A list of individual files to be dumped (for @code{backup}), or restored
5439 (for @code{restore}). These should be accessible from the machine on
5440 which the backup script is run.
5442 If the list of file systems is very long you may wish to store it
5443 in a separate file. This file is usually named
5444 @file{/etc/backup/files}, but this name may be overridden in
5445 @file{backup-specs} using @code{FILELIST} variable.
5448 @defvr {Backup variable} FILELIST
5450 A path to the file containing the list of the individual files to backup
5451 or restore. By default it is @file{/etc/backup/files}.
5454 @defvr {Backup variable} MT
5456 Full file name of @command{mt} binary.
5459 @defvr {Backup variable} RSH
5461 Full file name of @command{rsh} binary or its equivalent. You may wish to
5462 set it to @code{ssh}, to improve security. In this case you will have
5463 to use public key authentication.
5466 @defvr {Backup variable} RSH_COMMAND
5468 Full file name of @command{rsh} binary on remote mashines. This will
5469 be passed via @option{--rsh-command} option to the remote invocation
5473 @defvr {Backup variable} VOLNO_FILE
5475 Name of temporary file to hold volume numbers. This needs to be accessible
5476 by all the machines which have file systems to be dumped.
5479 @defvr {Backup variable} XLIST
5481 Name of @dfn{exclude file list}. An @dfn{exclude file list} is a file
5482 located on the remote machine and containing the list of files to
5483 be excluded from the backup. Exclude file lists are searched in
5484 /etc/tar-backup directory. A common use for exclude file lists
5485 is to exclude files containing security-sensitive information
5486 (e.g., @file{/etc/shadow} from backups).
5488 This variable affects only @code{backup}.
5491 @defvr {Backup variable} SLEEP_TIME
5493 Time to sleep between dumps of any two successive file systems
5495 This variable affects only @code{backup}.
5498 @defvr {Backup variable} DUMP_REMIND_SCRIPT
5500 Script to be run when it's time to insert a new tape in for the next
5501 volume. Administrators may want to tailor this script for their site.
5502 If this variable isn't set, @GNUTAR{} will display its built-in prompt
5503 @FIXME-xref{describe it somewhere!}, and will expect confirmation from
5507 @defvr {Backup variable} SLEEP_MESSAGE
5509 Message to display on the terminal while waiting for dump time. Usually
5510 this will just be some literal text.
5513 @defvr {Backup variable} TAR
5515 Full file name of the @GNUTAR{} executable. If this is not set, backup
5516 scripts will search @command{tar} in the current shell path.
5519 @node Magnetic Tape Control
5520 @subsection Magnetic Tape Control
5522 Backup scripts access tape device using special @dfn{hook functions}.
5523 These functions take a single argument -- the name of the tape
5524 device. Their names are kept in the following variables:
5526 @defvr {Backup variable} MT_BEGIN
5527 The name of @dfn{begin} function. This function is called before
5528 accessing the drive. By default it retensions the tape:
5534 mt -f "$1" retension
5539 @defvr {Backup variable} MT_REWIND
5540 The name of @dfn{rewind} function. The default definition is as
5553 @defvr {Backup variable} MT_OFFLINE
5554 The name of the function switching the tape off line. By default
5555 it is defined as follows:
5558 MT_OFFLINE=mt_offline
5566 @defvr {Backup variable} MT_STATUS
5567 The name of the function used to obtain the status of the archive device,
5568 including error count. Default definition:
5580 @subsection User Hooks
5582 @dfn{User hooks} are shell functions executed before and after
5583 each @command{tar} invocation. Thus, there are @dfn{backup
5584 hooks}, which are executed before and after dumping each file
5585 system, and @dfn{restore hooks}, executed before and
5586 after restoring a file system. Each user hook is a shell function
5587 taking four arguments:
5589 @deffn {User Hook Function} hook @var{level} @var{host} @var{fs} @var{fsname}
5594 Current backup or restore level.
5597 Name or IP address of the host machine being dumped or restored.
5600 Full path name to the file system being dumped or restored.
5603 File system name with directory separators replaced with colons. This
5604 is useful, e.g., for creating unique files.
5608 Following variables keep the names of user hook functions
5610 @defvr {Backup variable} DUMP_BEGIN
5611 Dump begin function. It is executed before dumping the file system.
5614 @defvr {Backup variable} DUMP_END
5615 Executed after dumping the file system.
5618 @defvr {Backup variable} RESTORE_BEGIN
5619 Executed before restoring the file system.
5622 @defvr {Backup variable} RESTORE_END
5623 Executed after restoring the file system.
5626 @node backup-specs example
5627 @subsection An Example Text of @file{Backup-specs}
5629 The following is an example of @file{backup-specs}:
5632 # site-specific parameters for file system backup.
5634 ADMINISTRATOR=friedman
5636 TAPE_FILE=/dev/nrsmt0
5638 # Use @code{ssh} instead of the less secure @code{rsh}
5640 RSH_COMMAND=/usr/bin/ssh
5642 # Override MT_STATUS function:
5648 # Disable MT_OFFLINE function
5665 apple-gunkies:/com/mailer/gnu
5666 apple-gunkies:/com/archive/gnu"
5668 BACKUP_FILES="/com/mailer/aliases /com/mailer/league*[a-z]"
5672 @node Scripted Backups
5673 @section Using the Backup Scripts
5675 The syntax for running a backup script is:
5678 backup --level=@var{level} --time=@var{time}
5681 The @option{level} option requests the dump level. Thus, to produce
5682 a full dump, specify @code{--level=0} (this is the default, so
5683 @option{--level} may be omitted if its value is @code{0}).
5684 @footnote{For backward compatibility, the @code{backup} will also
5685 try to deduce the requested dump level from the name of the
5686 script itself. If the name consists of a string @samp{level-}
5687 followed by a single decimal digit, that digit is taken as
5688 the dump level number. Thus, you may create a link from @code{backup}
5689 to @code{level-1} and then run @code{level-1} whenever you need to
5690 create a level one dump.}
5692 The @option{--time} option determines when should the backup be
5693 run. @var{Time} may take three forms:
5696 @item @var{hh}:@var{mm}
5698 The dump must be run at @var{hh} hours @var{mm} minutes.
5702 The dump must be run at @var{hh} hours
5706 The dump must be run immediately.
5709 You should start a script with a tape or disk mounted. Once you
5710 start a script, it prompts you for new tapes or disks as it
5711 needs them. Media volumes don't have to correspond to archive
5712 files --- a multi-volume archive can be started in the middle of a
5713 tape that already contains the end of another multi-volume archive.
5714 The @code{restore} script prompts for media by its archive volume,
5715 so to avoid an error message you should keep track of which tape
5716 (or disk) contains which volume of the archive (@pxref{Scripted
5719 The backup scripts write two files on the file system. The first is a
5720 record file in @file{/etc/tar-backup/}, which is used by the scripts
5721 to store and retrieve information about which files were dumped. This
5722 file is not meant to be read by humans, and should not be deleted by
5723 them. @xref{Snapshot Files}, for a more detailed explanation of this
5726 The second file is a log file containing the names of the file systems
5727 and files dumped, what time the backup was made, and any error
5728 messages that were generated, as well as how much space was left in
5729 the media volume after the last volume of the archive was written.
5730 You should check this log file after every backup. The file name is
5731 @file{log-@var{mm-dd-yyyy}-level-@var{n}}, where @var{mm-dd-yyyy}
5732 represents current date, and @var{n} represents current dump level number.
5734 The script also prints the name of each system being dumped to the
5737 Following is the full list of options accepted by @code{backup}
5741 @item -l @var{level}
5742 @itemx --level=@var{level}
5743 Do backup level @var{level} (default 0).
5747 Force backup even if today's log file already exists.
5749 @item -v[@var{level}]
5750 @itemx --verbose[=@var{level}]
5751 Set verbosity level. The higher the level is, the more debugging
5752 information will be output during execution. Devault @var{level}
5753 is 100, which means the highest debugging level.
5755 @item -t @var{start-time}
5756 @itemx --time=@var{start-time}
5757 Wait till @var{time}, then do backup.
5761 Display short help message and exit.
5765 Display information about the program's name, version, origin and legal
5766 status, all on standard output, and then exit successfully.
5770 @node Scripted Restoration
5771 @section Using the Restore Script
5773 To restore files that were archived using a scripted backup, use the
5774 @code{restore} script. Its usage is quite straightforward. In the
5775 simplest form, invoke @code{restore --all}, it will
5776 then restore all the file systems and files specified in
5777 @file{backup-specs} (@pxref{General-Purpose Variables,BACKUP_DIRS}).
5779 You may select the file systems (and/or files) to restore by
5780 giving @code{restore} list of @dfn{patterns} in its command
5781 line. For example, running
5788 will restore all file systems on the machine @samp{albert}. A more
5789 complicated example:
5792 restore 'albert:*' '*:/var'
5796 This command will restore all file systems on the machine @samp{albert}
5797 as well as @file{/var} file system on all machines.
5799 By default @code{restore} will start restoring files from the lowest
5800 available dump level (usually zero) and will continue through
5801 all available dump levels. There may be situations where such a
5802 thorough restore is not necessary. For example, you may wish to
5803 restore only files from the recent level one backup. To do so,
5804 use @option{--level} option, as shown in the example below:
5810 The full list of options accepted by @code{restore} follows:
5815 Restore all file systems and files specified in @file{backup-specs}
5817 @item -l @var{level}
5818 @itemx --level=@var{level}
5819 Start restoring from the given backup level, instead of the default 0.
5821 @item -v[@var{level}]
5822 @itemx --verbose[=@var{level}]
5823 Set verbosity level. The higher the level is, the more debugging
5824 information will be output during execution. Devault @var{level}
5825 is 100, which means the highest debugging level.
5829 Display short help message and exit.
5833 Display information about the program's name, version, origin and legal
5834 status, all on standard output, and then exit successfully.
5837 You should start the restore script with the media containing the
5838 first volume of the archive mounted. The script will prompt for other
5839 volumes as they are needed. If the archive is on tape, you don't need
5840 to rewind the tape to to its beginning---if the tape head is
5841 positioned past the beginning of the archive, the script will rewind
5842 the tape as needed. @FIXME-xref{Media, for a discussion of tape
5846 @strong{Warning:} The script will delete files from the active file
5847 system if they were not in the file system when the archive was made.
5850 @xref{Incremental Dumps}, for an explanation of how the script makes
5854 @chapter Choosing Files and Names for @command{tar}
5857 @FIXME{Melissa (still) Doesn't Really Like This ``Intro'' Paragraph!!!}
5859 Certain options to @command{tar} enable you to specify a name for your
5860 archive. Other options let you decide which files to include or exclude
5861 from the archive, based on when or whether files were modified, whether
5862 the file names do or don't match specified patterns, or whether files
5863 are in specified directories.
5866 * file:: Choosing the Archive's Name
5867 * Selecting Archive Members::
5868 * files:: Reading Names from a File
5869 * exclude:: Excluding Some Files
5871 * after:: Operating Only on New Files
5872 * recurse:: Descending into Directories
5873 * one:: Crossing File System Boundaries
5877 @section Choosing and Naming Archive Files
5880 @FIXME{should the title of this section actually be, "naming an
5883 @cindex Naming an archive
5884 @cindex Archive Name
5885 @cindex Choosing an archive file
5886 @cindex Where is the archive?
5887 By default, @command{tar} uses an archive file name that was compiled when
5888 it was built on the system; usually this name refers to some physical
5889 tape drive on the machine. However, the person who installed @command{tar}
5890 on the system may not set the default to a meaningful value as far as
5891 most users are concerned. As a result, you will usually want to tell
5892 @command{tar} where to find (or create) the archive. The @option{--file=@var{archive-name}} (@option{-f @var{archive-name}})
5893 option allows you to either specify or name a file to use as the archive
5894 instead of the default archive file location.
5897 @opindex file, short description
5898 @item --file=@var{archive-name}
5899 @itemx -f @var{archive-name}
5900 Name the archive to create or operate on. Use in conjunction with
5904 For example, in this @command{tar} command,
5907 $ @kbd{tar -cvf collection.tar blues folk jazz}
5911 @file{collection.tar} is the name of the archive. It must directly
5912 follow the @option{-f} option, since whatever directly follows @option{-f}
5913 @emph{will} end up naming the archive. If you neglect to specify an
5914 archive name, you may end up overwriting a file in the working directory
5915 with the archive you create since @command{tar} will use this file's name
5916 for the archive name.
5918 An archive can be saved as a file in the file system, sent through a
5919 pipe or over a network, or written to an I/O device such as a tape,
5920 floppy disk, or CD write drive.
5922 @cindex Writing new archives
5923 @cindex Archive creation
5924 If you do not name the archive, @command{tar} uses the value of the
5925 environment variable @env{TAPE} as the file name for the archive. If
5926 that is not available, @command{tar} uses a default, compiled-in archive
5927 name, usually that for tape unit zero (ie. @file{/dev/tu00}).
5928 @command{tar} always needs an archive name.
5930 If you use @file{-} as an @var{archive-name}, @command{tar} reads the
5931 archive from standard input (when listing or extracting files), or
5932 writes it to standard output (when creating an archive). If you use
5933 @file{-} as an @var{archive-name} when modifying an archive,
5934 @command{tar} reads the original archive from its standard input and
5935 writes the entire new archive to its standard output.
5937 @FIXME{might want a different example here; this is already used in
5938 "notable tar usages".}
5941 $ @kbd{cd sourcedir; tar -cf - . | (cd targetdir; tar -xf -)}
5946 @cindex Standard input and output
5947 @cindex tar to standard input and output
5949 To specify an archive file on a device attached to a remote machine,
5953 @kbd{--file=@var{hostname}:/@var{dev}/@var{file_name}}
5957 @command{tar} will complete the remote connection, if possible, and
5958 prompt you for a username and password. If you use
5959 @option{--file=@@@var{hostname}:/@var{dev}/@var{file_name}}, @command{tar}
5960 will complete the remote connection, if possible, using your username
5961 as the username on the remote machine.
5963 @cindex Local and remote archives
5964 @anchor{local and remote archives}
5965 If the archive file name includes a colon (@samp{:}), then it is assumed
5966 to be a file on another machine. If the archive file is
5967 @samp{@var{user}@@@var{host}:@var{file}}, then @var{file} is used on the
5968 host @var{host}. The remote host is accessed using the @command{rsh}
5969 program, with a username of @var{user}. If the username is omitted
5970 (along with the @samp{@@} sign), then your user name will be used.
5971 (This is the normal @command{rsh} behavior.) It is necessary for the
5972 remote machine, in addition to permitting your @command{rsh} access, to
5973 have the @file{rmt} program installed (This command is included in
5974 the @GNUTAR{} distribution and by default is installed under
5975 @file{@var{prefix}/libexec/rmt}, were @var{prefix} means your
5976 installation prefix). If you need to use a file whose name includes a
5977 colon, then the remote tape drive behavior
5978 can be inhibited by using the @option{--force-local} option.
5980 @FIXME{i know we went over this yesterday, but bob (and now i do again,
5981 too) thinks it's out of the middle of nowhere. it doesn't seem to tie
5982 into what came before it well enough <<i moved it now, is it better
5983 here?>>. bob also comments that if Amanda isn't free software, we
5984 shouldn't mention it..}
5986 When the archive is being created to @file{/dev/null}, @GNUTAR{}
5987 tries to minimize input and output operations. The
5988 Amanda backup system, when used with @GNUTAR{}, has
5989 an initial sizing pass which uses this feature.
5991 @node Selecting Archive Members
5992 @section Selecting Archive Members
5993 @cindex Specifying files to act on
5994 @cindex Specifying archive members
5996 @dfn{File Name arguments} specify which files in the file system
5997 @command{tar} operates on, when creating or adding to an archive, or which
5998 archive members @command{tar} operates on, when reading or deleting from
5999 an archive. @xref{Operations}.
6001 To specify file names, you can include them as the last arguments on
6002 the command line, as follows:
6004 @kbd{tar} @var{operation} [@var{option1} @var{option2} @dots{}] [@var{file name-1} @var{file name-2} @dots{}]
6007 If a file name begins with dash (@samp{-}), precede it with
6008 @option{--add-file} option to preventit from being treated as an
6011 If you specify a directory name as a file name argument, all the files
6012 in that directory are operated on by @command{tar}.
6014 If you do not specify files when @command{tar} is invoked with
6015 @option{--create} (@option{-c}), @command{tar} operates on all the non-directory files in
6016 the working directory. If you specify either @option{--list} (@option{-t}) or
6017 @option{--extract} (@option{--get}, @option{-x}), @command{tar} operates on all the archive members in the
6018 archive. If you specify any operation other than one of these three,
6019 @command{tar} does nothing.
6021 By default, @command{tar} takes file names from the command line. However,
6022 there are other ways to specify file or member names, or to modify the
6023 manner in which @command{tar} selects the files or members upon which to
6024 operate. @FIXME{add xref here}In general, these methods work both for
6025 specifying the names of files and archive members.
6028 @section Reading Names from a File
6030 @cindex Reading file names from a file
6031 @cindex Lists of file names
6032 @cindex File Name arguments, alternatives
6033 Instead of giving the names of files or archive members on the command
6034 line, you can put the names into a file, and then use the
6035 @option{--files-from=@var{file-of-names}} (@option{-T @var{file-of-names}}) option to @command{tar}. Give the name of the file
6036 which contains the list of files to include as the argument to
6037 @option{--files-from}. In the list, the file names should be separated by
6038 newlines. You will frequently use this option when you have generated
6039 the list of files to archive with the @command{find} utility.
6043 @item --files-from=@var{file_name}
6044 @itemx -T @var{file_name}
6045 Get names to extract or create from file @var{file_name}.
6048 If you give a single dash as a file name for @option{--files-from}, (i.e.,
6049 you specify either @code{--files-from=-} or @code{-T -}), then the file
6050 names are read from standard input.
6052 Unless you are running @command{tar} with @option{--create}, you can not use
6053 both @code{--files-from=-} and @code{--file=-} (@code{-f -}) in the same
6056 Any number of @option{-T} options can be given in the command line.
6058 @FIXME{add bob's example, from his message on 2-10-97}
6060 The following example shows how to use @command{find} to generate a list of
6061 files smaller than 400K in length and put that list into a file
6062 called @file{small-files}. You can then use the @option{-T} option to
6063 @command{tar} to specify the files from that file, @file{small-files}, to
6064 create the archive @file{little.tgz}. (The @option{-z} option to
6065 @command{tar} compresses the archive with @command{gzip}; @pxref{gzip} for
6069 $ @kbd{find . -size -400 -print > small-files}
6070 $ @kbd{tar -c -v -z -T small-files -f little.tgz}
6074 In the file list given by @option{-T} option, any file name beginning
6075 with @samp{-} character is considered a @command{tar} option and is
6076 processed accordingly.@footnote{Versions of @GNUTAR{} up to 1.15.1
6077 recognized only @option{-C} option in file lists, and only if the
6078 option and its argument occupied two consecutive lines.} For example,
6079 the common use of this feature is to change to another directory by
6080 specifying @option{-C} option:
6090 $ @kbd{tar -c -f foo.tar --files-from list}
6095 In this example, @command{tar} will first switch to @file{/etc}
6096 directory and add files @file{passwd} and @file{hosts} to the
6097 archive. Then it will change to @file{/lib} directory and will archive
6098 the file @file{libc.a}. Thus, the resulting archive @file{foo.tar} will
6103 $ @kbd{tar tf foo.tar}
6111 @opindex directory, using in @option{--files-from} argument
6112 Notice that the option parsing algorithm used with @option{-T} is
6113 stricter than the one used by shell. Namely, when specifying option
6114 arguments, you should observe the following rules:
6118 When using short (single-letter) option form, its argument must
6119 immediately follow the option letter, without any intervening
6120 whitespace. For example: @code{-Cdir}.
6123 When using long option form, the option argument must be separated
6124 from the option by a single equal sign. No whitespace is allowed on
6125 any side of the equal sign. For example: @code{--directory=dir}.
