2 bup 0.04: It backs things up
3 ============================
5 bup is a program that backs things up. It's short for "backup." Can you
6 believe that nobody else has named an open source program "bup" after all
9 Despite its unassuming name, bup is pretty cool. To give you an idea of
10 just how cool it is, I wrote you this poem:
13 What rhymes with awesome?
15 But that's irrelevant.
17 Hmm. Did that help? Maybe prose is more useful after all.
20 Reasons bup is awesome
21 ----------------------
23 bup has a few advantages over other backup software:
25 - It uses a rolling checksum algorithm (similar to rsync) to split large
26 files into chunks. The most useful result of this is you can backup huge
27 virtual machine (VM) disk images, databases, and XML files incrementally,
28 even though they're typically all in one huge file, and not use tons of
29 disk space for multiple versions.
31 - It uses the packfile format from git (the open source version control
32 system), so you can access the stored data even if you don't like bup's
35 - Unlike git, it writes packfiles *directly* (instead of having a separate
36 garbage collection / repacking stage) so it's fast even with gratuitously
39 - Data is "automagically" shared between incremental backups without having
40 to know which backup is based on which other one - even if the backups
41 are made from two different computers that don't even know about each
42 other. You just tell bup to back stuff up, and it saves only the minimum
43 amount of data needed.
45 - Even when a backup is incremental, you don't have to worry about
46 restoring the full backup, then each of the incrementals in turn; an
47 incremental backup *acts* as if it's a full backup, it just takes less
50 - It's written in python (with some C parts to make it faster) so it's easy
51 for you to extend and maintain.
54 Reasons you might want to avoid bup
55 -----------------------------------
57 - This is a very early version. Therefore it will most probably not work
58 for you, but we don't know why. It is also missing some
59 probably-critical features.
61 - It requires python 2.5, a C compiler, and an installed git version >= 1.5.2.
63 - It currently only works on Linux, MacOS X 10.5, or Windows (with Cygwin).
64 Patches to support other platforms are welcome.
66 - It has almost no documentation. Not even a man page! This file is all
73 - check out the bup source code using git:
75 git clone git://github.com/apenwarr/bup
77 - install the python 2.5 development libraries. On Debian or Ubuntu, this
79 apt-get install python2.5-dev
81 - build the python module and symlinks:
89 (The tests should pass. If they don't pass for you, stop here and send
92 - Try making a local backup as a tar file:
94 tar -cvf - /etc | bup split -n local-etc -vv
96 - Try restoring your backup tarball:
98 bup join local-etc | tar -tf -
100 - Look at how much disk space your backup took:
104 - Make another backup (which should be mostly identical to the last one;
105 notice that you don't have to *specify* that this backup is incremental,
106 it just saves space automatically):
108 tar -cvf - /etc | bup split -n local-etc -vv
110 - Look how little extra space your second backup used on top of the first:
114 - Restore your old backup again (the ~1 is git notation for "one older than
117 bup join local-etc~1 | tar -tf -
119 - get a list of your previous backups:
121 GIT_DIR=~/.bup git log local-etc
123 - make a backup on a remote server (which must already have the 'bup' command
124 somewhere in the PATH, and be accessible via ssh; make sure to replace
125 SERVERNAME with the actual hostname of your server):
127 tar -cvf - /etc | bup split -r SERVERNAME: -n local-etc -vv
129 - try restoring the remote backup tarball:
131 bup join -r SERVERNAME: local-etc | tar -tf -
133 - try using the new (slightly experimental) 'bup index' and 'bup save'
134 style backups, which bypass 'tar' but have some missing features (see
135 "Things that are stupid" below):
138 bup save -n local-etc /etc
140 - do it again and see how fast an incremental backup can be:
143 bup save -n local-etc /etc
145 (You can also use the "-r SERVERNAME:" option to 'bup save', just like
146 with 'bup split' and 'bup join'. The index itself is always local,
147 so you don't need -r there.)
149 That's all there is to it!
157 bup stores its data in a git-formatted repository. Unfortunately, git
158 itself doesn't actually behave very well for bup's use case (huge numbers of
159 files, files with huge sizes, retaining file permissions/ownership are
160 important), so we mostly don't use git's *code* except for a few helper
161 programs. For example, bup has its own git packfile writer written in
164 Basically, 'bup split' reads the data on stdin (or from files specified on
165 the command line), breaks it into chunks using a rolling checksum (similar to
166 rsync), and saves those chunks into a new git packfile. There is one git
169 When deciding whether to write a particular chunk into the new packfile, bup
170 first checks all the other packfiles that exist to see if they already have that
171 chunk. If they do, the chunk is skipped.
173 git packs come in two parts: the pack itself (*.pack) and the index (*.idx).
174 The index is pretty small, and contains a list of all the objects in the
175 pack. Thus, when generating a remote backup, we don't have to have a copy
176 of the packfiles from the remote server: the local end just downloads a copy
177 of the server's *index* files, and compares objects against those when
178 generating the new pack, which it sends directly to the server.
