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 - It's not remotely as well tested as something like tar, so it's
71 more likely to eat your data. It's also missing some
72 probably-critical features, though fewer than it used to be.
74 - It requires python >= 2.6, a C compiler, and an installed git
75 version >= 1.5.6. 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, and maybe with WSL). Patches to
81 support other platforms are welcome.
83 - Any items in "Things that are stupid" below.
86 Notable changes introduced by a release
87 =======================================
89 - <a href="note/0.29.2-from-0.29.1.md">Changes in 0.29.2 as compared to 0.29.1</a>
90 - <a href="note/0.29.1-from-0.29.md">Changes in 0.29.1 as compared to 0.29</a>
91 - <a href="note/0.29-from-0.28.1.md">Changes in 0.29 as compared to 0.28.1</a>
92 - <a href="note/0.28.1-from-0.28.md">Changes in 0.28.1 as compared to 0.28</a>
93 - <a href="note/0.28-from-0.27.1.md">Changes in 0.28 as compared to 0.27.1</a>
94 - <a href="note/0.27.1-from-0.27.md">Changes in 0.27.1 as compared to 0.27</a>
103 - Check out the bup source code using git:
105 git clone https://github.com/bup/bup
107 - This will leave you on the master branch, which is perfect if you
108 would like to help with development, but if you'd just like to use
109 bup, please check out the latest stable release like this:
113 You can see the latest stable release here:
114 https://github.com/bup/bup/releases.
116 - Install the required python libraries (including the development
119 On very recent Debian/Ubuntu versions, this may be sufficient (run
122 apt-get build-dep bup
124 Otherwise try this (substitute python2.6-dev if you have an older
127 apt-get install python2.7-dev python-fuse
128 apt-get install python-pyxattr python-pylibacl
129 apt-get install linux-libc-dev
130 apt-get install acl attr
131 apt-get install python-tornado # optional
133 On CentOS (for CentOS 6, at least), this should be sufficient (run
136 yum groupinstall "Development Tools"
137 yum install python python-devel
138 yum install fuse-python pyxattr pylibacl
139 yum install perl-Time-HiRes
141 In addition to the default CentOS repositories, you may need to add
142 RPMForge (for fuse-python) and EPEL (for pyxattr and pylibacl).
144 On Cygwin, install python, make, rsync, and gcc4.
146 If you would like to use the optional bup web server on systems
147 without a tornado package, you may want to try this:
151 - Build the python module and symlinks:
159 or if you're in a bit more of a hurry:
163 The tests should pass. If they don't pass for you, stop here and
164 send an email to bup-list@googlegroups.com. Though if there are
165 symbolic links along the current working directory path, the tests
166 may fail. Running something like this before "make test" should
167 sidestep the problem:
171 - You can install bup via "make install", and override the default
172 destination with DESTDIR and PREFIX.
174 Files are normally installed to "$DESTDIR/$PREFIX" where DESTDIR is
175 empty by default, and PREFIX is set to /usr/local. So if you wanted to
176 install bup to /opt/bup, you might do something like this:
178 make install DESTDIR=/opt/bup PREFIX=''
180 - The Python executable that bup will use is chosen by ./configure,
181 which will search for a reasonable version unless PYTHON is set in
182 the environment, in which case, bup will use that path. You can
183 see which Python executable was chosen by looking at the
184 configure output, or examining cmd/python-cmd.sh, and you can
185 change the selection by re-running ./configure.