6128 For both short and long option forms, the option argument can be given
6129 on the next line after the option name, e.g.:
6150 If you happen to have a file whose name starts with @samp{-},
6151 precede it with @option{--add-file} option to prevent it from
6152 being recognized as an option. For example: @code{--add-file --my-file}.
6159 @subsection @code{NUL} Terminated File Names
6161 @cindex File names, terminated by @code{NUL}
6162 @cindex @code{NUL} terminated file names
6163 The @option{--null} option causes @option{--files-from=@var{file-of-names}} (@option{-T @var{file-of-names}}) to read file
6164 names terminated by a @code{NUL} instead of a newline, so files whose
6165 names contain newlines can be archived using @option{--files-from}.
6170 Only consider @code{NUL} terminated file names, instead of files that
6171 terminate in a newline.
6174 The @option{--null} option is just like the one in @acronym{GNU}
6175 @command{xargs} and @command{cpio}, and is useful with the
6176 @option{-print0} predicate of @acronym{GNU} @command{find}. In
6177 @command{tar}, @option{--null} also disables special handling for
6178 file names that begin with dash.
6180 This example shows how to use @command{find} to generate a list of files
6181 larger than 800K in length and put that list into a file called
6182 @file{long-files}. The @option{-print0} option to @command{find} is just
6183 like @option{-print}, except that it separates files with a @code{NUL}
6184 rather than with a newline. You can then run @command{tar} with both the
6185 @option{--null} and @option{-T} options to specify that @command{tar} get the
6186 files from that file, @file{long-files}, to create the archive
6187 @file{big.tgz}. The @option{--null} option to @command{tar} will cause
6188 @command{tar} to recognize the @code{NUL} separator between files.
6191 $ @kbd{find . -size +800 -print0 > long-files}
6192 $ @kbd{tar -c -v --null --files-from=long-files --file=big.tar}
6195 @FIXME{say anything else here to conclude the section?}
6198 @section Excluding Some Files
6201 @cindex File names, excluding files by
6202 @cindex Excluding files by name and pattern
6203 @cindex Excluding files by file system
6204 To avoid operating on files whose names match a particular pattern,
6205 use the @option{--exclude} or @option{--exclude-from} options.
6209 @item --exclude=@var{pattern}
6210 Causes @command{tar} to ignore files that match the @var{pattern}.
6214 The @option{--exclude=@var{pattern}} option prevents any file or member whose name
6215 matches the shell wildcard (@var{pattern}) from being operated on.
6216 For example, to create an archive with all the contents of the directory
6217 @file{src} except for files whose names end in @file{.o}, use the
6218 command @samp{tar -cf src.tar --exclude='*.o' src}.
6220 You may give multiple @option{--exclude} options.
6223 @opindex exclude-from
6224 @item --exclude-from=@var{file}
6225 @itemx -X @var{file}
6226 Causes @command{tar} to ignore files that match the patterns listed in
6230 @findex exclude-from
6231 Use the @option{--exclude-from} option to read a
6232 list of patterns, one per line, from @var{file}; @command{tar} will
6233 ignore files matching those patterns. Thus if @command{tar} is
6234 called as @w{@samp{tar -c -X foo .}} and the file @file{foo} contains a
6235 single line @file{*.o}, no files whose names end in @file{.o} will be
6236 added to the archive.
6238 @FIXME{do the exclude options files need to have stuff separated by
6239 newlines the same as the files-from option does?}
6242 @opindex exclude-caches
6243 @item --exclude-caches
6244 Causes @command{tar} to ignore directories containing a cache directory tag.
6247 @findex exclude-caches
6248 When creating an archive, the @option{--exclude-caches} option causes
6249 @command{tar} to exclude all directories that contain a @dfn{cache
6250 directory tag}. A cache directory tag is a short file with the
6251 well-known name @file{CACHEDIR.TAG} and having a standard header
6252 specified in @url{http://www.brynosaurus.com/cachedir/spec.html}.
6253 Various applications write cache directory tags into directories they
6254 use to hold regenerable, non-precious data, so that such data can be
6255 more easily excluded from backups.
6258 * problems with exclude::
6261 @node problems with exclude
6262 @unnumberedsubsec Problems with Using the @code{exclude} Options
6264 @opindex exclude, potential problems with
6265 Some users find @samp{exclude} options confusing. Here are some common
6270 The main operating mode of @command{tar} does not act on a path name
6271 explicitly listed on the command line if one of its file name
6272 components is excluded. In the example above, if
6273 you create an archive and exclude files that end with @samp{*.o}, but
6274 explicitly name the file @samp{dir.o/foo} after all the options have been
6275 listed, @samp{dir.o/foo} will be excluded from the archive.
6278 You can sometimes confuse the meanings of @option{--exclude} and
6279 @option{--exclude-from}. Be careful: use @option{--exclude} when files
6280 to be excluded are given as a pattern on the command line. Use
6281 @option{--exclude-from} to introduce the name of a file which contains
6282 a list of patterns, one per line; each of these patterns can exclude
6283 zero, one, or many files.
6286 When you use @option{--exclude=@var{pattern}}, be sure to quote the @var{pattern}
6287 parameter, so @GNUTAR{} sees wildcard characters
6288 like @samp{*}. If you do not do this, the shell might expand the
6289 @samp{*} itself using files at hand, so @command{tar} might receive a
6290 list of files instead of one pattern, or none at all, making the
6291 command somewhat illegal. This might not correspond to what you want.
6296 $ @kbd{tar -c -f @var{archive.tar} --exclude '*.o' @var{directory}}
6303 $ @kbd{tar -c -f @var{archive.tar} --exclude *.o @var{directory}}
6307 You must use use shell syntax, or globbing, rather than @code{regexp}
6308 syntax, when using exclude options in @command{tar}. If you try to use
6309 @code{regexp} syntax to describe files to be excluded, your command
6313 In earlier versions of @command{tar}, what is now the
6314 @option{--exclude-from} option was called @option{--exclude} instead.
6315 Now, @option{--exclude} applies to patterns listed on the command
6316 line and @option{--exclude-from} applies to patterns listed in a
6322 @section Wildcards Patterns and Matching
6324 @dfn{Globbing} is the operation by which @dfn{wildcard} characters,
6325 @samp{*} or @samp{?} for example, are replaced and expanded into all
6326 existing files matching the given pattern. However, @command{tar} often
6327 uses wildcard patterns for matching (or globbing) archive members instead
6328 of actual files in the file system. Wildcard patterns are also used for
6329 verifying volume labels of @command{tar} archives. This section has the
6330 purpose of explaining wildcard syntax for @command{tar}.
6332 @FIXME{the next few paragraphs need work.}
6334 A @var{pattern} should be written according to shell syntax, using wildcard
6335 characters to effect globbing. Most characters in the pattern stand
6336 for themselves in the matched string, and case is significant: @samp{a}
6337 will match only @samp{a}, and not @samp{A}. The character @samp{?} in the
6338 pattern matches any single character in the matched string. The character
6339 @samp{*} in the pattern matches zero, one, or more single characters in
6340 the matched string. The character @samp{\} says to take the following
6341 character of the pattern @emph{literally}; it is useful when one needs to
6342 match the @samp{?}, @samp{*}, @samp{[} or @samp{\} characters, themselves.
6344 The character @samp{[}, up to the matching @samp{]}, introduces a character
6345 class. A @dfn{character class} is a list of acceptable characters
6346 for the next single character of the matched string. For example,
6347 @samp{[abcde]} would match any of the first five letters of the alphabet.
6348 Note that within a character class, all of the ``special characters''
6349 listed above other than @samp{\} lose their special meaning; for example,
6350 @samp{[-\\[*?]]} would match any of the characters, @samp{-}, @samp{\},
6351 @samp{[}, @samp{*}, @samp{?}, or @samp{]}. (Due to parsing constraints,
6352 the characters @samp{-} and @samp{]} must either come @emph{first} or
6353 @emph{last} in a character class.)
6355 @cindex Excluding characters from a character class
6356 @cindex Character class, excluding characters from
6357 If the first character of the class after the opening @samp{[}
6358 is @samp{!} or @samp{^}, then the meaning of the class is reversed.
6359 Rather than listing character to match, it lists those characters which
6360 are @emph{forbidden} as the next single character of the matched string.
6362 Other characters of the class stand for themselves. The special
6363 construction @samp{[@var{a}-@var{e}]}, using an hyphen between two
6364 letters, is meant to represent all characters between @var{a} and
6367 @FIXME{need to add a sentence or so here to make this clear for those
6368 who don't have dan around.}
6370 Periods (@samp{.}) or forward slashes (@samp{/}) are not considered
6371 special for wildcard matches. However, if a pattern completely matches
6372 a directory prefix of a matched string, then it matches the full matched
6373 string: excluding a directory also excludes all the files beneath it.
6376 * controlling pattern-matching::
6379 @node controlling pattern-matching
6380 @unnumberedsubsec Controlling Pattern-Matching
6383 Normally, a pattern matches a name if an initial subsequence of the
6384 name's components matches the pattern, where @samp{*}, @samp{?}, and
6385 @samp{[...]} are the usual shell wildcards, @samp{\} escapes wildcards,
6386 and wildcards can match @samp{/}.
6388 Other than optionally stripping leading @samp{/} from names
6389 (@pxref{absolute}), patterns and names are used as-is. For
6390 example, trailing @samp{/} is not trimmed from a user-specified name
6391 before deciding whether to exclude it.
6393 However, this matching procedure can be altered by the options listed
6394 below. These options accumulate. For example:
6397 --ignore-case --exclude='makefile' --no-ignore-case ---exclude='readme'
6400 ignores case when excluding @samp{makefile}, but not when excluding
6405 @opindex no-anchored
6407 @itemx --no-anchored
6408 If anchored, a pattern must match an initial subsequence
6409 of the name's components. Otherwise, the pattern can match any
6410 subsequence. Default is @option{--no-anchored}.
6412 @opindex ignore-case
6413 @opindex no-ignore-case
6415 @itemx --no-ignore-case
6416 When ignoring case, upper-case patterns match lower-case names and vice versa.
6417 When not ignoring case (the default), matching is case-sensitive.
6420 @opindex no-wildcards
6422 @itemx --no-wildcards
6423 When using wildcards (the default), @samp{*}, @samp{?}, and @samp{[...]}
6424 are the usual shell wildcards, and @samp{\} escapes wildcards.
6425 Otherwise, none of these characters are special, and patterns must match
6428 @opindex wildcards-match-slash
6429 @opindex no-wildcards-match-slash
6430 @item --wildcards-match-slash
6431 @itemx --no-wildcards-match-slash
6432 When wildcards match slash (the default), a wildcard like @samp{*} in
6433 the pattern can match a @samp{/} in the name. Otherwise, @samp{/} is
6434 matched only by @samp{/}.
6438 The @option{--recursion} and @option{--no-recursion} options
6439 (@pxref{recurse}) also affect how exclude patterns are interpreted. If
6440 recursion is in effect, a pattern excludes a name if it matches any of
6441 the name's parent directories.
6444 @section Operating Only on New Files
6447 @cindex Excluding file by age
6448 @cindex Data Modification time, excluding files by
6449 @cindex Modification time, excluding files by
6450 @cindex Age, excluding files by
6451 The @option{--after-date=@var{date}} (@option{--newer=@var{date}},
6452 @option{-N @var{date}}) option causes @command{tar} to only work on
6453 files whose data modification or status change times are newer than
6454 the @var{date} given. If @var{date} starts with @samp{/} or @samp{.},
6455 it is taken to be a file name; the data modification time of that file
6456 is used as the date. If you use this option when creating or appending
6457 to an archive, the archive will only include new files. If you use
6458 @option{--after-date} when extracting an archive, @command{tar} will
6459 only extract files newer than the @var{date} you specify.
6461 If you only want @command{tar} to make the date comparison based on
6462 modification of the file's data (rather than status
6463 changes), then use the @option{--newer-mtime=@var{date}} option.
6465 You may use these options with any operation. Note that these options
6466 differ from the @option{--update} (@option{-u}) operation in that they
6467 allow you to specify a particular date against which @command{tar} can
6468 compare when deciding whether or not to archive the files.
6473 @item --after-date=@var{date}
6474 @itemx --newer=@var{date}
6475 @itemx -N @var{date}
6476 Only store files newer than @var{date}.
6478 Acts on files only if their data modification or status change times are
6479 later than @var{date}. Use in conjunction with any operation.
6481 If @var{date} starts with @samp{/} or @samp{.}, it is taken to be a file
6482 name; the data modification time of that file is used as the date.
6484 @opindex newer-mtime
6485 @item --newer-mtime=@var{date}
6486 Acts like @option{--after-date}, but only looks at data modification times.
6489 These options limit @command{tar} to operate only on files which have
6490 been modified after the date specified. A file's status is considered to have
6491 changed if its contents have been modified, or if its owner,
6492 permissions, and so forth, have been changed. (For more information on
6493 how to specify a date, see @ref{Date input formats}; remember that the
6494 entire date argument must be quoted if it contains any spaces.)
6496 Gurus would say that @option{--after-date} tests both the data
6497 modification time (@code{mtime}, the time the contents of the file
6498 were last modified) and the status change time (@code{ctime}, the time
6499 the file's status was last changed: owner, permissions, etc.@:)
6500 fields, while @option{--newer-mtime} tests only the @code{mtime}
6503 To be precise, @option{--after-date} checks @emph{both} @code{mtime} and
6504 @code{ctime} and processes the file if either one is more recent than
6505 @var{date}, while @option{--newer-mtime} only checks @code{mtime} and
6506 disregards @code{ctime}. Neither does it use @code{atime} (the last time the
6507 contents of the file were looked at).
6509 Date specifiers can have embedded spaces. Because of this, you may need
6510 to quote date arguments to keep the shell from parsing them as separate
6513 @FIXME{Need example of --newer-mtime with quoted argument.}
6516 @strong{Please Note:} @option{--after-date} and @option{--newer-mtime}
6517 should not be used for incremental backups. Some files (such as those
6518 in renamed directories) are not selected properly by these options.
6519 @xref{Incremental Dumps}.
6523 @FIXME{which tells -- need to fill this in!}
6526 @section Descending into Directories
6528 @cindex Avoiding recursion in directories
6529 @cindex Descending directories, avoiding
6530 @cindex Directories, avoiding recursion
6531 @cindex Recursion in directories, avoiding
6533 @FIXME{arrggh! this is still somewhat confusing to me. :-< }
6535 @FIXME{show dan bob's comments, from 2-10-97}
6537 Usually, @command{tar} will recursively explore all directories (either
6538 those given on the command line or through the @option{--files-from}
6539 option) for the various files they contain. However, you may not always
6540 want @command{tar} to act this way.
6542 @opindex no-recursion
6543 The @option{--no-recursion} option inhibits @command{tar}'s recursive descent
6544 into specified directories. If you specify @option{--no-recursion}, you can
6545 use the @command{find} utility for hunting through levels of directories to
6546 construct a list of file names which you could then pass to @command{tar}.
6547 @command{find} allows you to be more selective when choosing which files to
6548 archive; see @ref{files} for more information on using @command{find} with
6549 @command{tar}, or look.
6552 @item --no-recursion
6553 Prevents @command{tar} from recursively descending directories.
6557 Requires @command{tar} to recursively descend directories.
6558 This is the default.
6561 When you use @option{--no-recursion}, @GNUTAR{} grabs
6562 directory entries themselves, but does not descend on them
6563 recursively. Many people use @command{find} for locating files they
6564 want to back up, and since @command{tar} @emph{usually} recursively
6565 descends on directories, they have to use the @samp{@w{! -d}} option
6566 to @command{find} @FIXME{needs more explanation or a cite to another
6567 info file}as they usually do not want all the files in a directory.
6568 They then use the @option{--files-from} option to archive the files
6569 located via @command{find}.
6571 The problem when restoring files archived in this manner is that the
6572 directories themselves are not in the archive; so the
6573 @option{--same-permissions} (@option{--preserve-permissions},
6574 @option{-p}) option does not affect them---while users might really
6575 like it to. Specifying @option{--no-recursion} is a way to tell
6576 @command{tar} to grab only the directory entries given to it, adding
6577 no new files on its own.
6579 The @option{--no-recursion} option also applies when extracting: it
6580 causes @command{tar} to extract only the matched directory entries, not
6581 the files under those directories.
6583 The @option{--no-recursion} option also affects how globbing patterns
6584 are interpreted (@pxref{controlling pattern-matching}).
6586 The @option{--no-recursion} and @option{--recursion} options apply to
6587 later options and operands, and can be overridden by later occurrences
6588 of @option{--no-recursion} and @option{--recursion}. For example:
6591 $ @kbd{tar -cf jams.tar --no-recursion grape --recursion grape/concord}
6595 creates an archive with one entry for @file{grape}, and the recursive
6596 contents of @file{grape/concord}, but no entries under @file{grape}
6597 other than @file{grape/concord}.
6600 @section Crossing File System Boundaries
6601 @cindex File system boundaries, not crossing
6604 @command{tar} will normally automatically cross file system boundaries in
6605 order to archive files which are part of a directory tree. You can
6606 change this behavior by running @command{tar} and specifying
6607 @option{--one-file-system}. This option only affects files that are
6608 archived because they are in a directory that is being archived;
6609 @command{tar} will still archive files explicitly named on the command line
6610 or through @option{--files-from}, regardless of where they reside.
6613 @opindex one-file-system
6614 @item --one-file-system
6616 Prevents @command{tar} from crossing file system boundaries when
6617 archiving. Use in conjunction with any write operation.
6620 The @option{--one-file-system} option causes @command{tar} to modify its
6621 normal behavior in archiving the contents of directories. If a file in
6622 a directory is not on the same file system as the directory itself, then
6623 @command{tar} will not archive that file. If the file is a directory
6624 itself, @command{tar} will not archive anything beneath it; in other words,
6625 @command{tar} will not cross mount points.
6627 It is reported that using this option, the mount point is is archived,
6628 but nothing under it.
6630 This option is useful for making full or incremental archival backups of
6631 a file system. If this option is used in conjunction with
6632 @option{--verbose} (@option{-v}), files that are excluded are mentioned by name on the
6636 * directory:: Changing Directory
6637 * absolute:: Absolute File Names
6641 @subsection Changing the Working Directory
6644 @FIXME{need to read over this node now for continuity; i've switched
6645 things around some.}
6647 @cindex Changing directory mid-stream
6648 @cindex Directory, changing mid-stream
6649 @cindex Working directory, specifying
6650 To change the working directory in the middle of a list of file names,
6651 either on the command line or in a file specified using
6652 @option{--files-from} (@option{-T}), use @option{--directory} (@option{-C}).
6653 This will change the working directory to the specified directory
6654 after that point in the list.
6658 @item --directory=@var{directory}
6659 @itemx -C @var{directory}
6660 Changes the working directory in the middle of a command line.
6666 $ @kbd{tar -c -f jams.tar grape prune -C food cherry}
6670 will place the files @file{grape} and @file{prune} from the current
6671 directory into the archive @file{jams.tar}, followed by the file
6672 @file{cherry} from the directory @file{food}. This option is especially
6673 useful when you have several widely separated files that you want to
6674 store in the same archive.
6676 Note that the file @file{cherry} is recorded in the archive under the
6677 precise name @file{cherry}, @emph{not} @file{food/cherry}. Thus, the
6678 archive will contain three files that all appear to have come from the
6679 same directory; if the archive is extracted with plain @samp{tar
6680 --extract}, all three files will be written in the current directory.
6682 Contrast this with the command,
6685 $ @kbd{tar -c -f jams.tar grape prune -C food red/cherry}
6689 which records the third file in the archive under the name
6690 @file{red/cherry} so that, if the archive is extracted using
6691 @samp{tar --extract}, the third file will be written in a subdirectory
6692 named @file{orange-colored}.
6694 You can use the @option{--directory} option to make the archive
6695 independent of the original name of the directory holding the files.
6696 The following command places the files @file{/etc/passwd},
6697 @file{/etc/hosts}, and @file{/lib/libc.a} into the archive
6701 $ @kbd{tar -c -f foo.tar -C /etc passwd hosts -C /lib libc.a}
6705 However, the names of the archive members will be exactly what they were
6706 on the command line: @file{passwd}, @file{hosts}, and @file{libc.a}.
