1 bup: It backs things up
2 =======================
4 bup is a program that backs things up. It's short for "backup." Can you
5 believe that nobody else has named an open source program "bup" after all
8 Despite its unassuming name, bup is pretty cool. To give you an idea of
9 just how cool it is, I wrote you this poem:
12 What rhymes with awesome?
14 But that's irrelevant.
16 Hmm. Did that help? Maybe prose is more useful after all.
19 Reasons bup is awesome
20 ----------------------
22 bup has a few advantages over other backup software:
24 - It uses a rolling checksum algorithm (similar to rsync) to split large
25 files into chunks. The most useful result of this is you can backup huge
26 virtual machine (VM) disk images, databases, and XML files incrementally,
27 even though they're typically all in one huge file, and not use tons of
28 disk space for multiple versions.
30 - It uses the packfile format from git (the open source version control
31 system), so you can access the stored data even if you don't like bup's
34 - Unlike git, it writes packfiles *directly* (instead of having a separate
35 garbage collection / repacking stage) so it's fast even with gratuitously
36 huge amounts of data. bup's improved index formats also allow you to
37 track far more filenames than git (millions) and keep track of far more
38 objects (hundreds or thousands of gigabytes).
40 - Data is "automagically" shared between incremental backups without having
41 to know which backup is based on which other one - even if the backups
42 are made from two different computers that don't even know about each
43 other. You just tell bup to back stuff up, and it saves only the minimum
44 amount of data needed.
46 - You can back up directly to a remote bup server, without needing tons of
47 temporary disk space on the computer being backed up. And if your backup
48 is interrupted halfway through, the next run will pick up where you left
49 off. And it's easy to set up a bup server: just install bup on any
50 machine where you have ssh access.
52 - Bup can use "par2" redundancy to recover corrupted backups even if your
53 disk has undetected bad sectors.
55 - Even when a backup is incremental, you don't have to worry about
56 restoring the full backup, then each of the incrementals in turn; an
57 incremental backup *acts* as if it's a full backup, it just takes less
60 - You can mount your bup repository as a FUSE filesystem and access the
61 content that way, and even export it over Samba.
63 - It's written in python (with some C parts to make it faster) so it's easy
64 for you to extend and maintain.
67 Reasons you might want to avoid bup
68 -----------------------------------
70 - This is a very early version. Therefore it will most probably not work
71 for you, but we don't know why. It is also missing some
72 probably-critical features.
74 - It requires python >= 2.6, a C compiler, and an installed git
75 version >= 1.5.3.1. It also requires par2 if you want fsck to be
76 able to generate the information needed to recover from some types
79 - It currently only works on Linux, FreeBSD, NetBSD, OS X >= 10.4,
80 Solaris, or Windows (with Cygwin). Patches to support other
81 platforms are welcome.
83 - Any items in "Things that are stupid" below.
86 Notable changes introduced by a release
87 =======================================
89 - <a href="note/0.29.1-from-0.29.md">Changes in 0.29.1 as compared to 0.29</a>
90 - <a href="note/0.29-from-0.28.1.md">Changes in 0.29 as compared to 0.28.1</a>
91 - <a href="note/0.28.1-from-0.28.md">Changes in 0.28.1 as compared to 0.28</a>
92 - <a href="note/0.28-from-0.27.1.md">Changes in 0.28 as compared to 0.27.1</a>
93 - <a href="note/0.27.1-from-0.27.md">Changes in 0.27.1 as compared to 0.27</a>
102 - Check out the bup source code using git:
104 git clone https://github.com/bup/bup
106 - Install the required python libraries (including the development
109 On very recent Debian/Ubuntu versions, this may be sufficient (run
112 apt-get build-dep bup
114 Otherwise try this (substitute python2.6-dev if you have an older
117 apt-get install python2.7-dev python-fuse
118 apt-get install python-pyxattr python-pylibacl
119 apt-get install linux-libc-dev
120 apt-get install acl attr
121 apt-get install python-tornado # optional
123 On CentOS (for CentOS 6, at least), this should be sufficient (run
126 yum groupinstall "Development Tools"
127 yum install python python-devel
128 yum install fuse-python pyxattr pylibacl
129 yum install perl-Time-HiRes
131 In addition to the default CentOS repositories, you may need to add
132 RPMForge (for fuse-python) and EPEL (for pyxattr and pylibacl).
134 On Cygwin, install python, make, rsync, and gcc4.
136 If you would like to use the optional bup web server on systems
137 without a tornado package, you may want to try this:
141 - Build the python module and symlinks:
149 The tests should pass. If they don't pass for you, stop here and
150 send an email to bup-list@googlegroups.com. Though if there are
151 symbolic links along the current working directory path, the tests
152 may fail. Running something like this before "make test" should
153 sidestep the problem:
157 - You can install bup via "make install", and override the default
158 destination with DESTDIR and PREFIX.