190 Binary packages of bup are known to be built for the following OSes:
193 http://packages.debian.org/search?searchon=names&keywords=bup
195 http://packages.ubuntu.com/search?searchon=names&keywords=bup
196 - pkgsrc (NetBSD, Dragonfly, and others)
197 http://pkgsrc.se/sysutils/bup
198 http://cvsweb.netbsd.org/bsdweb.cgi/pkgsrc/sysutils/bup/
200 https://www.archlinux.org/packages/?sort=&q=bup
202 https://apps.fedoraproject.org/packages/bup
208 - Get help for any bup command:
217 - Initialize the default BUP_DIR (~/.bup -- you can choose another by
218 either specifying `bup -d DIR ...` or setting the `BUP_DIR`
219 environment variable for a command):
223 - Make a local backup (-v or -vv will increase the verbosity):
226 bup save -n local-etc /etc
228 - Restore a local backup to ./dest:
230 bup restore -C ./dest local-etc/latest/etc
233 - Look at how much disk space your backup took:
237 - Make another backup (which should be mostly identical to the last one;
238 notice that you don't have to *specify* that this backup is incremental,
239 it just saves space automatically):
242 bup save -n local-etc /etc
244 - Look how little extra space your second backup used (on top of the first):
248 - Get a list of your previous backups:
252 - Restore your first backup again:
254 bup restore -C ./dest-2 local-etc/2013-11-23-11195/etc
256 - Make a backup to a remote server which must already have the 'bup' command
257 somewhere in its PATH (see /etc/profile, etc/environment, ~/.profile, or
258 ~/.bashrc), and be accessible via ssh.
259 Make sure to replace SERVERNAME with the actual hostname of your server:
261 bup init -r SERVERNAME:path/to/remote-bup-dir
263 bup save -r SERVERNAME:path/to/remote-bup-dir -n local-etc /etc
265 - Make a remote backup to ~/.bup on SERVER:
268 bup save -r SERVER: -n local-etc /etc
270 - See what saves are available in ~/.bup on SERVER:
274 - Restore the remote backup to ./dest:
276 bup restore -r SERVER: -C ./dest local-etc/latest/etc
279 - Defend your backups from death rays (OK fine, more likely from the
280 occasional bad disk block). This writes parity information
281 (currently via par2) for all of the existing data so that bup may
282 be able to recover from some amount of repository corruption:
286 - Use split/join instead of index/save/restore. Try making a local
289 tar -cvf - /etc | bup split -n local-etc -vv
291 - Try restoring the tarball:
293 bup join local-etc | tar -tf -
295 - Look at how much disk space your backup took:
299 - Make another tar backup:
301 tar -cvf - /etc | bup split -n local-etc -vv
303 - Look at how little extra space your second backup used on top of
308 - Restore the first tar backup again (the ~1 is git notation for "one
309 older than the most recent"):
311 bup join local-etc~1 | tar -tf -
313 - Get a list of your previous split-based backups:
315 GIT_DIR=~/.bup git log local-etc
317 - Save a tar archive to a remote server (without tar -z to facilitate
320 tar -cvf - /etc | bup split -r SERVERNAME: -n local-etc -vv
322 - Restore the archive:
324 bup join -r SERVERNAME: local-etc | tar -tf -
326 That's all there is to it!
332 - FreeBSD's default 'make' command doesn't like bup's Makefile. In order to
333 compile the code, run tests and install bup, you need to install GNU Make
334 from the port named 'gmake' and use its executable instead in the commands
335 seen above. (i.e. 'gmake test' runs bup's test suite)
337 - Python's development headers are automatically installed with the 'python'
338 port so there's no need to install them separately.
340 - To use the 'bup fuse' command, you need to install the fuse kernel module
341 from the 'fusefs-kmod' port in the 'sysutils' section and the libraries from
342 the port named 'py-fusefs' in the 'devel' section.
344 - The 'par2' command can be found in the port named 'par2cmdline'.
346 - In order to compile the documentation, you need pandoc which can be found in
347 the port named 'hs-pandoc' in the 'textproc' section.
350 Notes on NetBSD/pkgsrc
351 ----------------------
353 - See pkgsrc/sysutils/bup, which should be the most recent stable
354 release and includes man pages. It also has a reasonable set of
355 dependencies (git, par2, py-fuse-bindings).