6707 They will not appear to be related by file name to the original
6708 directories where those files were located.
6710 Note that @option{--directory} options are interpreted consecutively. If
6711 @option{--directory} specifies a relative file name, it is interpreted
6712 relative to the then current directory, which might not be the same as
6713 the original current working directory of @command{tar}, due to a previous
6714 @option{--directory} option.
6716 When using @option{--files-from} (@pxref{files}), you can put various
6717 @command{tar} options (including @option{-C}) in the file list. Notice,
6718 however, that in this case the option and its argument may not be
6719 separated by whitespace. If you use short option, its argument must
6720 either follow the option letter immediately, without any intervening
6721 whitespace, or occupy the next line. Otherwise, if you use long
6722 option, separate its argument by an equal sign.
6724 For instance, the file list for the above example will be:
6739 To use it, you would invoke @command{tar} as follows:
6742 $ @kbd{tar -c -f foo.tar --files-from list}
6745 Notice also that you can only use the short option variant in the file
6746 list, i.e., always use @option{-C}, not @option{--directory}.
6748 The interpretation of @option{--directory} is disabled by
6749 @option{--null} option.
6752 @subsection Absolute File Names
6756 @opindex absolute-names
6757 @item --absolute-names
6759 Do not strip leading slashes from file names, and permit file names
6760 containing a @file{..} file name component.
6763 By default, @GNUTAR{} drops a leading @samp{/} on
6764 input or output, and complains about file names containing a @file{..}
6765 component. This option turns off this behavior.
6767 When @command{tar} extracts archive members from an archive, it strips any
6768 leading slashes (@samp{/}) from the member name. This causes absolute
6769 member names in the archive to be treated as relative file names. This
6770 allows you to have such members extracted wherever you want, instead of
6771 being restricted to extracting the member in the exact directory named
6772 in the archive. For example, if the archive member has the name
6773 @file{/etc/passwd}, @command{tar} will extract it as if the name were
6774 really @file{etc/passwd}.
6776 File names containing @file{..} can cause problems when extracting, so
6777 @command{tar} normally warns you about such files when creating an
6778 archive, and rejects attempts to extracts such files.
6780 Other @command{tar} programs do not do this. As a result, if you
6781 create an archive whose member names start with a slash, they will be
6782 difficult for other people with a non-@GNUTAR{}
6783 program to use. Therefore, @GNUTAR{} also strips
6784 leading slashes from member names when putting members into the
6785 archive. For example, if you ask @command{tar} to add the file
6786 @file{/bin/ls} to an archive, it will do so, but the member name will
6787 be @file{bin/ls}.@footnote{A side effect of this is that when
6788 @option{--create} is used with @option{--verbose} the resulting output
6789 is not, generally speaking, the same as the one you'd get running
6790 @kbd{tar --list} command. This may be important if you use some
6791 scripts for comparing both outputs. @xref{listing member and file names},
6792 for the information on how to handle this case.}
6794 If you use the @option{--absolute-names} (@option{-P}) option,
6795 @command{tar} will do none of these transformations.
6797 To archive or extract files relative to the root directory, specify
6798 the @option{--absolute-names} (@option{-P}) option.
6800 Normally, @command{tar} acts on files relative to the working
6801 directory---ignoring superior directory names when archiving, and
6802 ignoring leading slashes when extracting.
6804 When you specify @option{--absolute-names} (@option{-P}),
6805 @command{tar} stores file names including all superior directory
6806 names, and preserves leading slashes. If you only invoked
6807 @command{tar} from the root directory you would never need the
6808 @option{--absolute-names} option, but using this option
6809 may be more convenient than switching to root.
6811 @FIXME{Should be an example in the tutorial/wizardry section using this
6812 to transfer files between systems.}
6814 @FIXME{Is write access an issue?}
6817 @item --absolute-names
6818 Preserves full file names (including superior directory names) when
6819 archiving files. Preserves leading slash when extracting files.
6823 @FIXME{this is still horrible; need to talk with dan on monday.}
6825 @command{tar} prints out a message about removing the @samp{/} from
6826 file names. This message appears once per @GNUTAR{}
6827 invocation. It represents something which ought to be told; ignoring
6828 what it means can cause very serious surprises, later.
6830 Some people, nevertheless, do not want to see this message. Wanting to
6831 play really dangerously, one may of course redirect @command{tar} standard
6832 error to the sink. For example, under @command{sh}:
6835 $ @kbd{tar -c -f archive.tar /home 2> /dev/null}
6839 Another solution, both nicer and simpler, would be to change to
6840 the @file{/} directory first, and then avoid absolute notation.
6844 $ @kbd{(cd / && tar -c -f archive.tar home)}
6845 $ @kbd{tar -c -f archive.tar -C / home}
6848 @include getdate.texi
6851 @chapter Controlling the Archive Format
6853 @cindex Tar archive formats
6854 Due to historical reasons, there are several formats of tar archives.
6855 All of them are based on the same principles, but have some subtle
6856 differences that often make them incompatible with each other.
6858 GNU tar is able to create and handle archives in a variety of formats.
6859 The most frequently used formats are (in alphabetical order):
6863 Format used by @GNUTAR{} versions up to 1.13.25. This format derived
6864 from an early @acronym{POSIX} standard, adding some improvements such as
6865 sparse file handling and incremental archives. Unfortunately these
6866 features were implemented in a way incompatible with other archive
6869 Archives in @samp{gnu} format are able to hold pathnames of unlimited
6873 Format used by @GNUTAR{} of versions prior to 1.12.
6876 Archive format, compatible with the V7 implementation of tar. This
6877 format imposes a number of limitations. The most important of them
6881 @item The maximum length of a file name is limited to 99 characters.
6882 @item The maximum length of a symbolic link is limited to 99 characters.
6883 @item It is impossible to store special files (block and character
6884 devices, fifos etc.)
6885 @item Maximum value of user or group ID is limited to 2097151 (7777777
6887 @item V7 archives do not contain symbolic ownership information (user
6888 and group name of the file owner).
6891 This format has traditionally been used by Automake when producing
6892 Makefiles. This practice will change in the future, in the meantime,
6893 however this means that projects containing filenames more than 99
6894 characters long will not be able to use @GNUTAR{} @value{VERSION} and
6895 Automake prior to 1.9.
6898 Archive format defined by @acronym{POSIX.1-1988} specification. It stores
6899 symbolic ownership information. It is also able to store
6900 special files. However, it imposes several restrictions as well:
6903 @item The maximum length of a file name is limited to 256 characters,
6904 provided that the filename can be split at directory separator in
6905 two parts, first of them being at most 155 bytes long. So, in most
6906 cases the maximum file name length will be shorter than 256
6908 @item The maximum length of a symbolic link name is limited to
6910 @item Maximum size of a file the archive is able to accomodate
6912 @item Maximum value of UID/GID is 2097151.
6913 @item Maximum number of bits in device major and minor numbers is 21.
6917 Format used by J@"org Schilling @command{star}
6918 implementation. @GNUTAR{} is able to read @samp{star} archives but
6919 currently does not produce them.
6922 Archive format defined by @acronym{POSIX.1-2001} specification. This is the
6923 most flexible and feature-rich format. It does not impose any
6924 restrictions on file sizes or filename lengths. This format is quite
6925 recent, so not all tar implementations are able to handle it properly.
6926 However, this format is designed in such a way that any tar
6927 implementation able to read @samp{ustar} archives will be able to read
6928 most @samp{posix} archives as well, with the only exception that any
6929 additional information (such as long file names etc.) will in such
6930 case be extracted as plain text files along with the files it refers to.
6932 This archive format will be the default format for future versions
6937 The following table summarizes the limitations of each of these
6940 @multitable @columnfractions .10 .20 .20 .20 .20
6941 @headitem Format @tab UID @tab File Size @tab Path Name @tab Devn
6942 @item gnu @tab 1.8e19 @tab Unlimited @tab Unlimited @tab 63
6943 @item oldgnu @tab 1.8e19 @tab Unlimited @tab Unlimited @tab 63
6944 @item v7 @tab 2097151 @tab 8GB @tab 99 @tab n/a
6945 @item ustar @tab 2097151 @tab 8GB @tab 256 @tab 21
6946 @item posix @tab Unlimited @tab Unlimited @tab Unlimited @tab Unlimited
6949 The default format for @GNUTAR{} is defined at compilation
6950 time. You may check it by running @command{tar --help}, and examining
6951 the last lines of its output. Usually, @GNUTAR{} is configured
6952 to create archives in @samp{gnu} format, however, future version will
6953 switch to @samp{posix}.
6956 * Portability:: Making @command{tar} Archives More Portable
6957 * Compression:: Using Less Space through Compression
6958 * Attributes:: Handling File Attributes
6959 * Standard:: The Standard Format
6960 * Extensions:: @acronym{GNU} Extensions to the Archive Format
6961 * cpio:: Comparison of @command{tar} and @command{cpio}
6965 @section Making @command{tar} Archives More Portable
6967 Creating a @command{tar} archive on a particular system that is meant to be
6968 useful later on many other machines and with other versions of @command{tar}
6969 is more challenging than you might think. @command{tar} archive formats
6970 have been evolving since the first versions of Unix. Many such formats
6971 are around, and are not always compatible with each other. This section
6972 discusses a few problems, and gives some advice about making @command{tar}
6973 archives more portable.
6975 One golden rule is simplicity. For example, limit your @command{tar}
6976 archives to contain only regular files and directories, avoiding
6977 other kind of special files. Do not attempt to save sparse files or
6978 contiguous files as such. Let's discuss a few more problems, in turn.
6980 @FIXME{Discuss GNU extensions (incremental backups, multi-volume
6981 archives and archive labels) in GNU and PAX formats.}
6984 * Portable Names:: Portable Names
6985 * dereference:: Symbolic Links
6986 * old:: Old V7 Archives
6987 * ustar:: Ustar Archives
6988 * gnu:: GNU and old GNU format archives.
6989 * posix:: @acronym{POSIX} archives
6990 * Checksumming:: Checksumming Problems
6991 * Large or Negative Values:: Large files, negative time stamps, etc.
6994 @node Portable Names
6995 @subsection Portable Names
6997 Use portable file and member names. A name is portable if it contains
6998 only ASCII letters and digits, @samp{/}, @samp{.}, @samp{_}, and
6999 @samp{-}; it cannot be empty, start with @samp{-} or @samp{//}, or
7000 contain @samp{/-}. Avoid deep directory nesting. For portability to
7001 old Unix hosts, limit your file name components to 14 characters or
7004 If you intend to have your @command{tar} archives to be read under
7005 MSDOS, you should not rely on case distinction for file names, and you
7006 might use the @acronym{GNU} @command{doschk} program for helping you
7007 further diagnosing illegal MSDOS names, which are even more limited
7011 @subsection Symbolic Links
7012 @cindex File names, using symbolic links
7013 @cindex Symbolic link as file name
7015 @opindex dereference
7016 Normally, when @command{tar} archives a symbolic link, it writes a
7017 block to the archive naming the target of the link. In that way, the
7018 @command{tar} archive is a faithful record of the file system contents.
7019 @option{--dereference} (@option{-h}) is used with @option{--create} (@option{-c}), and causes
7020 @command{tar} to archive the files symbolic links point to, instead of
7021 the links themselves. When this option is used, when @command{tar}
7022 encounters a symbolic link, it will archive the linked-to file,
7023 instead of simply recording the presence of a symbolic link.
7025 The name under which the file is stored in the file system is not
7026 recorded in the archive. To record both the symbolic link name and
7027 the file name in the system, archive the file under both names. If
7028 all links were recorded automatically by @command{tar}, an extracted file
7029 might be linked to a file name that no longer exists in the file
7032 If a linked-to file is encountered again by @command{tar} while creating
7033 the same archive, an entire second copy of it will be stored. (This
7034 @emph{might} be considered a bug.)
7036 So, for portable archives, do not archive symbolic links as such,
7037 and use @option{--dereference} (@option{-h}): many systems do not support
7038 symbolic links, and moreover, your distribution might be unusable if
7039 it contains unresolved symbolic links.
7042 @subsection Old V7 Archives
7043 @cindex Format, old style
7044 @cindex Old style format
7045 @cindex Old style archives
7046 @cindex v7 archive format
7048 Certain old versions of @command{tar} cannot handle additional
7049 information recorded by newer @command{tar} programs. To create an
7050 archive in V7 format (not ANSI), which can be read by these old
7051 versions, specify the @option{--format=v7} option in
7052 conjunction with the @option{--create} (@option{-c}) (@command{tar} also
7053 accepts @option{--portability} or @samp{op-old-archive} for this
7054 option). When you specify it,
7055 @command{tar} leaves out information about directories, pipes, fifos,
7056 contiguous files, and device files, and specifies file ownership by
7057 group and user IDs instead of group and user names.
7059 When updating an archive, do not use @option{--format=v7}
7060 unless the archive was created using this option.
7062 In most cases, a @emph{new} format archive can be read by an @emph{old}
7063 @command{tar} program without serious trouble, so this option should
7064 seldom be needed. On the other hand, most modern @command{tar}s are
7065 able to read old format archives, so it might be safer for you to
7066 always use @option{--format=v7} for your distributions.
7069 @subsection Ustar Archive Format
7071 @cindex ustar archive format
7072 Archive format defined by @acronym{POSIX}.1-1988 specification is called
7073 @code{ustar}. Although it is more flexible than the V7 format, it
7074 still has many restrictions (@xref{Formats,ustar}, for the detailed
7075 description of @code{ustar} format). Along with V7 format,
7076 @code{ustar} format is a good choice for archives intended to be read
7077 with other implementations of @command{tar}.
7079 To create archive in @code{ustar} format, use @option{--format=ustar}
7080 option in conjunction with the @option{--create} (@option{-c}).
7083 @subsection @acronym{GNU} and old @GNUTAR{} format
7085 @cindex GNU archive format
7086 @cindex Old GNU archive format
7087 @GNUTAR{} was based on an early draft of the
7088 @acronym{POSIX} 1003.1 @code{ustar} standard. @acronym{GNU} extensions to
7089 @command{tar}, such as the support for file names longer than 100
7090 characters, use portions of the @command{tar} header record which were
7091 specified in that @acronym{POSIX} draft as unused. Subsequent changes in
7092 @acronym{POSIX} have allocated the same parts of the header record for
7093 other purposes. As a result, @GNUTAR{} format is
7094 incompatible with the current @acronym{POSIX} specification, and with
7095 @command{tar} programs that follow it.
7097 In the majority of cases, @command{tar} will be configured to create
7098 this format by default. This will change in the future releases, since
7099 we plan to make @samp{posix} format the default.
7101 To force creation a @GNUTAR{} archive, use option
7102 @option{--format=gnu}.
7105 @subsection @GNUTAR{} and @acronym{POSIX} @command{tar}
7107 @cindex POSIX archive format
7108 @cindex PAX archive format
7109 The version @value{VERSION} of @GNUTAR{} is able
7110 to read and create archives conforming to @acronym{POSIX.1-2001} standard.
7112 A @acronym{POSIX} conformant archive will be created if @command{tar}
7113 was given @option{--format=posix} option.
7116 @subsection Checksumming Problems
7118 SunOS and HP-UX @command{tar} fail to accept archives created using
7119 @GNUTAR{} and containing non-ASCII file names, that
7120 is, file names having characters with the eight bit set, because they
7121 use signed checksums, while @GNUTAR{} uses unsigned
7122 checksums while creating archives, as per @acronym{POSIX} standards. On
7123 reading, @GNUTAR{} computes both checksums and
7124 accept any. It is somewhat worrying that a lot of people may go
7125 around doing backup of their files using faulty (or at least
7126 non-standard) software, not learning about it until it's time to
7127 restore their missing files with an incompatible file extractor, or
7130 @GNUTAR{} compute checksums both ways, and accept
7131 any on read, so @acronym{GNU} tar can read Sun tapes even with their
7132 wrong checksums. @GNUTAR{} produces the standard
7133 checksum, however, raising incompatibilities with Sun. That is to
7134 say, @GNUTAR{} has not been modified to
7135 @emph{produce} incorrect archives to be read by buggy @command{tar}'s.
7136 I've been told that more recent Sun @command{tar} now read standard
7137 archives, so maybe Sun did a similar patch, after all?
7139 The story seems to be that when Sun first imported @command{tar}
7140 sources on their system, they recompiled it without realizing that
7141 the checksums were computed differently, because of a change in
7142 the default signing of @code{char}'s in their compiler. So they
7143 started computing checksums wrongly. When they later realized their
7144 mistake, they merely decided to stay compatible with it, and with
7145 themselves afterwards. Presumably, but I do not really know, HP-UX
7146 has chosen that their @command{tar} archives to be compatible with Sun's.
7147 The current standards do not favor Sun @command{tar} format. In any
7148 case, it now falls on the shoulders of SunOS and HP-UX users to get
7149 a @command{tar} able to read the good archives they receive.
7151 @node Large or Negative Values
7152 @subsection Large or Negative Values
7153 @cindex large values
7154 @cindex future time stamps
7155 @cindex negative time stamps
7158 The above sections suggest to use @samp{oldest possible} archive
7159 format if in doubt. However, sometimes it is not possible. If you
7160 attempt to archive a file whose metadata cannot be represented using
7161 required format, @GNUTAR{} will print error message and ignore such a
7162 file. You will than have to switch to a format that is able to
7163 handle such values. The format summary table (@pxref{Formats}) will
7166 In particular, when trying to archive files larger than 8GB or with
7167 timestamps not in the range 1970-01-01 00:00:00 through 2242-03-16
7168 12:56:31 @sc{utc}, you will have to chose between @acronym{GNU} and
7169 @acronym{POSIX} archive formats. When considering which format to
7170 choose, bear in mind that the @acronym{GNU} format uses
7171 two's-complement base-256 notation to store values that do not fit
7172 into standard @acronym{ustar} range. Such archives can generally be
7173 read only by a @GNUTAR{} implementation. Moreover, they sometimes
7174 cannot be correctly restored on another hosts even by @GNUTAR{}. For
7175 example, using two's complement representation for negative time
7176 stamps that assumes a signed 32-bit @code{time_t} generates archives
7177 that are not portable to hosts with differing @code{time_t}
7180 On the other hand, @acronym{POSIX} archives, generally speaking, can
7181 be extracted by any tar implementation that understands older
7182 @acronym{ustar} format. The only exception are files larger than 8GB.
7184 @FIXME{Describe how @acronym{POSIX} archives are extracted by non
7188 @section Using Less Space through Compression
7191 * gzip:: Creating and Reading Compressed Archives
7192 * sparse:: Archiving Sparse Files
7196 @subsection Creating and Reading Compressed Archives
7197 @cindex Compressed archives
7198 @cindex Storing archives in compressed format
7200 @GNUTAR{} is able to create and read compressed archives. It supports
7201 @command{gzip} and @command{bzip2} compression programs. For backward
7202 compatibilty, it also supports @command{compress} command, although
7203 we strongly recommend against using it, since there is a patent
7204 covering the algorithm it uses and you could be sued for patent
7205 infringement merely by running @command{compress}! Besides, it is less
7206 effective than @command{gzip} and @command{bzip2}.
7208 Creating a compressed archive is simple: you just specify a
7209 @dfn{compression option} along with the usual archive creation
7210 commands. The compression option is @option{-z} (@option{--gzip}) to
7211 create a @command{gzip} compressed archive, @option{-j}
7212 (@option{--bzip2}) to create a @command{bzip2} compressed archive, and
7213 @option{-Z} (@option{--compress}) to use @command{compress} program.