160 Files are normally installed to "$DESTDIR/$PREFIX" where DESTDIR is
161 empty by default, and PREFIX is set to /usr/local. So if you wanted to
162 install bup to /opt/bup, you might do something like this:
164 make install DESTDIR=/opt/bup PREFIX=''
166 - The Python executable that bup will use is chosen by ./configure,
167 which will search for a reasonable version unless PYTHON is set in
168 the environment, in which case, bup will use that path. You can
169 see which Python executable was chosen by looking at the
170 configure output, or examining cmd/python-cmd.sh, and you can
171 change the selection by re-running ./configure.
176 Binary packages of bup are known to be built for the following OSes:
179 http://packages.debian.org/search?searchon=names&keywords=bup
181 http://packages.ubuntu.com/search?searchon=names&keywords=bup
182 - pkgsrc (NetBSD, Dragonfly, and others)
183 http://pkgsrc.se/sysutils/bup
184 http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/sysutils/bup/
186 https://www.archlinux.org/packages/?sort=&q=bup
188 https://apps.fedoraproject.org/packages/bup
194 - Get help for any bup command:
203 - Initialize the default BUP_DIR (~/.bup):
207 - Make a local backup (-v or -vv will increase the verbosity):
210 bup save -n local-etc /etc
212 - Restore a local backup to ./dest:
214 bup restore -C ./dest local-etc/latest/etc
217 - Look at how much disk space your backup took:
221 - Make another backup (which should be mostly identical to the last one;
222 notice that you don't have to *specify* that this backup is incremental,
223 it just saves space automatically):
226 bup save -n local-etc /etc
228 - Look how little extra space your second backup used (on top of the first):
232 - Get a list of your previous backups:
236 - Restore your first backup again:
238 bup restore -C ./dest-2 local-etc/2013-11-23-11195/etc
240 - Make a backup to a remote server which must already have the 'bup' command
241 somewhere in its PATH (see /etc/profile, etc/environment, ~/.profile, or
242 ~/.bashrc), and be accessible via ssh.
243 Make sure to replace SERVERNAME with the actual hostname of your server:
245 bup init -r SERVERNAME:path/to/remote-bup-dir
247 bup save -r SERVERNAME:path/to/remote-bup-dir -n local-etc /etc
249 - Restore a backup from a remote server. (FAIL: unfortunately,
250 unlike "bup join", "bup restore" does not yet support remote
251 restores. See both "bup join" and "Things that are stupid" below.)
253 - Defend your backups from death rays (OK fine, more likely from the
254 occasional bad disk block). This writes parity information
255 (currently via par2) for all of the existing data so that bup may
256 be able to recover from some amount of repository corruption:
260 - Use split/join instead of index/save/restore. Try making a local
263 tar -cvf - /etc | bup split -n local-etc -vv
265 - Try restoring the tarball:
267 bup join local-etc | tar -tf -
269 - Look at how much disk space your backup took:
273 - Make another tar backup:
275 tar -cvf - /etc | bup split -n local-etc -vv
277 - Look at how little extra space your second backup used on top of
282 - Restore the first tar backup again (the ~1 is git notation for "one
283 older than the most recent"):
285 bup join local-etc~1 | tar -tf -
287 - Get a list of your previous split-based backups:
289 GIT_DIR=~/.bup git log local-etc
291 - Make a backup on a remote server:
293 tar -cvf - /etc | bup split -r SERVERNAME: -n local-etc -vv
295 - Try restoring the remote backup tarball:
297 bup join -r SERVERNAME: local-etc | tar -tf -
299 That's all there is to it!
305 - FreeBSD's default 'make' command doesn't like bup's Makefile. In order to
306 compile the code, run tests and install bup, you need to install GNU Make
307 from the port named 'gmake' and use its executable instead in the commands
308 seen above. (i.e. 'gmake test' runs bup's test suite)
310 - Python's development headers are automatically installed with the 'python'
311 port so there's no need to install them separately.
313 - To use the 'bup fuse' command, you need to install the fuse kernel module
314 from the 'fusefs-kmod' port in the 'sysutils' section and the libraries from
315 the port named 'py-fusefs' in the 'devel' section.