180 The "-n" option to 'bup split' and 'bup save' is the name of the backup you
181 want to create, but it's actually implemented as a git branch. So you can
182 do cute things like checkout a particular branch using git, and receive a
183 bunch of chunk files corresponding to the file you split.
185 If you use '-b' or '-t' or '-c' instead of '-n', bup split will output a
186 list of blobs, a tree containing that list of blobs, or a commit containing
187 that tree, respectively, to stdout. You can use this to construct your own
188 scripts that do something with those values.
192 'bup index' walks through your filesystem and updates a file (whose name is,
193 by default, ~/.bup/bupindex) to contain the name, attributes, and an
194 optional git SHA1 (blob id) of each file and directory.
196 'bup save' basically just runs the equivalent of 'bup split' a whole bunch
197 of times, once per file in the index, and assembles a git tree
198 that contains all the resulting objects. Among other things, that makes
199 'git diff' much more useful (compared to splitting a tarball, which is
200 essentially a big binary blob). However, since bup splits large files into
201 smaller chunks, the resulting tree structure doesn't *exactly* correspond to
202 what git itself would have stored. Also, the tree format used by 'bup save'
203 will probably change in the future to support storing file ownership, more
204 complex file permissions, and so on.
206 If a file has previously been written by 'bup save', then its git blob/tree
207 id is stored in the index. This lets 'bup save' avoid reading that file to
208 produce future incremental backups, which means it can go *very* fast unless
209 a lot of files have changed.
212 Things that are stupid for now but which we'll fix later
213 --------------------------------------------------------
215 Help with any of these problems, or others, is very, very welcome. Let me
216 know if you'd like to help. Maybe we can start a mailing list.
218 - 'bup save' doesn't know about file metadata.
220 That means we aren't saving file attributes, mtimes, ownership, hard
221 links, MacOS resource forks, etc. Clearly this needs to be improved.
223 - There's no 'bup restore' yet.
225 'bup save' saves files in the standard git 'tree of blobs' format, so you
226 could then "restore" the files using something like 'git checkout'. But
227 that's a git command, not a bup command, so it's hard to explain and
228 doesn't support retrieving objects from a remote bup server without first
229 fetching and packing an entire (possibly huge) pack, which could be very
230 slow. Also, like 'bup save', you would need extra features in order to
231 properly restore file metadata. And files that bup has split into
232 chunks would need to be recombined somehow.
234 - 'bup index' is slower than it should be.
236 It's still rather fast: it can iterate through all the filenames on my
237 600,000 file filesystem in a few seconds. But sometimes you just want to
238 change a filename or two, so this is needlessly slow. There should be
239 a way to binary search through the file list rather than always going
240 through it sequentially. And if you only add a couple of filenames,
241 there's no need to rewrite the entire index; just leave the new files
242 in a second "extra index" file or something.
244 - bup could use inotify for *really* efficient incremental backups.
246 You could even have your system doing "continuous" backups: whenever a
247 file changes, we immediately send an image of it to the server. We could
248 give the continuous-backup process a really low CPU and I/O priority so
249 you wouldn't even know it was running.
251 - bup currently has no features that prune away *old* backups.
253 Because of the way the packfile system works, backups become "entangled"
254 in weird ways and it's not actually possible to delete one pack
255 (corresponding approximately to one backup) without risking screwing up
258 git itself has lots of ways of optimizing this sort of thing, but its
259 methods aren't really applicable here; bup packfiles are just too huge.
260 We'll have to do it in a totally different way. There are lots of
261 options. For now: make sure you've got lots of disk space :)
263 - bup doesn't ever validate existing backups/packs to ensure they're
266 This would be easy to implement (given that git uses hashes and CRCs all
267 over the place), but nobody has implemented it. For now, you could try
268 doing a test restore of your tarball; doing so should trigger git's error
269 handling if any of the objects are corrupted. 'git fsck' would
270 theoreticaly work too, but it's too slow for huge backups.
272 - bup has never been tested on anything but Linux, MacOS, and Linux+Cygwin.
274 There's nothing that makes it *inherently* non-portable, though, so
275 that's mostly a matter of someone putting in some effort. (For a
276 "native" Windows port, the most annoying thing is the absence of ssh in
277 a default Windows installation.)
279 - bup has no GUI. Actually, that's not stupid, but you might consider it
280 a limitation. There are a bunch of Linux GUI backup programs; someday
281 I expect someone will adapt one of them to use bup.
287 bup is a work in progress and there are many ways it can still be improved.
288 If you'd like to contribute patches, ideas, or bug reports, please join the
291 You can find the mailing list archives here:
293 http://groups.google.com/group/bup-list
295 and you can subscribe by sending a message to:
297 bup-list+subscribe@googlegroups.com