7217 $ @kbd{tar cfz archive.tar.gz .}
7220 Reading compressed archive is even simpler: you don't need to specify
7221 any additional options as @GNUTAR{} recognizes its format
7222 automatically. Thus, the following commands will list and extract the
7223 archive created in previous example:
7226 # List the compressed archive
7227 $ @kbd{tar tf archive.tar.gz}
7228 # Extract the compressed archive
7229 $ @kbd{tar xf archive.tar.gz}
7232 The only case when you have to specify a decompression option while
7233 reading the archive is when reading from a pipe or from a tape drive
7234 that does not support random access. However, in this case @GNUTAR{}
7235 will indicate which option you should use. For example:
7238 $ @kbd{cat archive.tar.gz | tar tf -}
7239 tar: Archive is compressed. Use -z option
7240 tar: Error is not recoverable: exiting now
7243 If you see such diagnostics, just add the suggested option to the
7244 invocation of @GNUTAR{}:
7247 $ @kbd{cat archive.tar.gz | tar tfz -}
7250 Notice also, that there are several restrictions on operations on
7251 compressed archives. First of all, compressed archives cannot be
7252 modified, i.e., you cannot update (@option{--update} (@option{-u})) them or delete
7253 (@option{--delete}) members from them. Likewise, you cannot append
7254 another @command{tar} archive to a compressed archive using
7255 @option{--append} (@option{-r})). Secondly, multi-volume archives cannot be
7258 The following table summarizes compression options used by @GNUTAR{}.
7266 Filter the archive through @command{gzip}.
7268 You can use @option{--gzip} and @option{--gunzip} on physical devices
7269 (tape drives, etc.) and remote files as well as on normal files; data
7270 to or from such devices or remote files is reblocked by another copy
7271 of the @command{tar} program to enforce the specified (or default) record
7272 size. The default compression parameters are used; if you need to
7273 override them, set @env{GZIP} environment variable, e.g.:
7276 $ @kbd{GZIP=--best tar cfz archive.tar.gz subdir}
7280 Another way would be to avoid the @option{--gzip} (@option{--gunzip}, @option{--ungzip}, @option{-z}) option and run
7281 @command{gzip} explicitly:
7284 $ @kbd{tar cf - subdir | gzip --best -c - > archive.tar.gz}
7287 @cindex corrupted archives
7288 About corrupted compressed archives: @command{gzip}'ed files have no
7289 redundancy, for maximum compression. The adaptive nature of the
7290 compression scheme means that the compression tables are implicitly
7291 spread all over the archive. If you lose a few blocks, the dynamic
7292 construction of the compression tables becomes unsynchronized, and there
7293 is little chance that you could recover later in the archive.
7295 There are pending suggestions for having a per-volume or per-file
7296 compression in @GNUTAR{}. This would allow for viewing the
7297 contents without decompression, and for resynchronizing decompression at
7298 every volume or file, in case of corrupted archives. Doing so, we might
7299 lose some compressibility. But this would have make recovering easier.
7300 So, there are pros and cons. We'll see!
7305 Filter the archive through @code{bzip2}. Otherwise like @option{--gzip}.
7312 Filter the archive through @command{compress}. Otherwise like @option{--gzip}.
7314 The @acronym{GNU} Project recommends you not use
7315 @command{compress}, because there is a patent covering the algorithm it
7316 uses. You could be sued for patent infringement merely by running
7319 @opindex use-compress-program
7320 @item --use-compress-program=@var{prog}
7321 Use external compression program @var{prog}. Use this option if you
7322 have a compression program that @GNUTAR{} does not support. There
7323 are two requirements to which @var{prog} should comply:
7325 First, when called without options, it should read data from standard
7326 input, compress it and output it on standard output.
7328 Secondly, if called with @option{-d} argument, it should do exactly
7329 the opposite, i.e., read the compressed data from the standard input
7330 and produce uncompressed data on the standard output.
7333 @FIXME{I have one question, or maybe it's a suggestion if there isn't a way
7334 to do it now. I would like to use @option{--gzip}, but I'd also like
7335 the output to be fed through a program like @acronym{GNU}
7336 @command{ecc} (actually, right now that's @samp{exactly} what I'd like
7337 to use :-)), basically adding ECC protection on top of compression.
7338 It seems as if this should be quite easy to do, but I can't work out
7339 exactly how to go about it. Of course, I can pipe the standard output
7340 of @command{tar} through @command{ecc}, but then I lose (though I
7341 haven't started using it yet, I confess) the ability to have
7342 @command{tar} use @command{rmt} for it's I/O (I think).
7344 I think the most straightforward thing would be to let me specify a
7345 general set of filters outboard of compression (preferably ordered,
7346 so the order can be automatically reversed on input operations, and
7347 with the options they require specifiable), but beggars shouldn't be
7348 choosers and anything you decide on would be fine with me.
7350 By the way, I like @command{ecc} but if (as the comments say) it can't
7351 deal with loss of block sync, I'm tempted to throw some time at adding
7352 that capability. Supposing I were to actually do such a thing and
7353 get it (apparently) working, do you accept contributed changes to
7354 utilities like that? (Leigh Clayton @file{loc@@soliton.com}, May 1995).
7356 Isn't that exactly the role of the @option{--use-compress-prog=@var{program}} option?
7357 I never tried it myself, but I suspect you may want to write a
7358 @var{prog} script or program able to filter stdin to stdout to
7359 way you want. It should recognize the @option{-d} option, for when
7360 extraction is needed rather than creation.
7362 It has been reported that if one writes compressed data (through the
7363 @option{--gzip} or @option{--compress} options) to a DLT and tries to use
7364 the DLT compression mode, the data will actually get bigger and one will
7365 end up with less space on the tape.}
7368 @subsection Archiving Sparse Files
7369 @cindex Sparse Files
7376 Handle sparse files efficiently.
7379 This option causes all files to be put in the archive to be tested for
7380 sparseness, and handled specially if they are. The @option{--sparse}
7381 (@option{-S}) option is useful when many @code{dbm} files, for example, are being
7382 backed up. Using this option dramatically decreases the amount of
7383 space needed to store such a file.
7385 In later versions, this option may be removed, and the testing and
7386 treatment of sparse files may be done automatically with any special
7387 @acronym{GNU} options. For now, it is an option needing to be specified on
7388 the command line with the creation or updating of an archive.
7390 Files in the file system occasionally have ``holes.'' A hole in a file
7391 is a section of the file's contents which was never written. The
7392 contents of a hole read as all zeros. On many operating systems,
7393 actual disk storage is not allocated for holes, but they are counted
7394 in the length of the file. If you archive such a file, @command{tar}
7395 could create an archive longer than the original. To have @command{tar}
7396 attempt to recognize the holes in a file, use @option{--sparse} (@option{-S}). When
7397 you use this option, then, for any file using less disk space than
7398 would be expected from its length, @command{tar} searches the file for
7399 consecutive stretches of zeros. It then records in the archive for
7400 the file where the consecutive stretches of zeros are, and only
7401 archives the ``real contents'' of the file. On extraction (using
7402 @option{--sparse} is not needed on extraction) any such
7403 files have holes created wherever the continuous stretches of zeros
7404 were found. Thus, if you use @option{--sparse}, @command{tar} archives
7405 won't take more space than the original.
7407 A file is sparse if it contains blocks of zeros whose existence is
7408 recorded, but that have no space allocated on disk. When you specify
7409 the @option{--sparse} option in conjunction with the @option{--create}
7410 (@option{-c}) operation, @command{tar} tests all files for sparseness
7411 while archiving. If @command{tar} finds a file to be sparse, it uses a
7412 sparse representation of the file in the archive. @xref{create}, for
7413 more information about creating archives.
7415 @option{--sparse} is useful when archiving files, such as dbm files,
7416 likely to contain many nulls. This option dramatically
7417 decreases the amount of space needed to store such an archive.
7420 @strong{Please Note:} Always use @option{--sparse} when performing file
7421 system backups, to avoid archiving the expanded forms of files stored
7422 sparsely in the system.
7424 Even if your system has no sparse files currently, some may be
7425 created in the future. If you use @option{--sparse} while making file
7426 system backups as a matter of course, you can be assured the archive
7427 will never take more space on the media than the files take on disk
7428 (otherwise, archiving a disk filled with sparse files might take
7429 hundreds of tapes). @FIXME-xref{incremental when node name is set.}
7432 @command{tar} ignores the @option{--sparse} option when reading an archive.
7437 Files stored sparsely in the file system are represented sparsely in
7438 the archive. Use in conjunction with write operations.
7441 However, users should be well aware that at archive creation time,
7442 @GNUTAR{} still has to read whole disk file to
7443 locate the @dfn{holes}, and so, even if sparse files use little space
7444 on disk and in the archive, they may sometimes require inordinate
7445 amount of time for reading and examining all-zero blocks of a file.
7446 Although it works, it's painfully slow for a large (sparse) file, even
7447 though the resulting tar archive may be small. (One user reports that
7448 dumping a @file{core} file of over 400 megabytes, but with only about
7449 3 megabytes of actual data, took about 9 minutes on a Sun Sparcstation
7450 ELC, with full CPU utilization.)
7452 This reading is required in all cases and is not related to the fact
7453 the @option{--sparse} option is used or not, so by merely @emph{not}
7454 using the option, you are not saving time@footnote{Well! We should say
7455 the whole truth, here. When @option{--sparse} is selected while creating
7456 an archive, the current @command{tar} algorithm requires sparse files to be
7457 read twice, not once. We hope to develop a new archive format for saving
7458 sparse files in which one pass will be sufficient.}.
7460 Programs like @command{dump} do not have to read the entire file; by
7461 examining the file system directly, they can determine in advance
7462 exactly where the holes are and thus avoid reading through them. The
7463 only data it need read are the actual allocated data blocks.
7464 @GNUTAR{} uses a more portable and straightforward
7465 archiving approach, it would be fairly difficult that it does
7466 otherwise. Elizabeth Zwicky writes to @file{comp.unix.internals}, on
7470 What I did say is that you cannot tell the difference between a hole and an
7471 equivalent number of nulls without reading raw blocks. @code{st_blocks} at
7472 best tells you how many holes there are; it doesn't tell you @emph{where}.
7473 Just as programs may, conceivably, care what @code{st_blocks} is (care
7474 to name one that does?), they may also care where the holes are (I have
7475 no examples of this one either, but it's equally imaginable).
7477 I conclude from this that good archivers are not portable. One can
7478 arguably conclude that if you want a portable program, you can in good
7479 conscience restore files with as many holes as possible, since you can't
7484 @section Handling File Attributes
7487 When @command{tar} reads files, it updates their access times. To
7488 avoid this, use the @option{--atime-preserve[=METHOD]} option, which can either
7489 reset the access time retroactively or avoid changing it in the first
7492 Handling of file attributes
7495 @opindex atime-preserve
7496 @item --atime-preserve
7497 @itemx --atime-preserve=replace
7498 @itemx --atime-preserve=system
7499 Preserve the access times of files that are read. This works only for
7500 files that you own, unless you have superuser privileges.
7502 @option{--atime-preserve=replace} works on most systems, but it also
7503 restores the data modification time and updates the status change
7504 time. Hence it doesn't interact with incremental dumps nicely
7505 (@pxref{Backups}), and it can set access or data modification times
7506 incorrectly if other programs access the file while @command{tar} is
7509 @option{--atime-preserve=system} avoids changing the access time in
7510 the first place, if the operating system supports this.
7511 Unfortunately, this may or may not work on any given operating system
7512 or file system. If @command{tar} knows for sure it won't work, it
7513 complains right away.
7515 Currently @option{--atime-preserve} with no operand defaults to
7516 @option{--atime-preserve=replace}, but this is intended to change to
7517 @option{--atime-preserve=system} when the latter is better-supported.
7522 Do not extract data modification time.
7524 When this option is used, @command{tar} leaves the data modification times
7525 of the files it extracts as the times when the files were extracted,
7526 instead of setting it to the times recorded in the archive.
7528 This option is meaningless with @option{--list} (@option{-t}).
7532 Create extracted files with the same ownership they have in the
7535 This is the default behavior for the superuser,
7536 so this option is meaningful only for non-root users, when @command{tar}
7537 is executed on those systems able to give files away. This is
7538 considered as a security flaw by many people, at least because it
7539 makes quite difficult to correctly account users for the disk space
7540 they occupy. Also, the @code{suid} or @code{sgid} attributes of
7541 files are easily and silently lost when files are given away.
7543 When writing an archive, @command{tar} writes the user id and user name
7544 separately. If it can't find a user name (because the user id is not
7545 in @file{/etc/passwd}), then it does not write one. When restoring,
7546 and doing a @code{chmod} like when you use @option{--same-permissions},
7547 @FIXME{same-owner?}it tries to look the name (if one was written)
7548 up in @file{/etc/passwd}. If it fails, then it uses the user id
7549 stored in the archive instead.
7551 @opindex no-same-owner
7552 @item --no-same-owner
7554 Do not attempt to restore ownership when extracting. This is the
7555 default behavior for ordinary users, so this option has an effect
7556 only for the superuser.
7558 @opindex numeric-owner
7559 @item --numeric-owner
7560 The @option{--numeric-owner} option allows (ANSI) archives to be written
7561 without user/group name information or such information to be ignored
7562 when extracting. It effectively disables the generation and/or use
7563 of user/group name information. This option forces extraction using
7564 the numeric ids from the archive, ignoring the names.
7566 This is useful in certain circumstances, when restoring a backup from
7567 an emergency floppy with different passwd/group files for example.
7568 It is otherwise impossible to extract files with the right ownerships
7569 if the password file in use during the extraction does not match the
7570 one belonging to the file system(s) being extracted. This occurs,
7571 for example, if you are restoring your files after a major crash and
7572 had booted from an emergency floppy with no password file or put your
7573 disk into another machine to do the restore.
7575 The numeric ids are @emph{always} saved into @command{tar} archives.
7576 The identifying names are added at create time when provided by the
7577 system, unless @option{--old-archive} (@option{-o}) is used. Numeric ids could be
7578 used when moving archives between a collection of machines using
7579 a centralized management for attribution of numeric ids to users
7580 and groups. This is often made through using the NIS capabilities.
7582 When making a @command{tar} file for distribution to other sites, it
7583 is sometimes cleaner to use a single owner for all files in the
7584 distribution, and nicer to specify the write permission bits of the
7585 files as stored in the archive independently of their actual value on
7586 the file system. The way to prepare a clean distribution is usually
7587 to have some Makefile rule creating a directory, copying all needed
7588 files in that directory, then setting ownership and permissions as
7589 wanted (there are a lot of possible schemes), and only then making a
7590 @command{tar} archive out of this directory, before cleaning
7591 everything out. Of course, we could add a lot of options to
7592 @GNUTAR{} for fine tuning permissions and ownership.
7593 This is not the good way, I think. @GNUTAR{} is
7594 already crowded with options and moreover, the approach just explained
7595 gives you a great deal of control already.
7597 @opindex same-permissions, short description
7598 @opindex preserve-permissions, short description
7600 @itemx --same-permissions
7601 @itemx --preserve-permissions
7602 Extract all protection information.
7604 This option causes @command{tar} to set the modes (access permissions) of
7605 extracted files exactly as recorded in the archive. If this option
7606 is not used, the current @code{umask} setting limits the permissions
7607 on extracted files. This option is by default enabled when
7608 @command{tar} is executed by a superuser.
7611 This option is meaningless with @option{--list} (@option{-t}).
7615 Same as both @option{--same-permissions} and @option{--same-order}.
7617 The @option{--preserve} option has no equivalent short option name.
7618 It is equivalent to @option{--same-permissions} plus @option{--same-order}.
7620 @FIXME{I do not see the purpose of such an option. (Neither I. FP.)}
7625 @section Basic Tar Format
7628 While an archive may contain many files, the archive itself is a
7629 single ordinary file. Like any other file, an archive file can be
7630 written to a storage device such as a tape or disk, sent through a
7631 pipe or over a network, saved on the active file system, or even
7632 stored in another archive. An archive file is not easy to read or
7633 manipulate without using the @command{tar} utility or Tar mode in
7634 @acronym{GNU} Emacs.
7636 Physically, an archive consists of a series of file entries terminated
7637 by an end-of-archive entry, which consists of two 512 blocks of zero
7639 entry usually describes one of the files in the archive (an
7640 @dfn{archive member}), and consists of a file header and the contents
7641 of the file. File headers contain file names and statistics, checksum
7642 information which @command{tar} uses to detect file corruption, and
7643 information about file types.
7645 Archives are permitted to have more than one member with the same
7646 member name. One way this situation can occur is if more than one
7647 version of a file has been stored in the archive. For information
7648 about adding new versions of a file to an archive, see @ref{update}.
7649 @FIXME-xref{To learn more about having more than one archive member with the
7650 same name, see -backup node, when it's written.}
7652 In addition to entries describing archive members, an archive may
7653 contain entries which @command{tar} itself uses to store information.
7654 @xref{label}, for an example of such an archive entry.
7656 A @command{tar} archive file contains a series of blocks. Each block
7657 contains @code{BLOCKSIZE} bytes. Although this format may be thought
7658 of as being on magnetic tape, other media are often used.
7660 Each file archived is represented by a header block which describes
7661 the file, followed by zero or more blocks which give the contents
7662 of the file. At the end of the archive file there are two 512-byte blocks
7663 filled with binary zeros as an end-of-file marker. A reasonable system
7664 should write such end-of-file marker at the end of an archive, but
7665 must not assume that such a block exists when reading an archive. In
7666 particular @GNUTAR{} always issues a warning if it does not encounter it.
7668 The blocks may be @dfn{blocked} for physical I/O operations.
7669 Each record of @var{n} blocks (where @var{n} is set by the
7670 @option{--blocking-factor=@var{512-size}} (@option{-b @var{512-size}}) option to @command{tar}) is written with a single
7671 @w{@samp{write ()}} operation. On magnetic tapes, the result of
7672 such a write is a single record. When writing an archive,
7673 the last record of blocks should be written at the full size, with
7674 blocks after the zero block containing all zeros. When reading
7675 an archive, a reasonable system should properly handle an archive
7676 whose last record is shorter than the rest, or which contains garbage
7677 records after a zero block.
7679 The header block is defined in C as follows. In the @GNUTAR{}
7680 distribution, this is part of file @file{src/tar.h}:
7683 @include header.texi
7686 All characters in header blocks are represented by using 8-bit
7687 characters in the local variant of ASCII. Each field within the
7688 structure is contiguous; that is, there is no padding used within
7689 the structure. Each character on the archive medium is stored
7692 Bytes representing the contents of files (after the header block
7693 of each file) are not translated in any way and are not constrained
7694 to represent characters in any character set. The @command{tar} format
7695 does not distinguish text files from binary files, and no translation
7696 of file contents is performed.
7698 The @code{name}, @code{linkname}, @code{magic}, @code{uname}, and
7699 @code{gname} are null-terminated character strings. All other fields
7700 are zero-filled octal numbers in ASCII. Each numeric field of width
7701 @var{w} contains @var{w} minus 1 digits, and a null.
7703 The @code{name} field is the file name of the file, with directory names
7704 (if any) preceding the file name, separated by slashes.
7706 @FIXME{how big a name before field overflows?}
7708 The @code{mode} field provides nine bits specifying file permissions
7709 and three bits to specify the Set UID, Set GID, and Save Text
7710 (@dfn{sticky}) modes. Values for these bits are defined above.
7711 When special permissions are required to create a file with a given
7712 mode, and the user restoring files from the archive does not hold such
7713 permissions, the mode bit(s) specifying those special permissions
7714 are ignored. Modes which are not supported by the operating system
7715 restoring files from the archive will be ignored. Unsupported modes
7716 should be faked up when creating or updating an archive; e.g., the
7717 group permission could be copied from the @emph{other} permission.
7719 The @code{uid} and @code{gid} fields are the numeric user and group
7720 ID of the file owners, respectively. If the operating system does
7721 not support numeric user or group IDs, these fields should be ignored.
7723 The @code{size} field is the size of the file in bytes; linked files
7724 are archived with this field specified as zero. @FIXME-xref{Modifiers, in
7725 particular the @option{--incremental} (@option{-G}) option.}
7727 The @code{mtime} field is the data modification time of the file at
7728 the time it was archived. It is the ASCII representation of the octal
7729 value of the last time the file's contents were modified, represented
7730 as an integer number of
7731 seconds since January 1, 1970, 00:00 Coordinated Universal Time.