317 - The 'par2' command can be found in the port named 'par2cmdline'.
319 - In order to compile the documentation, you need pandoc which can be found in
320 the port named 'hs-pandoc' in the 'textproc' section.
323 Notes on NetBSD/pkgsrc
324 ----------------------
326 - See pkgsrc/sysutils/bup, which should be the most recent stable
327 release and includes man pages. It also has a reasonable set of
328 dependencies (git, par2, py-fuse-bindings).
330 - The "fuse-python" package referred to is hard to locate, and is a
331 separate tarball for the python language binding distributed by the
332 fuse project on sourceforge. It is available as
333 pkgsrc/filesystems/py-fuse-bindings and on NetBSD 5, "bup fuse"
336 - "bup fuse" presents every directory/file as inode 0. The directory
337 traversal code ("fts") in NetBSD's libc will interpret this as a
338 cycle and error out, so "ls -R" and "find" will not work.
340 - There is no support for ACLs. If/when some entrprising person
341 fixes this, adjust t/compare-trees.
347 - There is no support for ACLs. If/when some enterprising person
348 fixes this, adjust t/compare-trees.
350 - In t/test.sh, two tests have been disabled. These tests check to
351 see that repeated saves produce identical trees and that an
352 intervening index doesn't change the SHA1. Apparently Cygwin has
353 some unusual behaviors with respect to access times (that probably
354 warrant further investigation). Possibly related:
355 http://cygwin.com/ml/cygwin/2007-06/msg00436.html
361 - There is no support for ACLs. If/when some enterprising person
362 fixes this, adjust t/compare-trees.
371 bup stores its data in a git-formatted repository. Unfortunately, git
372 itself doesn't actually behave very well for bup's use case (huge numbers of
373 files, files with huge sizes, retaining file permissions/ownership are
374 important), so we mostly don't use git's *code* except for a few helper
375 programs. For example, bup has its own git packfile writer written in
378 Basically, 'bup split' reads the data on stdin (or from files specified on
379 the command line), breaks it into chunks using a rolling checksum (similar to
380 rsync), and saves those chunks into a new git packfile. There is at least one
381 git packfile per backup.
383 When deciding whether to write a particular chunk into the new packfile, bup
384 first checks all the other packfiles that exist to see if they already have that
385 chunk. If they do, the chunk is skipped.
387 git packs come in two parts: the pack itself (*.pack) and the index (*.idx).
388 The index is pretty small, and contains a list of all the objects in the
389 pack. Thus, when generating a remote backup, we don't have to have a copy
390 of the packfiles from the remote server: the local end just downloads a copy
391 of the server's *index* files, and compares objects against those when
392 generating the new pack, which it sends directly to the server.
394 The "-n" option to 'bup split' and 'bup save' is the name of the backup you
395 want to create, but it's actually implemented as a git branch. So you can
396 do cute things like checkout a particular branch using git, and receive a
397 bunch of chunk files corresponding to the file you split.
399 If you use '-b' or '-t' or '-c' instead of '-n', bup split will output a
400 list of blobs, a tree containing that list of blobs, or a commit containing
401 that tree, respectively, to stdout. You can use this to construct your own
402 scripts that do something with those values.
407 'bup index' walks through your filesystem and updates a file (whose name is,
408 by default, ~/.bup/bupindex) to contain the name, attributes, and an
409 optional git SHA1 (blob id) of each file and directory.
411 'bup save' basically just runs the equivalent of 'bup split' a whole bunch
412 of times, once per file in the index, and assembles a git tree
413 that contains all the resulting objects. Among other things, that makes
414 'git diff' much more useful (compared to splitting a tarball, which is
415 essentially a big binary blob). However, since bup splits large files into
416 smaller chunks, the resulting tree structure doesn't *exactly* correspond to
417 what git itself would have stored. Also, the tree format used by 'bup save'
418 will probably change in the future to support storing file ownership, more
419 complex file permissions, and so on.
421 If a file has previously been written by 'bup save', then its git blob/tree
422 id is stored in the index. This lets 'bup save' avoid reading that file to
423 produce future incremental backups, which means it can go *very* fast unless
424 a lot of files have changed.
427 Things that are stupid for now but which we'll fix later
428 ========================================================
430 Help with any of these problems, or others, is very welcome. Join the
431 mailing list (see below) if you'd like to help.
433 - 'bup restore' can't pull directly from a remote server.