357 - The "fuse-python" package referred to is hard to locate, and is a
358 separate tarball for the python language binding distributed by the
359 fuse project on sourceforge. It is available as
360 pkgsrc/filesystems/py-fuse-bindings and on NetBSD 5, "bup fuse"
363 - "bup fuse" presents every directory/file as inode 0. The directory
364 traversal code ("fts") in NetBSD's libc will interpret this as a
365 cycle and error out, so "ls -R" and "find" will not work.
367 - There is no support for ACLs. If/when some entrprising person
368 fixes this, adjust t/compare-trees.
374 - There is no support for ACLs. If/when some enterprising person
375 fixes this, adjust t/compare-trees.
377 - In t/test.sh, two tests have been disabled. These tests check to
378 see that repeated saves produce identical trees and that an
379 intervening index doesn't change the SHA1. Apparently Cygwin has
380 some unusual behaviors with respect to access times (that probably
381 warrant further investigation). Possibly related:
382 http://cygwin.com/ml/cygwin/2007-06/msg00436.html
388 - There is no support for ACLs. If/when some enterprising person
389 fixes this, adjust t/compare-trees.
398 bup stores its data in a git-formatted repository. Unfortunately, git
399 itself doesn't actually behave very well for bup's use case (huge numbers of
400 files, files with huge sizes, retaining file permissions/ownership are
401 important), so we mostly don't use git's *code* except for a few helper
402 programs. For example, bup has its own git packfile writer written in
405 Basically, 'bup split' reads the data on stdin (or from files specified on
406 the command line), breaks it into chunks using a rolling checksum (similar to
407 rsync), and saves those chunks into a new git packfile. There is at least one
408 git packfile per backup.
410 When deciding whether to write a particular chunk into the new packfile, bup
411 first checks all the other packfiles that exist to see if they already have that
412 chunk. If they do, the chunk is skipped.
414 git packs come in two parts: the pack itself (*.pack) and the index (*.idx).
415 The index is pretty small, and contains a list of all the objects in the
416 pack. Thus, when generating a remote backup, we don't have to have a copy
417 of the packfiles from the remote server: the local end just downloads a copy
418 of the server's *index* files, and compares objects against those when
419 generating the new pack, which it sends directly to the server.
421 The "-n" option to 'bup split' and 'bup save' is the name of the backup you
422 want to create, but it's actually implemented as a git branch. So you can
423 do cute things like checkout a particular branch using git, and receive a
424 bunch of chunk files corresponding to the file you split.
426 If you use '-b' or '-t' or '-c' instead of '-n', bup split will output a
427 list of blobs, a tree containing that list of blobs, or a commit containing
428 that tree, respectively, to stdout. You can use this to construct your own
429 scripts that do something with those values.
434 'bup index' walks through your filesystem and updates a file (whose name is,
435 by default, ~/.bup/bupindex) to contain the name, attributes, and an
436 optional git SHA1 (blob id) of each file and directory.
438 'bup save' basically just runs the equivalent of 'bup split' a whole bunch
439 of times, once per file in the index, and assembles a git tree
440 that contains all the resulting objects. Among other things, that makes
441 'git diff' much more useful (compared to splitting a tarball, which is
442 essentially a big binary blob). However, since bup splits large files into
443 smaller chunks, the resulting tree structure doesn't *exactly* correspond to
444 what git itself would have stored. Also, the tree format used by 'bup save'
445 will probably change in the future to support storing file ownership, more
446 complex file permissions, and so on.
448 If a file has previously been written by 'bup save', then its git blob/tree
449 id is stored in the index. This lets 'bup save' avoid reading that file to
450 produce future incremental backups, which means it can go *very* fast unless
451 a lot of files have changed.
454 Things that are stupid for now but which we'll fix later
455 ========================================================
457 Help with any of these problems, or others, is very welcome. Join the
458 mailing list (see below) if you'd like to help.
460 - 'bup save' and 'bup restore' have immature metadata support.