7733 The @code{chksum} field is the ASCII representation of the octal value
7734 of the simple sum of all bytes in the header block. Each 8-bit
7735 byte in the header is added to an unsigned integer, initialized to
7736 zero, the precision of which shall be no less than seventeen bits.
7737 When calculating the checksum, the @code{chksum} field is treated as
7738 if it were all blanks.
7740 The @code{typeflag} field specifies the type of file archived. If a
7741 particular implementation does not recognize or permit the specified
7742 type, the file will be extracted as if it were a regular file. As this
7743 action occurs, @command{tar} issues a warning to the standard error.
7745 The @code{atime} and @code{ctime} fields are used in making incremental
7746 backups; they store, respectively, the particular file's access and
7747 status change times.
7749 The @code{offset} is used by the @option{--multi-volume} (@option{-M}) option, when
7750 making a multi-volume archive. The offset is number of bytes into
7751 the file that we need to restart at to continue the file on the next
7752 tape, i.e., where we store the location that a continued file is
7755 The following fields were added to deal with sparse files. A file
7756 is @dfn{sparse} if it takes in unallocated blocks which end up being
7757 represented as zeros, i.e., no useful data. A test to see if a file
7758 is sparse is to look at the number blocks allocated for it versus the
7759 number of characters in the file; if there are fewer blocks allocated
7760 for the file than would normally be allocated for a file of that
7761 size, then the file is sparse. This is the method @command{tar} uses to
7762 detect a sparse file, and once such a file is detected, it is treated
7763 differently from non-sparse files.
7765 Sparse files are often @code{dbm} files, or other database-type files
7766 which have data at some points and emptiness in the greater part of
7767 the file. Such files can appear to be very large when an @samp{ls
7768 -l} is done on them, when in truth, there may be a very small amount
7769 of important data contained in the file. It is thus undesirable
7770 to have @command{tar} think that it must back up this entire file, as
7771 great quantities of room are wasted on empty blocks, which can lead
7772 to running out of room on a tape far earlier than is necessary.
7773 Thus, sparse files are dealt with so that these empty blocks are
7774 not written to the tape. Instead, what is written to the tape is a
7775 description, of sorts, of the sparse file: where the holes are, how
7776 big the holes are, and how much data is found at the end of the hole.
7777 This way, the file takes up potentially far less room on the tape,
7778 and when the file is extracted later on, it will look exactly the way
7779 it looked beforehand. The following is a description of the fields
7780 used to handle a sparse file:
7782 The @code{sp} is an array of @code{struct sparse}. Each @code{struct
7783 sparse} contains two 12-character strings which represent an offset
7784 into the file and a number of bytes to be written at that offset.
7785 The offset is absolute, and not relative to the offset in preceding
7788 The header can hold four of these @code{struct sparse} at the moment;
7789 if more are needed, they are not stored in the header.
7791 The @code{isextended} flag is set when an @code{extended_header}
7792 is needed to deal with a file. Note that this means that this flag
7793 can only be set when dealing with a sparse file, and it is only set
7794 in the event that the description of the file will not fit in the
7795 allotted room for sparse structures in the header. In other words,
7796 an extended_header is needed.
7798 The @code{extended_header} structure is used for sparse files which
7799 need more sparse structures than can fit in the header. The header can
7800 fit 4 such structures; if more are needed, the flag @code{isextended}
7801 gets set and the next block is an @code{extended_header}.
7803 Each @code{extended_header} structure contains an array of 21
7804 sparse structures, along with a similar @code{isextended} flag
7805 that the header had. There can be an indeterminate number of such
7806 @code{extended_header}s to describe a sparse file.
7810 @item @code{REGTYPE}
7811 @itemx @code{AREGTYPE}
7812 These flags represent a regular file. In order to be compatible
7813 with older versions of @command{tar}, a @code{typeflag} value of
7814 @code{AREGTYPE} should be silently recognized as a regular file.
7815 New archives should be created using @code{REGTYPE}. Also, for
7816 backward compatibility, @command{tar} treats a regular file whose name
7817 ends with a slash as a directory.
7819 @item @code{LNKTYPE}
7820 This flag represents a file linked to another file, of any type,
7821 previously archived. Such files are identified in Unix by each
7822 file having the same device and inode number. The linked-to name is
7823 specified in the @code{linkname} field with a trailing null.
7825 @item @code{SYMTYPE}
7826 This represents a symbolic link to another file. The linked-to name
7827 is specified in the @code{linkname} field with a trailing null.
7829 @item @code{CHRTYPE}
7830 @itemx @code{BLKTYPE}
7831 These represent character special files and block special files
7832 respectively. In this case the @code{devmajor} and @code{devminor}
7833 fields will contain the major and minor device numbers respectively.
7834 Operating systems may map the device specifications to their own
7835 local specification, or may ignore the entry.
7837 @item @code{DIRTYPE}
7838 This flag specifies a directory or sub-directory. The directory
7839 name in the @code{name} field should end with a slash. On systems where
7840 disk allocation is performed on a directory basis, the @code{size} field
7841 will contain the maximum number of bytes (which may be rounded to
7842 the nearest disk block allocation unit) which the directory may
7843 hold. A @code{size} field of zero indicates no such limiting. Systems
7844 which do not support limiting in this manner should ignore the
7847 @item @code{FIFOTYPE}
7848 This specifies a FIFO special file. Note that the archiving of a
7849 FIFO file archives the existence of this file and not its contents.
7851 @item @code{CONTTYPE}
7852 This specifies a contiguous file, which is the same as a normal
7853 file except that, in operating systems which support it, all its
7854 space is allocated contiguously on the disk. Operating systems
7855 which do not allow contiguous allocation should silently treat this
7856 type as a normal file.
7858 @item @code{A} @dots{} @code{Z}
7859 These are reserved for custom implementations. Some of these are
7860 used in the @acronym{GNU} modified format, as described below.
7864 Other values are reserved for specification in future revisions of
7865 the P1003 standard, and should not be used by any @command{tar} program.
7867 The @code{magic} field indicates that this archive was output in
7868 the P1003 archive format. If this field contains @code{TMAGIC},
7869 the @code{uname} and @code{gname} fields will contain the ASCII
7870 representation of the owner and group of the file respectively.
7871 If found, the user and group IDs are used rather than the values in
7872 the @code{uid} and @code{gid} fields.
7874 For references, see ISO/IEC 9945-1:1990 or IEEE Std 1003.1-1990, pages
7875 169-173 (section 10.1) for @cite{Archive/Interchange File Format}; and
7876 IEEE Std 1003.2-1992, pages 380-388 (section 4.48) and pages 936-940
7877 (section E.4.48) for @cite{pax - Portable archive interchange}.
7880 @section @acronym{GNU} Extensions to the Archive Format
7883 The @acronym{GNU} format uses additional file types to describe new types of
7884 files in an archive. These are listed below.
7887 @item GNUTYPE_DUMPDIR
7889 This represents a directory and a list of files created by the
7890 @option{--incremental} (@option{-G}) option. The @code{size} field gives the total
7891 size of the associated list of files. Each file name is preceded by
7892 either a @samp{Y} (the file should be in this archive) or an @samp{N}.
7893 (The file is a directory, or is not stored in the archive.) Each file
7894 name is terminated by a null. There is an additional null after the
7897 @item GNUTYPE_MULTIVOL
7899 This represents a file continued from another volume of a multi-volume
7900 archive created with the @option{--multi-volume} (@option{-M}) option. The original
7901 type of the file is not given here. The @code{size} field gives the
7902 maximum size of this piece of the file (assuming the volume does
7903 not end before the file is written out). The @code{offset} field
7904 gives the offset from the beginning of the file where this part of
7905 the file begins. Thus @code{size} plus @code{offset} should equal
7906 the original size of the file.
7908 @item GNUTYPE_SPARSE
7910 This flag indicates that we are dealing with a sparse file. Note
7911 that archiving a sparse file requires special operations to find
7912 holes in the file, which mark the positions of these holes, along
7913 with the number of bytes of data to be found after the hole.
7915 @item GNUTYPE_VOLHDR
7917 This file type is used to mark the volume header that was given with
7918 the @option{--label=@var{archive-label}} (@option{-V @var{archive-label}}) option when the archive was created. The @code{name}
7919 field contains the @code{name} given after the @option{--label=@var{archive-label}} (@option{-V @var{archive-label}}) option.
7920 The @code{size} field is zero. Only the first file in each volume
7921 of an archive should have this type.
7925 You may have trouble reading a @acronym{GNU} format archive on a
7926 non-@acronym{GNU} system if the options @option{--incremental} (@option{-G}),
7927 @option{--multi-volume} (@option{-M}), @option{--sparse} (@option{-S}), or @option{--label=@var{archive-label}} (@option{-V @var{archive-label}}) were
7928 used when writing the archive. In general, if @command{tar} does not
7929 use the @acronym{GNU}-added fields of the header, other versions of
7930 @command{tar} should be able to read the archive. Otherwise, the
7931 @command{tar} program will give an error, the most likely one being a
7935 @section Comparison of @command{tar} and @command{cpio}
7938 @FIXME{Reorganize the following material}
7940 The @command{cpio} archive formats, like @command{tar}, do have maximum
7941 pathname lengths. The binary and old ASCII formats have a max path
7942 length of 256, and the new ASCII and CRC ASCII formats have a max
7943 path length of 1024. @acronym{GNU} @command{cpio} can read and write archives
7944 with arbitrary pathname lengths, but other @command{cpio} implementations
7945 may crash unexplainedly trying to read them.
7947 @command{tar} handles symbolic links in the form in which it comes in BSD;
7948 @command{cpio} doesn't handle symbolic links in the form in which it comes
7949 in System V prior to SVR4, and some vendors may have added symlinks
7950 to their system without enhancing @command{cpio} to know about them.
7951 Others may have enhanced it in a way other than the way I did it
7952 at Sun, and which was adopted by AT&T (and which is, I think, also
7953 present in the @command{cpio} that Berkeley picked up from AT&T and put
7954 into a later BSD release---I think I gave them my changes).
7956 (SVR4 does some funny stuff with @command{tar}; basically, its @command{cpio}
7957 can handle @command{tar} format input, and write it on output, and it
7958 probably handles symbolic links. They may not have bothered doing
7959 anything to enhance @command{tar} as a result.)
7961 @command{cpio} handles special files; traditional @command{tar} doesn't.
7963 @command{tar} comes with V7, System III, System V, and BSD source;
7964 @command{cpio} comes only with System III, System V, and later BSD
7965 (4.3-tahoe and later).
7967 @command{tar}'s way of handling multiple hard links to a file can handle
7968 file systems that support 32-bit inumbers (e.g., the BSD file system);
7969 @command{cpio}s way requires you to play some games (in its "binary"
7970 format, i-numbers are only 16 bits, and in its "portable ASCII" format,
7971 they're 18 bits---it would have to play games with the "file system ID"
7972 field of the header to make sure that the file system ID/i-number pairs
7973 of different files were always different), and I don't know which
7974 @command{cpio}s, if any, play those games. Those that don't might get
7975 confused and think two files are the same file when they're not, and
7976 make hard links between them.
7978 @command{tar}s way of handling multiple hard links to a file places only
7979 one copy of the link on the tape, but the name attached to that copy
7980 is the @emph{only} one you can use to retrieve the file; @command{cpio}s
7981 way puts one copy for every link, but you can retrieve it using any
7985 What type of check sum (if any) is used, and how is this calculated.
7988 See the attached manual pages for @command{tar} and @command{cpio} format.
7989 @command{tar} uses a checksum which is the sum of all the bytes in the
7990 @command{tar} header for a file; @command{cpio} uses no checksum.
7993 If anyone knows why @command{cpio} was made when @command{tar} was present
7997 It wasn't. @command{cpio} first showed up in PWB/UNIX 1.0; no
7998 generally-available version of UNIX had @command{tar} at the time. I don't
7999 know whether any version that was generally available @emph{within AT&T}
8000 had @command{tar}, or, if so, whether the people within AT&T who did
8001 @command{cpio} knew about it.
8003 On restore, if there is a corruption on a tape @command{tar} will stop at
8004 that point, while @command{cpio} will skip over it and try to restore the
8007 The main difference is just in the command syntax and header format.
8009 @command{tar} is a little more tape-oriented in that everything is blocked
8010 to start on a record boundary.
8013 Is there any differences between the ability to recover crashed
8014 archives between the two of them. (Is there any chance of recovering
8015 crashed archives at all.)
8018 Theoretically it should be easier under @command{tar} since the blocking
8019 lets you find a header with some variation of @samp{dd skip=@var{nn}}.
8020 However, modern @command{cpio}'s and variations have an option to just
8021 search for the next file header after an error with a reasonable chance
8022 of resyncing. However, lots of tape driver software won't allow you to
8023 continue past a media error which should be the only reason for getting
8024 out of sync unless a file changed sizes while you were writing the
8028 If anyone knows why @command{cpio} was made when @command{tar} was present
8029 at the unix scene, please tell me about this too.
8032 Probably because it is more media efficient (by not blocking everything
8033 and using only the space needed for the headers where @command{tar}
8034 always uses 512 bytes per file header) and it knows how to archive
8037 You might want to look at the freely available alternatives. The
8038 major ones are @command{afio}, @GNUTAR{}, and
8039 @command{pax}, each of which have their own extensions with some
8040 backwards compatibility.
8042 Sparse files were @command{tar}red as sparse files (which you can
8043 easily test, because the resulting archive gets smaller, and
8044 @acronym{GNU} @command{cpio} can no longer read it).
8047 @chapter Tapes and Other Archive Media
8050 A few special cases about tape handling warrant more detailed
8051 description. These special cases are discussed below.
8053 Many complexities surround the use of @command{tar} on tape drives. Since
8054 the creation and manipulation of archives located on magnetic tape was
8055 the original purpose of @command{tar}, it contains many features making
8056 such manipulation easier.
8058 Archives are usually written on dismountable media---tape cartridges,
8059 mag tapes, or floppy disks.
8061 The amount of data a tape or disk holds depends not only on its size,
8062 but also on how it is formatted. A 2400 foot long reel of mag tape
8063 holds 40 megabytes of data when formatted at 1600 bits per inch. The
8064 physically smaller EXABYTE tape cartridge holds 2.3 gigabytes.
8066 Magnetic media are re-usable---once the archive on a tape is no longer
8067 needed, the archive can be erased and the tape or disk used over.
8068 Media quality does deteriorate with use, however. Most tapes or disks
8069 should be discarded when they begin to produce data errors. EXABYTE
8070 tape cartridges should be discarded when they generate an @dfn{error
8071 count} (number of non-usable bits) of more than 10k.
8073 Magnetic media are written and erased using magnetic fields, and
8074 should be protected from such fields to avoid damage to stored data.
8075 Sticking a floppy disk to a filing cabinet using a magnet is probably
8079 * Device:: Device selection and switching
8080 * Remote Tape Server::
8081 * Common Problems and Solutions::
8082 * Blocking:: Blocking
8083 * Many:: Many archives on one tape
8084 * Using Multiple Tapes:: Using Multiple Tapes
8085 * label:: Including a Label in the Archive
8087 * Write Protection::
8091 @section Device Selection and Switching
8095 @item -f [@var{hostname}:]@var{file}
8096 @itemx --file=[@var{hostname}:]@var{file}
8097 Use archive file or device @var{file} on @var{hostname}.
8100 This option is used to specify the file name of the archive @command{tar}
8103 If the file name is @samp{-}, @command{tar} reads the archive from standard
8104 input (when listing or extracting), or writes it to standard output
8105 (when creating). If the @samp{-} file name is given when updating an
8106 archive, @command{tar} will read the original archive from its standard
8107 input, and will write the entire new archive to its standard output.
8109 If the file name contains a @samp{:}, it is interpreted as
8110 @samp{hostname:file name}. If the @var{hostname} contains an @dfn{at}
8111 sign (@samp{@@}), it is treated as @samp{user@@hostname:file name}. In
8112 either case, @command{tar} will invoke the command @command{rsh} (or
8113 @command{remsh}) to start up an @command{/usr/libexec/rmt} on the remote
8114 machine. If you give an alternate login name, it will be given to the
8116 Naturally, the remote machine must have an executable
8117 @command{/usr/libexec/rmt}. This program is free software from the
8118 University of California, and a copy of the source code can be found
8119 with the sources for @command{tar}; it's compiled and installed by default.
8120 The exact path to this utility is determined when configuring the package.
8121 It is @file{@var{prefix}/libexec/rmt}, where @var{prefix} stands for
8122 your installation prefix. This location may also be overridden at
8123 runtime by using @option{rmt-command=@var{command}} option (@xref{Option Summary,
8124 ---rmt-command}, for detailed description of this option. @xref{Remote
8125 Tape Server}, for the description of @command{rmt} command).
8127 If this option is not given, but the environment variable @env{TAPE}
8128 is set, its value is used; otherwise, old versions of @command{tar}
8129 used a default archive name (which was picked when @command{tar} was
8130 compiled). The default is normally set up to be the @dfn{first} tape
8131 drive or other transportable I/O medium on the system.
8133 Starting with version 1.11.5, @GNUTAR{} uses
8134 standard input and standard output as the default device, and I will
8135 not try anymore supporting automatic device detection at installation
8136 time. This was failing really in too many cases, it was hopeless.
8137 This is now completely left to the installer to override standard
8138 input and standard output for default device, if this seems
8139 preferable. Further, I think @emph{most} actual usages of
8140 @command{tar} are done with pipes or disks, not really tapes,
8141 cartridges or diskettes.
8143 Some users think that using standard input and output is running
8144 after trouble. This could lead to a nasty surprise on your screen if
8145 you forget to specify an output file name---especially if you are going
8146 through a network or terminal server capable of buffering large amounts
8147 of output. We had so many bug reports in that area of configuring
8148 default tapes automatically, and so many contradicting requests, that
8149 we finally consider the problem to be portably intractable. We could
8150 of course use something like @samp{/dev/tape} as a default, but this
8151 is @emph{also} running after various kind of trouble, going from hung
8152 processes to accidental destruction of real tapes. After having seen
8153 all this mess, using standard input and output as a default really
8154 sounds like the only clean choice left, and a very useful one too.
8156 @GNUTAR{} reads and writes archive in records, I
8157 suspect this is the main reason why block devices are preferred over
8158 character devices. Most probably, block devices are more efficient
8159 too. The installer could also check for @samp{DEFTAPE} in
8160 @file{<sys/mtio.h>}.
8163 @opindex force-local, short description
8165 Archive file is local even if it contains a colon.
8167 @opindex rsh-command
8168 @item --rsh-command=@var{command}
8169 Use remote @var{command} instead of @command{rsh}. This option exists
8170 so that people who use something other than the standard @command{rsh}
8171 (e.g., a Kerberized @command{rsh}) can access a remote device.
8173 When this command is not used, the shell command found when
8174 the @command{tar} program was installed is used instead. This is
8175 the first found of @file{/usr/ucb/rsh}, @file{/usr/bin/remsh},
8176 @file{/usr/bin/rsh}, @file{/usr/bsd/rsh} or @file{/usr/bin/nsh}.
8177 The installer may have overridden this by defining the environment
8178 variable @env{RSH} @emph{at installation time}.
8181 Specify drive and density.
8183 @opindex multi-volume, short description
8185 @itemx --multi-volume
8186 Create/list/extract multi-volume archive.
8188 This option causes @command{tar} to write a @dfn{multi-volume} archive---one
8189 that may be larger than will fit on the medium used to hold it.
8190 @xref{Multi-Volume Archives}.
8192 @opindex tape-length, short description
8194 @itemx --tape-length=@var{num}
8195 Change tape after writing @var{num} x 1024 bytes.
8197 This option might be useful when your tape drivers do not properly
8198 detect end of physical tapes. By being slightly conservative on the
8199 maximum tape length, you might avoid the problem entirely.
8201 @opindex info-script, short description
8202 @opindex new-volume-script, short description
8204 @itemx --info-script=@var{file}
8205 @itemx --new-volume-script=@var{file}
8206 Execute @file{file} at end of each tape. This implies
8207 @option{--multi-volume} (@option{-M}). @xref{info-script}, for a detailed
8208 description of this option.