435 So in one sense "save -r" is a dead-end right now. Obviously you
436 can use "ssh SERVER bup restore -C ./dest..." to create a tree you
437 can transfer elsewhere via rsync/tar/whatever, but that's *lame*.
439 Until we fix it, you may be able to mount the remote BUP_DIR via
440 sshfs and then restore "normally", though that hasn't been
443 - 'bup save' and 'bup restore' have immature metadata support.
445 On the plus side, they actually do have support now, but it's new,
446 and not remotely as well tested as tar/rsync/whatever's. However,
447 you have to start somewhere, and as of 0.25, we think it's ready
448 for more general use. Please let us know if you have any trouble.
450 Also, if any strip or graft-style options are specified to 'bup
451 save', then no metadata will be written for the root directory.
452 That's obviously less than ideal.
454 - bup is overly optimistic about mmap. Right now bup just assumes
455 that it can mmap as large a block as it likes, and that mmap will
456 never fail. Yeah, right... If nothing else, this has failed on
457 32-bit architectures (and 31-bit is even worse -- looking at you,
460 To fix this, we might just implement a FakeMmap[1] class that uses
461 normal file IO and handles all of the mmap methods[2] that bup
462 actually calls. Then we'd swap in one of those whenever mmap
465 This would also require implementing some of the methods needed to
466 support "[]" array access, probably at a minimum __getitem__,
467 __setitem__, and __setslice__ [3].
469 [1] http://comments.gmane.org/gmane.comp.sysutils.backup.bup/613
470 [2] http://docs.python.org/2/library/mmap.html
471 [3] http://docs.python.org/2/reference/datamodel.html#emulating-container-types
473 - 'bup index' is slower than it should be.
475 It's still rather fast: it can iterate through all the filenames on my
476 600,000 file filesystem in a few seconds. But it still needs to rewrite
477 the entire index file just to add a single filename, which is pretty
478 nasty; it should just leave the new files in a second "extra index" file
481 - bup could use inotify for *really* efficient incremental backups.
483 You could even have your system doing "continuous" backups: whenever a
484 file changes, we immediately send an image of it to the server. We could
485 give the continuous-backup process a really low CPU and I/O priority so
486 you wouldn't even know it was running.
488 - bup only has experimental support for pruning old backups.
490 While you should now be able to drop old saves and branches with
491 `bup rm`, and reclaim the space occupied by data that's no longer
492 needed by other backups with `bup gc`, these commands are
493 experimental, and should be handled with great care. See the
494 man pages for more information.
496 Unless you want to help test the new commands, one possible
497 workaround is to just start a new BUP_DIR occasionally,
498 i.e. bup-2013, bup-2014...
500 - bup has never been tested on anything but Linux, FreeBSD, NetBSD,
501 OS X, and Windows+Cygwin.
503 There's nothing that makes it *inherently* non-portable, though, so
504 that's mostly a matter of someone putting in some effort. (For a
505 "native" Windows port, the most annoying thing is the absence of ssh in
506 a default Windows installation.)
508 - bup needs better documentation.
510 According to a recent article about bup in Linux Weekly News
511 (https://lwn.net/Articles/380983/), "it's a bit short on examples and
512 a user guide would be nice." Documentation is the sort of thing that
513 will never be great unless someone from outside contributes it (since
514 the developers can never remember which parts are hard to understand).
516 - bup is "relatively speedy" and has "pretty good" compression.
518 ...according to the same LWN article. Clearly neither of those is good
519 enough. We should have awe-inspiring speed and crazy-good compression.
520 Must work on that. Writing more parts in C might help with the speed.
524 Actually, that's not stupid, but you might consider it a
525 limitation. See the ["Related Projects"](https://bup.github.io/)
526 list for some possible options.
531 bup has an extensive set of man pages. Try using 'bup help' to get
532 started, or use 'bup help SUBCOMMAND' for any bup subcommand (like split,
533 join, index, save, etc.) to get details on that command.
535 For further technical details, please see ./DESIGN.
541 bup is a work in progress and there are many ways it can still be improved.
542 If you'd like to contribute patches, ideas, or bug reports, please join the
545 You can find the mailing list archives here:
547 http://groups.google.com/group/bup-list
549 and you can subscribe by sending a message to:
551 bup-list+subscribe@googlegroups.com
553 Please see <a href="HACKING">./HACKING</a> for
554 additional information, i.e. how to submit patches (hint - no pull
555 requests), how we handle branches, etc.