462 On the plus side, they actually do have support now, but it's new,
463 and not remotely as well tested as tar/rsync/whatever's. However,
464 you have to start somewhere, and as of 0.25, we think it's ready
465 for more general use. Please let us know if you have any trouble.
467 Also, if any strip or graft-style options are specified to 'bup
468 save', then no metadata will be written for the root directory.
469 That's obviously less than ideal.
471 - bup is overly optimistic about mmap. Right now bup just assumes
472 that it can mmap as large a block as it likes, and that mmap will
473 never fail. Yeah, right... If nothing else, this has failed on
474 32-bit architectures (and 31-bit is even worse -- looking at you,
477 To fix this, we might just implement a FakeMmap[1] class that uses
478 normal file IO and handles all of the mmap methods[2] that bup
479 actually calls. Then we'd swap in one of those whenever mmap
482 This would also require implementing some of the methods needed to
483 support "[]" array access, probably at a minimum __getitem__,
484 __setitem__, and __setslice__ [3].
486 [1] http://comments.gmane.org/gmane.comp.sysutils.backup.bup/613
487 [2] http://docs.python.org/2/library/mmap.html
488 [3] http://docs.python.org/2/reference/datamodel.html#emulating-container-types
490 - 'bup index' is slower than it should be.
492 It's still rather fast: it can iterate through all the filenames on my
493 600,000 file filesystem in a few seconds. But it still needs to rewrite
494 the entire index file just to add a single filename, which is pretty
495 nasty; it should just leave the new files in a second "extra index" file
498 - bup could use inotify for *really* efficient incremental backups.
500 You could even have your system doing "continuous" backups: whenever a
501 file changes, we immediately send an image of it to the server. We could
502 give the continuous-backup process a really low CPU and I/O priority so
503 you wouldn't even know it was running.
505 - bup only has experimental support for pruning old backups.
507 While you should now be able to drop old saves and branches with
508 `bup rm`, and reclaim the space occupied by data that's no longer
509 needed by other backups with `bup gc`, these commands are
510 experimental, and should be handled with great care. See the
511 man pages for more information.
513 Unless you want to help test the new commands, one possible
514 workaround is to just start a new BUP_DIR occasionally,
515 i.e. bup-2013, bup-2014...
517 - bup has never been tested on anything but Linux, FreeBSD, NetBSD,
518 OS X, and Windows+Cygwin.
520 There's nothing that makes it *inherently* non-portable, though, so
521 that's mostly a matter of someone putting in some effort. (For a
522 "native" Windows port, the most annoying thing is the absence of ssh in
523 a default Windows installation.)
525 - bup needs better documentation.
527 According to an article about bup in Linux Weekly News
528 (https://lwn.net/Articles/380983/), "it's a bit short on examples and
529 a user guide would be nice." Documentation is the sort of thing that
530 will never be great unless someone from outside contributes it (since
531 the developers can never remember which parts are hard to understand).
533 - bup is "relatively speedy" and has "pretty good" compression.
535 ...according to the same LWN article. Clearly neither of those is good
536 enough. We should have awe-inspiring speed and crazy-good compression.
537 Must work on that. Writing more parts in C might help with the speed.
541 Actually, that's not stupid, but you might consider it a
542 limitation. See the ["Related Projects"](https://bup.github.io/)
543 list for some possible options.
548 bup has an extensive set of man pages. Try using 'bup help' to get
549 started, or use 'bup help SUBCOMMAND' for any bup subcommand (like split,
550 join, index, save, etc.) to get details on that command.
552 For further technical details, please see ./DESIGN.
558 bup is a work in progress and there are many ways it can still be improved.
559 If you'd like to contribute patches, ideas, or bug reports, please join the
562 You can find the mailing list archives here:
564 http://groups.google.com/group/bup-list
566 and you can subscribe by sending a message to:
568 bup-list+subscribe@googlegroups.com
570 Please see <a href="HACKING">./HACKING</a> for
571 additional information, i.e. how to submit patches (hint - no pull
572 requests), how we handle branches, etc.