8211 @node Remote Tape Server
8212 @section The Remote Tape Server
8214 @cindex remote tape drive
8216 In order to access the tape drive on a remote machine, @command{tar}
8217 uses the remote tape server written at the University of California at
8218 Berkeley. The remote tape server must be installed as
8219 @file{@var{prefix}/libexec/rmt} on any machine whose tape drive you
8220 want to use. @command{tar} calls @command{rmt} by running an
8221 @command{rsh} or @command{remsh} to the remote machine, optionally
8222 using a different login name if one is supplied.
8224 A copy of the source for the remote tape server is provided. It is
8225 Copyright @copyright{} 1983 by the Regents of the University of
8226 California, but can be freely distributed. It is compiled and
8227 installed by default.
8229 @cindex absolute file names
8230 Unless you use the @option{--absolute-names} (@option{-P}) option,
8231 @GNUTAR{} will not allow you to create an archive that contains
8232 absolute file names (a file name beginning with @samp{/}.) If you try,
8233 @command{tar} will automatically remove the leading @samp{/} from the
8234 file names it stores in the archive. It will also type a warning
8235 message telling you what it is doing.
8237 When reading an archive that was created with a different
8238 @command{tar} program, @GNUTAR{} automatically
8239 extracts entries in the archive which have absolute file names as if
8240 the file names were not absolute. This is an important feature. A
8241 visitor here once gave a @command{tar} tape to an operator to restore;
8242 the operator used Sun @command{tar} instead of @GNUTAR{},
8243 and the result was that it replaced large portions of
8244 our @file{/bin} and friends with versions from the tape; needless to
8245 say, we were unhappy about having to recover the file system from
8248 For example, if the archive contained a file @file{/usr/bin/computoy},
8249 @GNUTAR{} would extract the file to @file{usr/bin/computoy},
8250 relative to the current directory. If you want to extract the files in
8251 an archive to the same absolute names that they had when the archive
8252 was created, you should do a @samp{cd /} before extracting the files
8253 from the archive, or you should either use the @option{--absolute-names}
8254 option, or use the command @samp{tar -C / @dots{}}.
8256 @cindex Ultrix 3.1 and write failure
8257 Some versions of Unix (Ultrix 3.1 is known to have this problem),
8258 can claim that a short write near the end of a tape succeeded,
8259 when it actually failed. This will result in the -M option not
8260 working correctly. The best workaround at the moment is to use a
8261 significantly larger blocking factor than the default 20.
8263 In order to update an archive, @command{tar} must be able to backspace the
8264 archive in order to reread or rewrite a record that was just read (or
8265 written). This is currently possible only on two kinds of files: normal
8266 disk files (or any other file that can be backspaced with @samp{lseek}),
8267 and industry-standard 9-track magnetic tape (or any other kind of tape
8268 that can be backspaced with the @code{MTIOCTOP} @code{ioctl}.
8270 This means that the @option{--append}, @option{--concatenate}, and
8271 @option{--delete} commands will not work on any other kind of file.
8272 Some media simply cannot be backspaced, which means these commands and
8273 options will never be able to work on them. These non-backspacing
8274 media include pipes and cartridge tape drives.
8276 Some other media can be backspaced, and @command{tar} will work on them
8277 once @command{tar} is modified to do so.
8279 Archives created with the @option{--multi-volume}, @option{--label}, and
8280 @option{--incremental} (@option{-G}) options may not be readable by other version
8281 of @command{tar}. In particular, restoring a file that was split over
8282 a volume boundary will require some careful work with @command{dd}, if
8283 it can be done at all. Other versions of @command{tar} may also create
8284 an empty file whose name is that of the volume header. Some versions
8285 of @command{tar} may create normal files instead of directories archived
8286 with the @option{--incremental} (@option{-G}) option.
8288 @node Common Problems and Solutions
8289 @section Some Common Problems and their Solutions
8296 no such file or directory
8299 errors from @command{tar}:
8300 directory checksum error
8303 errors from media/system:
8314 @dfn{Block} and @dfn{record} terminology is rather confused, and it
8315 is also confusing to the expert reader. On the other hand, readers
8316 who are new to the field have a fresh mind, and they may safely skip
8317 the next two paragraphs, as the remainder of this manual uses those
8318 two terms in a quite consistent way.
8320 John Gilmore, the writer of the public domain @command{tar} from which
8321 @GNUTAR{} was originally derived, wrote (June 1995):
8324 The nomenclature of tape drives comes from IBM, where I believe
8325 they were invented for the IBM 650 or so. On IBM mainframes, what
8326 is recorded on tape are tape blocks. The logical organization of
8327 data is into records. There are various ways of putting records into
8328 blocks, including @code{F} (fixed sized records), @code{V} (variable
8329 sized records), @code{FB} (fixed blocked: fixed size records, @var{n}
8330 to a block), @code{VB} (variable size records, @var{n} to a block),
8331 @code{VSB} (variable spanned blocked: variable sized records that can
8332 occupy more than one block), etc. The @code{JCL} @samp{DD RECFORM=}
8333 parameter specified this to the operating system.
8335 The Unix man page on @command{tar} was totally confused about this.
8336 When I wrote @code{PD TAR}, I used the historically correct terminology
8337 (@command{tar} writes data records, which are grouped into blocks).
8338 It appears that the bogus terminology made it into @acronym{POSIX} (no surprise
8339 here), and now Fran@,{c}ois has migrated that terminology back
8340 into the source code too.
8343 The term @dfn{physical block} means the basic transfer chunk from or
8344 to a device, after which reading or writing may stop without anything
8345 being lost. In this manual, the term @dfn{block} usually refers to
8346 a disk physical block, @emph{assuming} that each disk block is 512
8347 bytes in length. It is true that some disk devices have different
8348 physical blocks, but @command{tar} ignore these differences in its own
8349 format, which is meant to be portable, so a @command{tar} block is always
8350 512 bytes in length, and @dfn{block} always mean a @command{tar} block.
8351 The term @dfn{logical block} often represents the basic chunk of
8352 allocation of many disk blocks as a single entity, which the operating
8353 system treats somewhat atomically; this concept is only barely used
8356 The term @dfn{physical record} is another way to speak of a physical
8357 block, those two terms are somewhat interchangeable. In this manual,
8358 the term @dfn{record} usually refers to a tape physical block,
8359 @emph{assuming} that the @command{tar} archive is kept on magnetic tape.
8360 It is true that archives may be put on disk or used with pipes,
8361 but nevertheless, @command{tar} tries to read and write the archive one
8362 @dfn{record} at a time, whatever the medium in use. One record is made
8363 up of an integral number of blocks, and this operation of putting many
8364 disk blocks into a single tape block is called @dfn{reblocking}, or
8365 more simply, @dfn{blocking}. The term @dfn{logical record} refers to
8366 the logical organization of many characters into something meaningful
8367 to the application. The term @dfn{unit record} describes a small set
8368 of characters which are transmitted whole to or by the application,
8369 and often refers to a line of text. Those two last terms are unrelated
8370 to what we call a @dfn{record} in @GNUTAR{}.
8372 When writing to tapes, @command{tar} writes the contents of the archive
8373 in chunks known as @dfn{records}. To change the default blocking
8374 factor, use the @option{--blocking-factor=@var{512-size}} (@option{-b
8375 @var{512-size}}) option. Each record will then be composed of
8376 @var{512-size} blocks. (Each @command{tar} block is 512 bytes.
8377 @xref{Standard}.) Each file written to the archive uses at least one
8378 full record. As a result, using a larger record size can result in
8379 more wasted space for small files. On the other hand, a larger record
8380 size can often be read and written much more efficiently.
8382 Further complicating the problem is that some tape drives ignore the
8383 blocking entirely. For these, a larger record size can still improve
8384 performance (because the software layers above the tape drive still
8385 honor the blocking), but not as dramatically as on tape drives that
8388 When reading an archive, @command{tar} can usually figure out the
8389 record size on itself. When this is the case, and a non-standard
8390 record size was used when the archive was created, @command{tar} will
8391 print a message about a non-standard blocking factor, and then operate
8392 normally. On some tape devices, however, @command{tar} cannot figure
8393 out the record size itself. On most of those, you can specify a
8394 blocking factor (with @option{--blocking-factor}) larger than the
8395 actual blocking factor, and then use the @option{--read-full-records}
8396 (@option{-B}) option. (If you specify a blocking factor with
8397 @option{--blocking-factor} and don't use the
8398 @option{--read-full-records} option, then @command{tar} will not
8399 attempt to figure out the recording size itself.) On some devices,
8400 you must always specify the record size exactly with
8401 @option{--blocking-factor} when reading, because @command{tar} cannot
8402 figure it out. In any case, use @option{--list} (@option{-t}) before
8403 doing any extractions to see whether @command{tar} is reading the archive
8406 @command{tar} blocks are all fixed size (512 bytes), and its scheme for
8407 putting them into records is to put a whole number of them (one or
8408 more) into each record. @command{tar} records are all the same size;
8409 at the end of the file there's a block containing all zeros, which
8410 is how you tell that the remainder of the last record(s) are garbage.
8412 In a standard @command{tar} file (no options), the block size is 512
8413 and the record size is 10240, for a blocking factor of 20. What the
8414 @option{--blocking-factor} option does is sets the blocking factor,
8415 changing the record size while leaving the block size at 512 bytes.
8416 20 was fine for ancient 800 or 1600 bpi reel-to-reel tape drives;
8417 most tape drives these days prefer much bigger records in order to
8418 stream and not waste tape. When writing tapes for myself, some tend
8419 to use a factor of the order of 2048, say, giving a record size of
8420 around one megabyte.
8422 If you use a blocking factor larger than 20, older @command{tar}
8423 programs might not be able to read the archive, so we recommend this
8424 as a limit to use in practice. @GNUTAR{}, however,
8425 will support arbitrarily large record sizes, limited only by the
8426 amount of virtual memory or the physical characteristics of the tape
8430 * Format Variations:: Format Variations
8431 * Blocking Factor:: The Blocking Factor of an Archive
8434 @node Format Variations
8435 @subsection Format Variations
8436 @cindex Format Parameters
8437 @cindex Format Options
8438 @cindex Options, archive format specifying
8439 @cindex Options, format specifying
8442 Format parameters specify how an archive is written on the archive
8443 media. The best choice of format parameters will vary depending on
8444 the type and number of files being archived, and on the media used to
8447 To specify format parameters when accessing or creating an archive,
8448 you can use the options described in the following sections.
8449 If you do not specify any format parameters, @command{tar} uses
8450 default parameters. You cannot modify a compressed archive.
8451 If you create an archive with the @option{--blocking-factor} option
8452 specified (@pxref{Blocking Factor}), you must specify that
8453 blocking-factor when operating on the archive. @xref{Formats}, for other
8454 examples of format parameter considerations.
8456 @node Blocking Factor
8457 @subsection The Blocking Factor of an Archive
8458 @cindex Blocking Factor
8460 @cindex Number of blocks per record
8461 @cindex Number of bytes per record
8462 @cindex Bytes per record
8463 @cindex Blocks per record
8466 @opindex blocking-factor
8467 The data in an archive is grouped into blocks, which are 512 bytes.
8468 Blocks are read and written in whole number multiples called
8469 @dfn{records}. The number of blocks in a record (ie. the size of a
8470 record in units of 512 bytes) is called the @dfn{blocking factor}.
8471 The @option{--blocking-factor=@var{512-size}} (@option{-b
8472 @var{512-size}}) option specifies the blocking factor of an archive.
8473 The default blocking factor is typically 20 (i.e., 10240 bytes), but
8474 can be specified at installation. To find out the blocking factor of
8475 an existing archive, use @samp{tar --list --file=@var{archive-name}}.
8476 This may not work on some devices.
8478 Records are separated by gaps, which waste space on the archive media.
8479 If you are archiving on magnetic tape, using a larger blocking factor
8480 (and therefore larger records) provides faster throughput and allows you
8481 to fit more data on a tape (because there are fewer gaps). If you are
8482 archiving on cartridge, a very large blocking factor (say 126 or more)
8483 greatly increases performance. A smaller blocking factor, on the other
8484 hand, may be useful when archiving small files, to avoid archiving lots
8485 of nulls as @command{tar} fills out the archive to the end of the record.
8486 In general, the ideal record size depends on the size of the
8487 inter-record gaps on the tape you are using, and the average size of the
8488 files you are archiving. @xref{create}, for information on
8491 @FIXME{Need example of using a cartridge with blocking factor=126 or more.}
8493 Archives with blocking factors larger than 20 cannot be read
8494 by very old versions of @command{tar}, or by some newer versions
8495 of @command{tar} running on old machines with small address spaces.
8496 With @GNUTAR{}, the blocking factor of an archive is limited
8497 only by the maximum record size of the device containing the archive,
8498 or by the amount of available virtual memory.
8500 Also, on some systems, not using adequate blocking factors, as sometimes
8501 imposed by the device drivers, may yield unexpected diagnostics. For
8502 example, this has been reported:
8505 Cannot write to /dev/dlt: Invalid argument
8509 In such cases, it sometimes happen that the @command{tar} bundled by
8510 the system is aware of block size idiosyncrasies, while @GNUTAR{}
8511 requires an explicit specification for the block size,
8512 which it cannot guess. This yields some people to consider
8513 @GNUTAR{} is misbehaving, because by comparison,
8514 @cite{the bundle @command{tar} works OK}. Adding @w{@kbd{-b 256}},
8515 for example, might resolve the problem.
8517 If you use a non-default blocking factor when you create an archive, you
8518 must specify the same blocking factor when you modify that archive. Some
8519 archive devices will also require you to specify the blocking factor when
8520 reading that archive, however this is not typically the case. Usually, you
8521 can use @option{--list} (@option{-t}) without specifying a blocking factor---@command{tar}
8522 reports a non-default record size and then lists the archive members as
8523 it would normally. To extract files from an archive with a non-standard
8524 blocking factor (particularly if you're not sure what the blocking factor
8525 is), you can usually use the @option{--read-full-records} (@option{-B}) option while
8526 specifying a blocking factor larger then the blocking factor of the archive
8527 (ie. @samp{tar --extract --read-full-records --blocking-factor=300}.
8528 @xref{list}, for more information on the @option{--list} (@option{-t})
8529 operation. @xref{Reading}, for a more detailed explanation of that option.
8532 @item --blocking-factor=@var{number}
8533 @itemx -b @var{number}
8534 Specifies the blocking factor of an archive. Can be used with any
8535 operation, but is usually not necessary with @option{--list} (@option{-t}).
8541 @item -b @var{blocks}
8542 @itemx --blocking-factor=@var{blocks}
8543 Set record size to @math{@var{blocks} * 512} bytes.
8545 This option is used to specify a @dfn{blocking factor} for the archive.
8546 When reading or writing the archive, @command{tar}, will do reads and writes
8547 of the archive in records of @math{@var{block}*512} bytes. This is true
8548 even when the archive is compressed. Some devices requires that all
8549 write operations be a multiple of a certain size, and so, @command{tar}
8550 pads the archive out to the next record boundary.
8552 The default blocking factor is set when @command{tar} is compiled, and is
8553 typically 20. Blocking factors larger than 20 cannot be read by very
8554 old versions of @command{tar}, or by some newer versions of @command{tar}
8555 running on old machines with small address spaces.
8557 With a magnetic tape, larger records give faster throughput and fit
8558 more data on a tape (because there are fewer inter-record gaps).
8559 If the archive is in a disk file or a pipe, you may want to specify
8560 a smaller blocking factor, since a large one will result in a large
8561 number of null bytes at the end of the archive.
8563 When writing cartridge or other streaming tapes, a much larger
8564 blocking factor (say 126 or more) will greatly increase performance.
8565 However, you must specify the same blocking factor when reading or
8566 updating the archive.
8568 Apparently, Exabyte drives have a physical block size of 8K bytes.
8569 If we choose our blocksize as a multiple of 8k bytes, then the problem
8570 seems to disappear. Id est, we are using block size of 112 right
8571 now, and we haven't had the problem since we switched@dots{}
8573 With @GNUTAR{} the blocking factor is limited only
8574 by the maximum record size of the device containing the archive, or by
8575 the amount of available virtual memory.
8577 However, deblocking or reblocking is virtually avoided in a special
8578 case which often occurs in practice, but which requires all the
8579 following conditions to be simultaneously true:
8582 the archive is subject to a compression option,
8584 the archive is not handled through standard input or output, nor
8585 redirected nor piped,
8587 the archive is directly handled to a local disk, instead of any special
8590 @option{--blocking-factor} is not explicitly specified on the @command{tar}
8594 If the output goes directly to a local disk, and not through
8595 stdout, then the last write is not extended to a full record size.
8596 Otherwise, reblocking occurs. Here are a few other remarks on this
8602 @command{gzip} will complain about trailing garbage if asked to
8603 uncompress a compressed archive on tape, there is an option to turn
8604 the message off, but it breaks the regularity of simply having to use
8605 @samp{@var{prog} -d} for decompression. It would be nice if gzip was
8606 silently ignoring any number of trailing zeros. I'll ask Jean-loup
8607 Gailly, by sending a copy of this message to him.
8610 @command{compress} does not show this problem, but as Jean-loup pointed
8611 out to Michael, @samp{compress -d} silently adds garbage after
8612 the result of decompression, which tar ignores because it already
8613 recognized its end-of-file indicator. So this bug may be safely
8617 @samp{gzip -d -q} will be silent about the trailing zeros indeed,
8618 but will still return an exit status of 2 which tar reports in turn.
8619 @command{tar} might ignore the exit status returned, but I hate doing
8620 that, as it weakens the protection @command{tar} offers users against
8621 other possible problems at decompression time. If @command{gzip} was
8622 silently skipping trailing zeros @emph{and} also avoiding setting the
8623 exit status in this innocuous case, that would solve this situation.
8626 @command{tar} should become more solid at not stopping to read a pipe at
8627 the first null block encountered. This inelegantly breaks the pipe.
8628 @command{tar} should rather drain the pipe out before exiting itself.
8631 @opindex ignore-zeros, short description
8633 @itemx --ignore-zeros
8634 Ignore blocks of zeros in archive (means EOF).
8636 The @option{--ignore-zeros} (@option{-i}) option causes @command{tar} to ignore blocks
8637 of zeros in the archive. Normally a block of zeros indicates the
8638 end of the archive, but when reading a damaged archive, or one which
8639 was created by concatenating several archives together, this option
8640 allows @command{tar} to read the entire archive. This option is not on
8641 by default because many versions of @command{tar} write garbage after
8644 Note that this option causes @command{tar} to read to the end of the
8645 archive file, which may sometimes avoid problems when multiple files
8646 are stored on a single physical tape.
8648 @opindex read-full-records, short description
8650 @itemx --read-full-records
8651 Reblock as we read (for reading 4.2BSD pipes).
8653 If @option{--read-full-records} is used, @command{tar}
8654 will not panic if an attempt to read a record from the archive does
8655 not return a full record. Instead, @command{tar} will keep reading
8656 until it has obtained a full
8659 This option is turned on by default when @command{tar} is reading
8660 an archive from standard input, or from a remote machine. This is
8661 because on BSD Unix systems, a read of a pipe will return however
8662 much happens to be in the pipe, even if it is less than @command{tar}
8663 requested. If this option was not used, @command{tar} would fail as
8664 soon as it read an incomplete record from the pipe.
8666 This option is also useful with the commands for updating an archive.
8672 @FIXME{Appropriate options should be moved here from elsewhere.}
8674 @cindex blocking factor
8675 @cindex tape blocking
8677 When handling various tapes or cartridges, you have to take care of
8678 selecting a proper blocking, that is, the number of disk blocks you
8679 put together as a single tape block on the tape, without intervening
8680 tape gaps. A @dfn{tape gap} is a small landing area on the tape
8681 with no information on it, used for decelerating the tape to a
8682 full stop, and for later regaining the reading or writing speed.
