3 BUGS ---------------------------------------------------------------
5 There seems to be a bug with hardlinks
7 mbp/2 build$ ls -l /tmp/a /tmp/b -i
10 2568307 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a1
11 2568307 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a2
12 2568307 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a3
13 2568310 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 a4
14 2568310 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 a5
15 2568310 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b1
16 2568310 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b2
17 2568310 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b3
21 2568309 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a1
22 2568309 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a2
23 2568309 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a3
24 2568311 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 a4
25 2568311 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 a5
26 2568311 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b1
27 2568311 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b2
28 2568311 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b3
29 mbp/2 build$ rm -r /tmp/b && ./rsync -avH /tmp/a/ /tmp/b
30 building file list ... done
31 created directory /tmp/b
37 wrote 350 bytes read 52 bytes 804.00 bytes/sec
38 total size is 232 speedup is 0.58
39 mbp/2 build$ rm -r /tmp/b
40 mbp/2 build$ ls -l /tmp/b
41 ls: /tmp/b: No such file or directory
42 mbp/2 build$ rm -r /tmp/b && ./rsync -avH /tmp/a/ /tmp/b
43 rm: cannot remove `/tmp/b': No such file or directory
44 mbp/2 build$ rm -f -r /tmp/b && ./rsync -avH /tmp/a/ /tmp/b
45 building file list ... done
46 created directory /tmp/b
52 wrote 350 bytes read 52 bytes 804.00 bytes/sec
53 total size is 232 speedup is 0.58
54 mbp/2 build$ ls -l /tmp/b
56 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a1
57 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a2
58 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a3
59 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 a4
60 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 a5
61 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b1
62 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b2
63 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b3
64 mbp/2 build$ ls -l /tmp/a
66 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a1
67 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a2
68 -rw-rw-r-- 3 mbp mbp 29 Mar 25 17:30 a3
69 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 a4
70 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 a5
71 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b1
72 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b2
73 -rw-rw-r-- 5 mbp mbp 29 Mar 25 17:30 b3
76 Progress indicator can produce corrupt output when transferring directories:
79 main/binary-arm/admin/
81 main/binary-arm/comm/8.56kB/s 0:00:52
82 main/binary-arm/devel/
84 main/binary-arm/editors/
85 main/binary-arm/electronics/s 0:00:53
86 main/binary-arm/games/
87 main/binary-arm/graphics/
88 main/binary-arm/hamradio/
89 main/binary-arm/interpreters/
90 main/binary-arm/libs/6.61kB/s 0:00:54
97 I don't think we handle this properly on systems that don't have the
101 DAEMON --------------------------------------------------------------
103 server-imposed bandwidth limits
107 There are already some patches to do this.
109 BitKeeper uses a server whose login shell is set to bkd. That's
110 probably a reasonable approach.
113 FEATURES ------------------------------------------------------------
116 --dry-run is insufficiently dry
118 Mark Santcroos points out that -n fails to list files which have
119 only metadata changes, though it probably should.
121 There may be a Debian bug about this as well.
126 If the platform doesn't support it, then don't even try.
128 If running as non-root, then don't fail, just give a warning.
129 (There was a thread about this a while ago?)
131 http://lists.samba.org/pipermail/rsync/2001-August/thread.html
132 http://lists.samba.org/pipermail/rsync/2001-September/thread.html
137 Avoids traversal. Better option than a pile of --include statements
138 for people who want to generate the file list using a find(1)
142 File list structure in memory
144 Rather than one big array, perhaps have a tree in memory mirroring
147 This might make sorting much faster! (I'm not sure it's a big CPU
150 It might also reduce memory use in storing repeated directory names
151 -- again I'm not sure this is a problem.
155 Traverse just one directory at a time. Tridge says it's possible.
157 At the moment rsync reads the whole file list into memory at the
158 start, which makes us use a lot of memory and also not pipeline
159 network access as much as we could.
162 Handling duplicate names
164 We need to be careful of duplicate names getting into the file list.
165 See clean_flist(). This could happen if multiple arguments include
168 I think duplicates are only a problem if they're both flowing
169 through the pipeline at the same time. For example we might have
170 updated the first occurrence after reading the checksums for the
171 second. So possibly we just need to make sure that we don't have
172 both in the pipeline at the same time.