8683 When the tape driver starts reading a record, the record has to
8684 be read whole without stopping, as a tape gap is needed to stop the
8685 tape motion without loosing information.
8687 @cindex Exabyte blocking
8688 @cindex DAT blocking
8689 Using higher blocking (putting more disk blocks per tape block) will use
8690 the tape more efficiently as there will be less tape gaps. But reading
8691 such tapes may be more difficult for the system, as more memory will be
8692 required to receive at once the whole record. Further, if there is a
8693 reading error on a huge record, this is less likely that the system will
8694 succeed in recovering the information. So, blocking should not be too
8695 low, nor it should be too high. @command{tar} uses by default a blocking of
8696 20 for historical reasons, and it does not really matter when reading or
8697 writing to disk. Current tape technology would easily accommodate higher
8698 blockings. Sun recommends a blocking of 126 for Exabytes and 96 for DATs.
8699 We were told that for some DLT drives, the blocking should be a multiple
8700 of 4Kb, preferably 64Kb (@w{@kbd{-b 128}}) or 256 for decent performance.
8701 Other manufacturers may use different recommendations for the same tapes.
8702 This might also depends of the buffering techniques used inside modern
8703 tape controllers. Some imposes a minimum blocking, or a maximum blocking.
8704 Others request blocking to be some exponent of two.
8706 So, there is no fixed rule for blocking. But blocking at read time
8707 should ideally be the same as blocking used at write time. At one place
8708 I know, with a wide variety of equipment, they found it best to use a
8709 blocking of 32 to guarantee that their tapes are fully interchangeable.
8711 I was also told that, for recycled tapes, prior erasure (by the same
8712 drive unit that will be used to create the archives) sometimes lowers
8713 the error rates observed at rewriting time.
8715 I might also use @option{--number-blocks} instead of
8716 @option{--block-number}, so @option{--block} will then expand to
8717 @option{--blocking-factor} unambiguously.
8720 @section Many Archives on One Tape
8722 @FIXME{Appropriate options should be moved here from elsewhere.}
8724 @findex ntape @r{device}
8725 Most tape devices have two entries in the @file{/dev} directory, or
8726 entries that come in pairs, which differ only in the minor number for
8727 this device. Let's take for example @file{/dev/tape}, which often
8728 points to the only or usual tape device of a given system. There might
8729 be a corresponding @file{/dev/nrtape} or @file{/dev/ntape}. The simpler
8730 name is the @emph{rewinding} version of the device, while the name
8731 having @samp{nr} in it is the @emph{no rewinding} version of the same
8734 A rewinding tape device will bring back the tape to its beginning point
8735 automatically when this device is opened or closed. Since @command{tar}
8736 opens the archive file before using it and closes it afterwards, this
8737 means that a simple:
8740 $ @kbd{tar cf /dev/tape @var{directory}}
8744 will reposition the tape to its beginning both prior and after saving
8745 @var{directory} contents to it, thus erasing prior tape contents and
8746 making it so that any subsequent write operation will destroy what has
8749 @cindex tape positioning
8750 So, a rewinding device is normally meant to hold one and only one file.
8751 If you want to put more than one @command{tar} archive on a given tape, you
8752 will need to avoid using the rewinding version of the tape device. You
8753 will also have to pay special attention to tape positioning. Errors in
8754 positioning may overwrite the valuable data already on your tape. Many
8755 people, burnt by past experiences, will only use rewinding devices and
8756 limit themselves to one file per tape, precisely to avoid the risk of
8757 such errors. Be fully aware that writing at the wrong position on a
8758 tape loses all information past this point and most probably until the
8759 end of the tape, and this destroyed information @emph{cannot} be
8762 To save @var{directory-1} as a first archive at the beginning of a
8763 tape, and leave that tape ready for a second archive, you should use:
8766 $ @kbd{mt -f /dev/nrtape rewind}
8767 $ @kbd{tar cf /dev/nrtape @var{directory-1}}
8771 @dfn{Tape marks} are special magnetic patterns written on the tape
8772 media, which are later recognizable by the reading hardware. These
8773 marks are used after each file, when there are many on a single tape.
8774 An empty file (that is to say, two tape marks in a row) signal the
8775 logical end of the tape, after which no file exist. Usually,
8776 non-rewinding tape device drivers will react to the close request issued
8777 by @command{tar} by first writing two tape marks after your archive, and by
8778 backspacing over one of these. So, if you remove the tape at that time
8779 from the tape drive, it is properly terminated. But if you write
8780 another file at the current position, the second tape mark will be
8781 erased by the new information, leaving only one tape mark between files.
8783 So, you may now save @var{directory-2} as a second archive after the
8784 first on the same tape by issuing the command:
8787 $ @kbd{tar cf /dev/nrtape @var{directory-2}}
8791 and so on for all the archives you want to put on the same tape.
8793 Another usual case is that you do not write all the archives the same
8794 day, and you need to remove and store the tape between two archive
8795 sessions. In general, you must remember how many files are already
8796 saved on your tape. Suppose your tape already has 16 files on it, and
8797 that you are ready to write the 17th. You have to take care of skipping
8798 the first 16 tape marks before saving @var{directory-17}, say, by using
8802 $ @kbd{mt -f /dev/nrtape rewind}
8803 $ @kbd{mt -f /dev/nrtape fsf 16}
8804 $ @kbd{tar cf /dev/nrtape @var{directory-17}}
8807 In all the previous examples, we put aside blocking considerations, but
8808 you should do the proper things for that as well. @xref{Blocking}.
8811 * Tape Positioning:: Tape Positions and Tape Marks
8812 * mt:: The @command{mt} Utility
8815 @node Tape Positioning
8816 @subsection Tape Positions and Tape Marks
8819 Just as archives can store more than one file from the file system,
8820 tapes can store more than one archive file. To keep track of where
8821 archive files (or any other type of file stored on tape) begin and
8822 end, tape archive devices write magnetic @dfn{tape marks} on the
8823 archive media. Tape drives write one tape mark between files,
8824 two at the end of all the file entries.
8826 If you think of data as a series of records "rrrr"'s, and tape marks as
8827 "*"'s, a tape might look like the following:
8830 rrrr*rrrrrr*rrrrr*rr*rrrrr**-------------------------
8833 Tape devices read and write tapes using a read/write @dfn{tape
8834 head}---a physical part of the device which can only access one
8835 point on the tape at a time. When you use @command{tar} to read or
8836 write archive data from a tape device, the device will begin reading
8837 or writing from wherever on the tape the tape head happens to be,
8838 regardless of which archive or what part of the archive the tape
8839 head is on. Before writing an archive, you should make sure that no
8840 data on the tape will be overwritten (unless it is no longer needed).
8841 Before reading an archive, you should make sure the tape head is at
8842 the beginning of the archive you want to read. You can do it manually
8843 via @code{mt} utility (@pxref{mt}). The @code{restore} script does
8844 that automatically (@pxref{Scripted Restoration}).
8846 If you want to add new archive file entries to a tape, you should
8847 advance the tape to the end of the existing file entries, backspace
8848 over the last tape mark, and write the new archive file. If you were
8849 to add two archives to the example above, the tape might look like the
8853 rrrr*rrrrrr*rrrrr*rr*rrrrr*rrr*rrrr**----------------
8857 @subsection The @command{mt} Utility
8860 @FIXME{Is it true that this only works on non-block devices?
8861 should explain the difference, (fixed or variable).}
8862 @xref{Blocking Factor}.
8864 You can use the @command{mt} utility to advance or rewind a tape past a
8865 specified number of archive files on the tape. This will allow you
8866 to move to the beginning of an archive before extracting or reading
8867 it, or to the end of all the archives before writing a new one.
8868 @FIXME{Why isn't there an "advance 'til you find two tape marks
8871 The syntax of the @command{mt} command is:
8874 @kbd{mt [-f @var{tapename}] @var{operation} [@var{number}]}
8877 where @var{tapename} is the name of the tape device, @var{number} is
8878 the number of times an operation is performed (with a default of one),
8879 and @var{operation} is one of the following:
8881 @FIXME{is there any use for record operations?}
8886 Writes @var{number} tape marks at the current position on the tape.
8889 Moves tape position forward @var{number} files.
8892 Moves tape position back @var{number} files.
8895 Rewinds the tape. (Ignores @var{number}).
8899 Rewinds the tape and takes the tape device off-line. (Ignores @var{number}).
8902 Prints status information about the tape unit.
8906 @FIXME{Is there a better way to frob the spacing on the list?}
8908 If you don't specify a @var{tapename}, @command{mt} uses the environment
8909 variable @env{TAPE}; if @env{TAPE} is not set, @command{mt} uses the device
8912 @command{mt} returns a 0 exit status when the operation(s) were
8913 successful, 1 if the command was unrecognized, and 2 if an operation
8916 @node Using Multiple Tapes
8917 @section Using Multiple Tapes
8920 Often you might want to write a large archive, one larger than will fit
8921 on the actual tape you are using. In such a case, you can run multiple
8922 @command{tar} commands, but this can be inconvenient, particularly if you
8923 are using options like @option{--exclude=@var{pattern}} or dumping entire file systems.
8924 Therefore, @command{tar} supports multiple tapes automatically.
8926 Use @option{--multi-volume} (@option{-M}) on the command line, and
8927 then @command{tar} will, when it reaches the end of the tape, prompt
8928 for another tape, and continue the archive. Each tape will have an
8929 independent archive, and can be read without needing the other. (As
8930 an exception to this, the file that @command{tar} was archiving when
8931 it ran out of tape will usually be split between the two archives; in
8932 this case you need to extract from the first archive, using
8933 @option{--multi-volume}, and then put in the second tape when
8934 prompted, so @command{tar} can restore both halves of the file.)
8936 @GNUTAR{} multi-volume archives do not use a truly portable format.
8937 You need @GNUTAR{} at both ends to process them properly.
8939 When prompting for a new tape, @command{tar} accepts any of the following
8944 Request @command{tar} to explain possible responses
8946 Request @command{tar} to exit immediately.
8947 @item n @var{file_name}
8948 Request @command{tar} to write the next volume on the file @var{file_name}.
8950 Request @command{tar} to run a subshell. This option can be disabled
8951 by giving @option{--restrict} command line option to @command{tar}.
8953 Request @command{tar} to begin writing the next volume.
8956 (You should only type @samp{y} after you have changed the tape;
8957 otherwise @command{tar} will write over the volume it just finished.)
8959 @cindex End-of-archive info script
8961 @anchor{info-script}
8962 @opindex info-script
8963 @opindex new-volume-script
8964 If you want more elaborate behavior than this, give @command{tar} the
8965 @option{--info-script=@var{script-name}}
8966 (@option{--new-volume-script=@var{script-name}}, @option{-F
8967 @var{script-name}}) option. The file @var{script-name} is expected to
8968 be a program (or shell script) to be run instead of the normal
8969 prompting procedure. It is executed without any command line
8970 arguments. Additional data is passed to it via the following
8971 environment variables:
8974 @vrindex TAR_VERSION, info script environment variable
8976 @GNUTAR{} version number.
8978 @vrindex TAR_ARCHIVE, info script environment variable
8980 The name of the archive @command{tar} is processing.
8982 @vrindex TAR_VOLUME, info script environment variable
8984 Ordinal number of the volume @command{tar} is about to start.
8986 @vrindex TAR_SUBCOMMAND, info script environment variable
8987 @item TAR_SUBCOMMAND
8988 Short option describing the operation @command{tar} is executed.
8989 @xref{Operations}, for a complete list of subcommand options.
8991 @vrindex TAR_FORMAT, info script environment variable
8993 Format of the archive being processed. @xref{Formats}, for a complete
8994 list of archive format names.
8997 The info script can instruct @command{tar} to use new archive name,
8998 by writing in to file descriptor 3 (see below for an
9001 If the info script fails, @command{tar} exits; otherwise, it begins
9002 writing the next volume.
9004 The method @command{tar} uses to detect end of tape is not perfect, and
9005 fails on some operating systems or on some devices. You can use the
9006 @option{--tape-length=@var{size}} (@option{-L @var{size}}) option if
9007 @command{tar} can't detect the end of the tape itself. This option
9008 selects @option{--multi-volume} (@option{-M}) automatically. The
9009 @var{size} argument should then be the usable size of the tape in
9010 units of 1024 bytes. But for many devices, and floppy disks in
9011 particular, this option is never required for real, as far as we know.
9013 @cindex Volume number file
9017 The volume number used by @command{tar} in its tape-change prompt
9018 can be changed; if you give the
9019 @option{--volno-file=@var{file-of-number}} option, then
9020 @var{file-of-number} should be an unexisting file to be created, or
9021 else, a file already containing a decimal number. That number will be
9022 used as the volume number of the first volume written. When
9023 @command{tar} is finished, it will rewrite the file with the
9024 now-current volume number. (This does not change the volume number
9025 written on a tape label, as per @ref{label}, it @emph{only} affects
9026 the number used in the prompt.)
9028 If you want @command{tar} to cycle through a series of files or tape
9029 drives, there are three approaches to choose from. First of all, you
9030 can give @command{tar} multiple @option{--file} options. In this case
9031 the specified files will be used, in sequence, as the successive
9032 volumes of the archive. Only when the first one in the sequence needs
9033 to be used again will @command{tar} prompt for a tape change (or run
9034 the info script). Secondly, you can use the @samp{n} response to the
9035 tape-change prompt, and, finally, you can use an info script, that
9036 writes new archive name to file descriptor. The following example
9037 illustrates this approach:
9042 echo Preparing volume $TAR_VOLUME of $TAR_ARCHIVE.
9044 name=`expr $TAR_ARCHIVE : '\(.*\)-.*'`
9045 case $TAR_SUBCOMMAND in
9047 -d|-x|-t) test -r $@{name:-$TAR_ARCHIVE@}-$TAR_VOLUME || exit 1
9052 echo $@{name:-$TAR_ARCHIVE@}-$TAR_VOLUME >&3
9056 Each volume of a multi-volume archive is an independent @command{tar}
9057 archive, complete in itself. For example, you can list or extract any
9058 volume alone; just don't specify @option{--multi-volume}
9059 (@option{-M}). However, if one file in the archive is split across
9060 volumes, the only way to extract it successfully is with a
9061 multi-volume extract command @option{--extract --multi-volume}
9062 (@option{-xM}) starting on or before the volume where the file begins.
9064 For example, let's presume someone has two tape drives on a system
9065 named @file{/dev/tape0} and @file{/dev/tape1}. For having @GNUTAR{}
9066 to switch to the second drive when it needs to write the
9067 second tape, and then back to the first tape, etc., just do either of:
9070 $ @kbd{tar --create --multi-volume --file=/dev/tape0 --file=/dev/tape1 @var{files}}
9071 $ @kbd{tar cMff /dev/tape0 /dev/tape1 @var{files}}
9075 * Multi-Volume Archives:: Archives Longer than One Tape or Disk
9076 * Tape Files:: Tape Files
9077 * Tarcat:: Concatenate Volumes into a Single Archive
9081 @node Multi-Volume Archives
9082 @subsection Archives Longer than One Tape or Disk
9083 @cindex Multi-volume archives
9086 @opindex multi-volume
9087 To create an archive that is larger than will fit on a single unit of
9088 the media, use the @option{--multi-volume} (@option{-M}) option in conjunction with
9089 the @option{--create} option (@pxref{create}). A @dfn{multi-volume}
9090 archive can be manipulated like any other archive (provided the
9091 @option{--multi-volume} option is specified), but is stored on more
9092 than one tape or disk.
9094 When you specify @option{--multi-volume}, @command{tar} does not report an
9095 error when it comes to the end of an archive volume (when reading), or
9096 the end of the media (when writing). Instead, it prompts you to load
9097 a new storage volume. If the archive is on a magnetic tape, you
9098 should change tapes when you see the prompt; if the archive is on a
9099 floppy disk, you should change disks; etc.
9101 You can read each individual volume of a multi-volume archive as if it
9102 were an archive by itself. For example, to list the contents of one
9103 volume, use @option{--list}, without @option{--multi-volume} specified.
9104 To extract an archive member from one volume (assuming it is described
9105 that volume), use @option{--extract}, again without
9106 @option{--multi-volume}.
9108 If an archive member is split across volumes (ie. its entry begins on
9109 one volume of the media and ends on another), you need to specify
9110 @option{--multi-volume} to extract it successfully. In this case, you
9111 should load the volume where the archive member starts, and use
9112 @samp{tar --extract --multi-volume}---@command{tar} will prompt for later
9113 volumes as it needs them. @xref{extracting archives}, for more
9114 information about extracting archives.
9116 @option{--info-script=@var{script-name}}
9117 (@option{--new-volume-script=@var{script-name}}, @option{-F
9118 @var{script-name}}) (@pxref{info-script}) is like
9119 @option{--multi-volume} (@option{-M}), except that @command{tar} does
9120 not prompt you directly to change media volumes when a volume is
9121 full---instead, @command{tar} runs commands you have stored in
9122 @var{script-name}. For example, this option can be used to eject
9123 cassettes, or to broadcast messages such as @samp{Someone please come
9124 change my tape} when performing unattended backups. When
9125 @var{script-name} is done, @command{tar} will assume that the media
9128 Multi-volume archives can be modified like any other archive. To add
9129 files to a multi-volume archive, you need to only mount the last
9130 volume of the archive media (and new volumes, if needed). For all
9131 other operations, you need to use the entire archive.
9133 If a multi-volume archive was labeled using
9134 @option{--label=@var{archive-label}} (@option{-V @var{archive-label}})
9135 (@pxref{label}) when it was created, @command{tar} will not
9136 automatically label volumes which are added later. To label
9137 subsequent volumes, specify @option{--label=@var{archive-label}} again
9138 in conjunction with the @option{--append}, @option{--update} or
9139 @option{--concatenate} operation.
9141 @cindex Labeling multi-volume archives
9144 @FIXME{There should be a sample program here, including an exit
9145 before end. Is the exit status even checked in tar? :-(}
9148 @item --multi-volume
9150 Creates a multi-volume archive, when used in conjunction with
9151 @option{--create} (@option{-c}). To perform any other operation on a multi-volume
9152 archive, specify @option{--multi-volume} in conjunction with that
9155 @item --info-script=@var{program-file}
9156 @itemx --new-volume-script=@var{program-file}
9157 @itemx -F @var{program-file}
9158 Creates a multi-volume archive via a script. Used in conjunction with
9159 @option{--create} (@option{-c}). @xref{info-script}, dor a detailed discussion.
9162 Beware that there is @emph{no} real standard about the proper way, for
9163 a @command{tar} archive, to span volume boundaries. If you have a
9164 multi-volume created by some vendor's @command{tar}, there is almost
9165 no chance you could read all the volumes with @GNUTAR{}.
9166 The converse is also true: you may not expect
9167 multi-volume archives created by @GNUTAR{} to be
9168 fully recovered by vendor's @command{tar}. Since there is little
9169 chance that, in mixed system configurations, some vendor's
9170 @command{tar} will work on another vendor's machine, and there is a
9171 great chance that @GNUTAR{} will work on most of
9172 them, your best bet is to install @GNUTAR{} on all
9173 machines between which you know exchange of files is possible.
9176 @subsection Tape Files
9179 To give the archive a name which will be recorded in it, use the
9180 @option{--label=@var{volume-label}} (@option{-V @var{volume-label}})
9181 option. This will write a special block identifying
9182 @var{volume-label} as the name of the archive to the front of the
9183 archive which will be displayed when the archive is listed with
9184 @option{--list}. If you are creating a multi-volume archive with
9185 @option{--multi-volume} (@pxref{Using Multiple Tapes}), then the
9186 volume label will have @samp{Volume @var{nnn}} appended to the name
9187 you give, where @var{nnn} is the number of the volume of the archive.
9188 (If you use the @option{--label=@var{volume-label}}) option when
9189 reading an archive, it checks to make sure the label on the tape
9190 matches the one you give. @xref{label}.
9192 When @command{tar} writes an archive to tape, it creates a single
9193 tape file. If multiple archives are written to the same tape, one
9194 after the other, they each get written as separate tape files. When
9195 extracting, it is necessary to position the tape at the right place
9196 before running @command{tar}. To do this, use the @command{mt} command.
9197 For more information on the @command{mt} command and on the organization
9198 of tapes into a sequence of tape files, see @ref{mt}.
9200 People seem to often do:
9203 @kbd{--label="@var{some-prefix} `date +@var{some-format}`"}
9206 or such, for pushing a common date in all volumes or an archive set.
9209 @subsection Concatenate Volumes into a Single Archive
9212 Sometimes it is necessary to convert existing @GNUTAR{} multi-volume
9213 archive to a single @command{tar} archive. Simply concatenating all
9214 volumes into one will not work, since each volume carries an additional
9215 information at the beginning. @GNUTAR{} is shipped with the shell
9216 script @command{tarcat} designed for this purpose.
9218 The script takes a list of files comprising a multi-volume archive
9219 and creates the resulting archive at the standard output. For example:
9222 @kbd{tarcat vol.1 vol.2 vol.3 | tar tf -}
9225 The script implements a simple heuristics to determine the format of
9226 the first volume file and to decide how to process the rest of the
9227 files. However, it makes no attempt to verify whether the files are
9228 given in order or even if they are valid @command{tar} archives.
9229 It uses @command{dd} and does not filter its standard error, so you
9230 will usually see lots of spurious messages.
9232 @FIXME{The script is not installed. Should we install it?}
9235 @section Including a Label in the Archive
9236 @cindex Labeling an archive
9237 @cindex Labels on the archive media
9241 To avoid problems caused by misplaced paper labels on the archive
9242 media, you can include a @dfn{label} entry---an archive member which
9243 contains the name of the archive---in the archive itself. Use the
9244 @option{--label=@var{archive-label}} (@option{-V @var{archive-label}})
9245 option in conjunction with the @option{--create} operation to include
9246 a label entry in the archive as it is being created.
9249 @item --label=@var{archive-label}
9250 @itemx -V @var{archive-label}
9251 Includes an @dfn{archive-label} at the beginning of the archive when
9252 the archive is being created, when used in conjunction with the
9253 @option{--create} operation. Checks to make sure the archive label
9254 matches the one specified (when used in conjunction with any other
9258 If you create an archive using both
9259 @option{--label=@var{archive-label}} (@option{-V @var{archive-label}})
9260 and @option{--multi-volume} (@option{-M}), each volume of the archive
9261 will have an archive label of the form @samp{@var{archive-label}
9262 Volume @var{n}}, where @var{n} is 1 for the first volume, 2 for the
9263 next, and so on. @xref{Using Multiple Tapes}, for information on
9264 creating multiple volume archives.
9266 @cindex Volume label, listing
9267 @cindex Listing volume label
9268 The volume label will be displayed by @option{--list} along with
9269 the file contents. If verbose display is requested, it will also be
9270 explicitely marked as in the example below:
9274 $ @kbd{tar --verbose --list --file=iamanarchive}
9275 V--------- 0 0 0 1992-03-07 12:01 iamalabel--Volume Header--
9276 -rw-r--r-- ringo user 40 1990-05-21 13:30 iamafilename
9281 @anchor{--test-label option}
9282 However, @option{--list} option will cause listing entire
9283 contents of the archive, which may be undesirable (for example, if the
9284 archive is stored on a tape). You can request checking only the volume
9285 by specifying @option{--test-label} option. This option reads only the
9286 first block of an archive, so it can be used with slow storage
9287 devices. For example:
9291 $ @kbd{tar --test-label --file=iamanarchive}
9296 If @option{--test-label} is used with a single command line
9297 argument, @command{tar} compares the volume label with the
9298 argument. It exits with code 0 if the two strings match, and with code
9299 2 otherwise. In this case no output is displayed. For example:
9303 $ @kbd{tar --test-label --file=iamanarchive 'iamalable'}
9305 $ @kbd{tar --test-label --file=iamanarchive 'iamalable' alabel}
9310 If you request any operation, other than @option{--create}, along
9311 with using @option{--label} option, @command{tar} will first check if
9312 the archive label matches the one specified and will refuse to proceed
9313 if it does not. Use this as a safety precaution to avoid accidentally
9314 overwriting existing archives. For example, if you wish to add files
9315 to @file{archive}, presumably labelled with string @samp{My volume},
9320 $ @kbd{tar -rf archive --label 'My volume' .}
9321 tar: Archive not labeled to match `My volume'
9326 in case its label does not match. This will work even if
9327 @file{archive} is not labelled at all.
9329 Similarly, @command{tar} will refuse to list or extract the
9330 archive if its label doesn't match the @var{archive-label}
9331 specified. In those cases, @var{archive-label} argument is interpreted
9332 as a globbing-style pattern which must match the actual magnetic
9333 volume label. @xref{exclude}, for a precise description of how match
9334 is attempted@footnote{Previous versions of @command{tar} used full
9335 regular expression matching, or before that, only exact string
9336 matching, instead of wildcard matchers. We decided for the sake of
9337 simplicity to use a uniform matching device through
9338 @command{tar}.}. If the switch @option{--multi-volume} (@option{-M}) is being used,
9339 the volume label matcher will also suffix @var{archive-label} by
9340 @w{@samp{ Volume [1-9]*}} if the initial match fails, before giving
9341 up. Since the volume numbering is automatically added in labels at
9342 creation time, it sounded logical to equally help the user taking care
9343 of it when the archive is being read.
9345 The @option{--label} was once called @option{--volume}, but is not
9346 available under that name anymore.
9348 You can also use @option{--label} to get a common information on
9349 all tapes of a series. For having this information different in each
9350 series created through a single script used on a regular basis, just
9351 manage to get some date string as part of the label. For example:
9355 $ @kbd{tar cfMV /dev/tape "Daily backup for `date +%Y-%m-%d`"}
9356 $ @kbd{tar --create --file=/dev/tape --multi-volume \
9357 --volume="Daily backup for `date +%Y-%m-%d`"}
9361 Also note that each label has its own date and time, which corresponds
9362 to when @GNUTAR{} initially attempted to write it,
9363 often soon after the operator launches @command{tar} or types the
9364 carriage return telling that the next tape is ready. Comparing date
9365 labels does give an idea of tape throughput only if the delays for
9366 rewinding tapes and the operator switching them were negligible, which
9367 is usually not the case.
9370 @section Verifying Data as It is Stored
9371 @cindex Verifying a write operation
9372 @cindex Double-checking a write operation
9377 @opindex verify, short description
9378 Attempt to verify the archive after writing.
9381 This option causes @command{tar} to verify the archive after writing it.
9382 Each volume is checked after it is written, and any discrepancies
9383 are recorded on the standard error output.
9385 Verification requires that the archive be on a back-space-able medium.
9386 This means pipes, some cartridge tape drives, and some other devices
9389 You can insure the accuracy of an archive by comparing files in the
9390 system with archive members. @command{tar} can compare an archive to the
9391 file system as the archive is being written, to verify a write
9392 operation, or can compare a previously written archive, to insure that
9395 @opindex verify, using with @option{--create}
9396 @opindex create, using with @option{--verify}
9397 To check for discrepancies in an archive immediately after it is
9398 written, use the @option{--verify} (@option{-W}) option in conjunction with
9399 the @option{--create} operation. When this option is
9400 specified, @command{tar} checks archive members against their counterparts
9401 in the file system, and reports discrepancies on the standard error.
9403 To verify an archive, you must be able to read it from before the end
9404 of the last written entry. This option is useful for detecting data
9405 errors on some tapes. Archives written to pipes, some cartridge tape
9406 drives, and some other devices cannot be verified.
9408 One can explicitly compare an already made archive with the file
9409 system by using the @option{--compare} (@option{--diff}, @option{-d})
9410 option, instead of using the more automatic @option{--verify} option.
9413 Note that these two options have a slightly different intent. The
9414 @option{--compare} option checks how identical are the logical contents of some
9415 archive with what is on your disks, while the @option{--verify} option is
9416 really for checking if the physical contents agree and if the recording
9417 media itself is of dependable quality. So, for the @option{--verify}
9418 operation, @command{tar} tries to defeat all in-memory cache pertaining to
9419 the archive, while it lets the speed optimization undisturbed for the
9420 @option{--compare} option. If you nevertheless use @option{--compare} for
9421 media verification, you may have to defeat the in-memory cache yourself,
9422 maybe by opening and reclosing the door latch of your recording unit,
9423 forcing some doubt in your operating system about the fact this is really
9424 the same volume as the one just written or read.
9426 The @option{--verify} option would not be necessary if drivers were indeed
9427 able to detect dependably all write failures. This sometimes require many
9428 magnetic heads, some able to read after the writes occurred. One would
9429 not say that drivers unable to detect all cases are necessarily flawed,
9430 as long as programming is concerned.
9432 The @option{--verify} (@option{-W}) option will not work in
9433 conjunction with the @option{--multi-volume} (@option{-M}) option or
9434 the @option{--append} (@option{-r}), @option{--update} (@option{-u})
9435 and @option{--delete} operations. @xref{Operations}, for more
9436 information on these operations.
9438 Also, since @command{tar} normally strips leading @samp{/} from file
9439 names (@pxref{absolute}), a command like @samp{tar --verify -cf
9440 /tmp/foo.tar /etc} will work as desired only if the working directory is
9441 @file{/}, as @command{tar} uses the archive's relative member names
9442 (e.g., @file{etc/motd}) when verifying the archive.
9444 @node Write Protection
9445 @section Write Protection
9447 Almost all tapes and diskettes, and in a few rare cases, even disks can
9448 be @dfn{write protected}, to protect data on them from being changed.
9449 Once an archive is written, you should write protect the media to prevent
9450 the archive from being accidentally overwritten or deleted. (This will
9451 protect the archive from being changed with a tape or floppy drive---it
9452 will not protect it from magnet fields or other physical hazards).
9454 The write protection device itself is usually an integral part of the
9455 physical media, and can be a two position (write enabled/write
9456 disabled) switch, a notch which can be popped out or covered, a ring
9457 which can be removed from the center of a tape reel, or some other
9463 This appendix lists some important user-visible changes between
9464 version @GNUTAR{} @value{VERSION} and previous versions. An up-to-date
9465 version of this document is available at
9466 @uref{http://www.gnu.org/@/software/@/tar/manual/changes.html,the
9467 @GNUTAR{} documentation page}.
9470 @item Use of globbing patterns when listing and extracting.
9472 Previous versions of GNU tar assumed shell-style globbing when
9473 extracting from or listing an archive. For example:
9476 $ @kbd{tar xf foo.tar '*.c'}
9479 would extract all files whose names end in @samp{.c}. This behavior
9480 was not documented and was incompatible with traditional tar
9481 implementations. Therefore, starting from version 1.15.91, GNU tar
9482 no longer uses globbing by default. For example, the above invocation
9483 is now interpreted as a request to extract from the archive the file
9486 To facilitate transition to the new behavior for those users who got
9487 used to the previous incorrect one, @command{tar} will print a warning
9488 if it finds out that a requested member was not found in the archive
9489 and its name looks like a globbing pattern. For example:
9492 $ @kbd{tar xf foo.tar '*.c'}
9493 tar: Pattern matching characters used in file names. Please,
9494 tar: use --wildcards to enable pattern matching, or --no-wildcards to
9495 tar: suppress this warning.
9496 tar: *.c: Not found in archive
9497 tar: Error exit delayed from previous errors
9500 To treat member names as globbing patterns, use --wildcards option.
9501 If you want to tar to mimic the behavior of versions prior to 1.15.91,
9502 add this option to your @env{TAR_OPTIONS} variable.
9504 @xref{Wildcards}, for the detailed discussion of the use of globbing
9505 patterns by @GNUTAR{}.
9506 @FIXME{Check this reference.}
9508 @item Use of short option @option{-o}.
9510 Earlier versions of @GNUTAR{} understood @option{-o} command line
9511 option as a synonym for @option{--old-archive}.
9513 @GNUTAR{} starting from version 1.13.90 understands this option as
9514 a synonym for @option{--no-same-owner}. This is compatible with
9515 UNIX98 @command{tar} implementations.
9517 However, to facilitate transition, @option{-o} option retains its
9518 old semantics when it is used with one of archive-creation commands.
9519 Users are encouraged to use @option{--format=oldgnu} instead.
9521 It is especially important, since versions of @acronym{GNU} Automake
9522 up to and including 1.8.4 invoke tar with this option to produce
9523 distribution tarballs. @xref{Formats,v7}, for the detailed discussion
9524 of this issue and its implications.
9526 @FIXME{Refer to tar-v7 description in automake.info.}
9528 Future versions of @GNUTAR{} will understand @option{-o} only as a
9529 synonym for @option{--no-same-owner}.
9531 @item Use of short option @option{-l}
9533 Earlier versions of @GNUTAR{} understood @option{-l} option as a
9534 synonym for @option{--one-file-system}. Since such usage contradicted
9535 to UNIX98 specification and harmed compatibility with other
9536 implementation, it was declared deprecated in version 1.14. However,
9537 to facilitate transition to its new semantics, it was supported by
9538 versions 1.15 and 1.15.90. The present use of @option{-l} as a short
9539 variant of @option{--check-links} was introduced in version 1.15.91.
9541 @item Use of options @option{--portability} and @option{--old-archive}
9543 These options are deprecated. Please use @option{--format=v7} instead.
9545 @item Use of option @option{--posix}
9547 This option is deprecated. Please use @option{--format=posix} instead.
9550 @node Configuring Help Summary
9551 @appendix Configuring Help Summary
9553 Running @kbd{tar --help} displays the short @command{tar} option
9554 summary (@pxref{help}). This summary is organised by @dfn{groups} of
9555 semantically close options. The options within each group are printed
9556 in the following order: a short option, eventually followed by a list
9557 of corresponding long option names, followed by a short description of
9558 the option. For example, here is an excerpt from the actual @kbd{tar
9562 Main operation mode:
9564 -A, --catenate, --concatenate append tar files to an archive
9565 -c, --create create a new archive
9566 -d, --diff, --compare find differences between archive and
9568 --delete delete from the archive
9571 @vrindex ARGP_HELP_FMT, environment variable
9572 The exact visual representation of the help output is configurable via
9573 @env{ARGP_HELP_FMT} environment variable. The value of this variable
9574 is a comma-separated list of @dfn{format variable} assignments. There
9575 are two kinds of format variables. An @dfn{offset variable} keeps the
9576 offset of some part of help output text from the leftmost column on
9577 the screen. A @dfn{boolean} variable is a flag that toggles some
9578 output feature on or off. Depending on the type of the corresponding
9579 variable, there are two kinds of assignments:
9582 @item Offset assignment
9584 The assignment to an offset variable has the following syntax:
9587 @var{variable}=@var{value}
9591 where @var{variable} is the variable name, and @var{value} is a
9592 numeric value to be assigned to the variable.
9594 @item Boolean assignment
9596 To assign @code{true} value to a variable, simply put this variable name. To
9597 assign @code{false} value, prefix the variable name with @samp{no-}. For
9602 # Assign @code{true} value:
9604 # Assign @code{false} value:
9610 Following variables are declared:
9612 @deftypevr {Help Output} boolean dup-args
9613 If true, arguments for an option are shown with both short and long
9614 options, even when a given option has both forms, for example:
9617 -f ARCHIVE, --file=ARCHIVE use archive file or device ARCHIVE
9620 If false, then if an option has both short and long forms, the
9621 argument is only shown with the long one, for example:
9624 -f, --file=ARCHIVE use archive file or device ARCHIVE
9628 and a message indicating that the argument is applicable to both
9629 forms is printed below the options. This message can be disabled
9630 using @code{dup-args-note} (see below).
9632 The default is false.
9635 @deftypevr {Help Output} boolean dup-args-note
9636 If this variable is true, which is the default, the following notice
9637 is displayed at the end of the help output:
9640 Mandatory or optional arguments to long options are also mandatory or
9641 optional for any corresponding short options.
9644 Setting @code{no-dup-args-note} inhibits this message. Normally, only one of
9645 variables @code{dup-args} or @code{dup-args-note} should be set.
9648 @deftypevr {Help Output} offset short-opt-col
9649 Column in which short options start. Default is 2.
9653 $ @kbd{tar --help|grep ARCHIVE}
9654 -f, --file=ARCHIVE use archive file or device ARCHIVE
9655 $ @kbd{ARGP_HELP_FMT=short-opt-col=6 tar --help|grep ARCHIVE}
9656 -f, --file=ARCHIVE use archive file or device ARCHIVE
9661 @deftypevr {Help Output} offset long-opt-col
9662 Column in which long options start. Default is 6. For example:
9666 $ @kbd{tar --help|grep ARCHIVE}
9667 -f, --file=ARCHIVE use archive file or device ARCHIVE
9668 $ @kbd{ARGP_HELP_FMT=long-opt-col=16 tar --help|grep ARCHIVE}
9669 -f, --file=ARCHIVE use archive file or device ARCHIVE
9674 @deftypevr {Help Output} offset doc-opt-col
9675 Column in which @dfn{doc options} start. A doc option isn't actually
9676 an option, but rather an arbitrary piece of documentation that is
9677 displayed in much the same manner as the options. For example, in
9678 the description of @option{--format} option:
9682 -H, --format=FORMAT create archive of the given format.
9684 FORMAT is one of the following:
9686 gnu GNU tar 1.13.x format
9687 oldgnu GNU format as per tar <= 1.12
9688 pax POSIX 1003.1-2001 (pax) format
9690 ustar POSIX 1003.1-1988 (ustar) format
9691 v7 old V7 tar format
9696 the format names are doc options. Thus, if you set
9697 @kbd{ARGP_HELP_FMT=doc-opt-col=6} the above part of the help output
9698 will look as follows:
9702 -H, --format=FORMAT create archive of the given format.
9704 FORMAT is one of the following:
9706 gnu GNU tar 1.13.x format
9707 oldgnu GNU format as per tar <= 1.12
9708 pax POSIX 1003.1-2001 (pax) format
9710 ustar POSIX 1003.1-1988 (ustar) format
9711 v7 old V7 tar format
9716 @deftypevr {Help Output} offset opt-doc-col
9717 Column in which option description starts. Default is 29.
9721 $ @kbd{tar --help|grep ARCHIVE}
9722 -f, --file=ARCHIVE use archive file or device ARCHIVE
9723 $ @kbd{ARGP_HELP_FMT=opt-doc-col=19 tar --help|grep ARCHIVE}
9724 -f, --file=ARCHIVE use archive file or device ARCHIVE
9725 $ @kbd{ARGP_HELP_FMT=opt-doc-col=9 tar --help|grep ARCHIVE}
9727 use archive file or device ARCHIVE
9732 Notice, that the description starts on a separate line if
9733 @code{opt-doc-col} value is too small.
9736 @deftypevr {Help Output} offset header-col
9737 Column in which @dfn{group headers} are printed. A group header is a
9738 descriptive text preceding an option group. For example, in the
9742 Main operation mode:
9744 -A, --catenate, --concatenate append tar files to
9746 -c, --create create a new archive
9749 @samp{Main operation mode:} is the group header.
9751 The default value is 1.
9754 @deftypevr {Help Output} offset usage-indent
9755 Indentation of wrapped usage lines. Affects @option{--usage}
9756 output. Default is 12.
9759 @deftypevr {Help Output} offset rmargin
9760 Right margin of the text output. Used for wrapping.
9765 @include genfile.texi
9767 @node Snapshot Files
9768 @appendix Format of the Incremental Snapshot Files
9769 @include snapshot.texi
9771 @node Free Software Needs Free Documentation
9772 @appendix Free Software Needs Free Documentation
9773 @include freemanuals.texi
9775 @node Copying This Manual
9776 @appendix Copying This Manual
9779 * GNU Free Documentation License:: License for copying this manual
9784 @node Index of Command Line Options
9785 @appendix Index of Command Line Options
9787 This appendix contains an index of all @GNUTAR{} long command line
9788 options. The options are listed without the preceeding double-dash.
9791 @item Make sure @emph{all} options are indexed.
9792 @item Provide an index of short options
9807 @c texinfo-column-for-description: 32