174 Possibly if we did one directory at a time that would be sufficient.
176 Alternatively we could pre-process the arguments to make sure no
177 duplicates will ever be inserted. There could be some bad cases
178 when we're collapsing symlinks.
180 We could have a hash table.
182 The root of the problem is that we do not want more than one file
183 list entry referring to the same file. At first glance there are
184 several ways this could happen: symlinks, hardlinks, and repeated
185 names on the command line.
187 If names are repeated on the command line, they may be present in
188 different forms, perhaps by traversing directory paths in different
189 ways, traversing paths including symlinks. Also we need to allow
190 for expansion of globs by rsync.
192 At the moment, clean_flist() requires having the entire file list in
193 memory. Duplicate names are detected just by a string comparison.
195 We don't need to worry about hard links causing duplicates because
196 files are never updated in place. Similarly for symlinks.
198 I think even if we're using a different symlink mode we don't need
201 Unless we're really clever this will introduce a protocol
202 incompatibility, so we need to be able to accept the old format as
208 At exit, show how much memory was used for the file list, etc.
210 Also we do a wierd exponential-growth allocation in flist.c. I'm
211 not sure this makes sense with modern mallocs. At any rate it will
212 make us allocate a huge amount of memory for large file lists.
217 At the moment hardlink handling is very expensive, so it's off by
218 default. It does not need to be so.
220 Since most of the solutions are rather intertwined with the file
221 list it is probably better to fix that first, although fixing
222 hardlinks is possibly simpler.
224 We can rule out hardlinked directories since they will probably
225 screw us up in all kinds of ways. They simply should not be used.
227 At the moment rsync only cares about hardlinks to regular files. I
228 guess you could also use them for sockets, devices and other beasts,
229 but I have not seen them.
231 When trying to reproduce hard links, we only need to worry about
232 files that have more than one name (nlinks>1 && !S_ISDIR).
234 The basic point of this is to discover alternate names that refer to
235 the same file. All operations, including creating the file and
236 writing modifications to it need only to be done for the first name.
237 For all later names, we just create the link and then leave it
240 If hard links are to be preserved:
242 Before the generator/receiver fork, the list of files is received
243 from the sender (recv_file_list), and a table for detecting hard
246 The generator looks for hard links within the file list and does
247 not send checksums for them, though it does send other metadata.
249 The sender sends the device number and inode with file entries, so
250 that files are uniquely identified.
252 The receiver goes through and creates hard links (do_hard_links)
253 after all data has been written, but before directory permissions
256 At the moment device and inum are sent as 4-byte integers, which
257 will probably cause problems on large filesystems. On Linux the
258 kernel uses 64-bit ino_t's internally, and people will soon have
259 filesystems big enough to use them. We ought to follow NFS4 in
260 using 64-bit device and inode identification, perhaps with a
261 protocol version bump.
263 Once we've seen all the names for a particular file, we no longer
264 need to think about it and we can deallocate the memory.
266 We can also have the case where there are links to a file that are
267 not in the tree being transferred. There's nothing we can do about
268 that. Because we rename the destination into place after writing,
269 any hardlinks to the old file are always going to be orphaned. In
270 fact that is almost necessary because otherwise we'd get really
271 confused if we were generating checksums for one name of a file and
274 At the moment the code seems to make a whole second copy of the file
275 list, which seems unnecessary.
277 We should have a test case that exercises hard links. Since it
278 might be hard to compare ./tls output where the inodes change we
279 might need a little program to check whether several names refer to
284 Implement suggestions from http://www.kame.net/newsletter/19980604/
285 and ftp://ftp.iij.ad.jp/pub/RFC/rfc2553.txt
287 If a host has multiple addresses, then listen try to connect to all
288 in order until we get through. (getaddrinfo may return multiple
289 addresses.) This is kind of implemented already.
291 Possibly also when starting as a server we may need to listen on
292 multiple passive addresses. This might be a bit harder, because we
293 may need to select on all of them. Hm.
295 Define a syntax for IPv6 literal addresses. Since they include
296 colons, they tend to break most naming systems, including ours.
297 Based on the HTTP IPv6 syntax, I think we should use
299 rsync://[::1]/foo/bar
302 which should just take a small change to the parser code.
307 If we hang or get SIGINT, then explain where we were up to. Perhaps
308 have a static buffer that contains the current function name, or
309 some kind of description of what we were trying to do. This is a
310 little easier on people than needing to run strace/truss.
312 "The dungeon collapses! You are killed." Rather than "unexpected
313 eof" give a message that is more detailed if possible and also more
316 If we get an error writing to a socket, then we should perhaps
317 continue trying to read to see if an error message comes across
318 explaining why the socket is closed. I'm not sure if this would
319 work, but it would certainly make our messages more helpful.
321 What happens if a directory is missing -x attributes. Do we lose
322 our load? (Debian #28416) Probably fixed now, but a test case
328 Device major/minor numbers should be at least 32 bits each. See
329 http://lists.samba.org/pipermail/rsync/2001-November/005357.html
331 Transfer ACLs. Need to think of a standard representation.
332 Probably better not to even try to convert between NT and POSIX.
333 Possibly can share some code with Samba.
337 With the current common --include '*/' --exclude '*' pattern, people
338 can end up with many empty directories. We might avoid this by
339 lazily creating such directories.
344 Perhaps don't use our own zlib.
348 - will automatically be up to date with bugfixes in zlib
350 - can leave it out for small rsync on e.g. recovery disks
352 - can use a shared library
354 - avoids people breaking rsync by trying to do this themselves and
357 Should we ship zlib for systems that don't have it, or require
358 people to install it separately?
360 Apparently this will make us incompatible with versions of rsync
361 that use the patched version of rsync. Probably the simplest way to
362 do this is to just disable gzip (with a warning) when talking to old
368 Allow RSYNC_PROXY to be http://user:pass@proxy.foo:3128/, and do
369 HTTP Basic Proxy-Authentication.
371 Multiple schemes are possible, up to and including the insanity that
372 is NTLM, but Basic probably covers most cases.
376 Add --with-socks, and then perhaps a command-line option to put them
377 on or off. This might be more reliable than LD_PRELOAD hacks.
381 <Rasmus> mbp: hey, how about an rsync option that just gives you the
382 summary without the list of files? And perhaps gives more
383 information like the number of new files, number of changed,
385 <mbp> Rasmus: nice idea
386 <mbp> there is --stats
387 <mbp> but at the moment it's very tridge-oriented
388 <mbp> rather than user-friendly
389 <mbp> it would be nice to improve it
390 <mbp> that would also work well with --dryrun
394 Rather than storing the file list in memory, store it in a TDB.
396 This *might* make memory usage lower while building the file list.
398 Hashtable lookup will mean files are not transmitted in order,
401 This would neatly eliminate one of the major post-fork shared data
407 On 12 Mar 2002, Dave Dykstra <dwd@bell-labs.com> wrote:
408 > If we would add an option to do that functionality, I would vote for one
409 > that was more general which could mask off any set of permission bits and
410 > possibly add any set of bits. Perhaps a chmod-like syntax if it could be
411 > implemented simply.
413 I think that would be good too. For example, people uploading files
414 to a web server might like to say
416 rsync -avzP --chmod a+rX ./ sourcefrog.net:/home/www/sourcefrog/
418 Ideally the patch would implement as many of the gnu chmod semantics
419 as possible. I think the mode parser should be a separate function
420 that passes back something like (mask,set) description to the rest of
421 the program. For bonus points there would be a test case for the
429 Allow people to specify the diff command. (Might want to use wdiff,
432 Just diff the temporary file with the destination file, and delete
433 the tmp file rather than moving it into place.
435 Interaction with --partial.
437 Security interactions with daemon mode?
439 (Suggestion from david.e.sewell)
442 Incorrect timestamps (Debian #100295)
444 A bit hard to believe, but apparently it happens.
447 Check "refuse options works"
449 We need a test case for this...
451 Was this broken when we changed to popt?
454 PERFORMANCE ----------------------------------------------------------
458 If we're doing a local transfer, or using -W, then perhaps don't
459 send the file checksum. If we're doing a local transfer, then
460 calculating MD4 checksums uses 90% of CPU and is unlikely to be
463 Indeed for transfers over zlib or ssh we can also rely on the
464 transport to have quite strong protection against corruption.
466 Perhaps we should have an option to disable this, analogous to
467 --whole-file, although it would default to disabled. The file
468 checksum takes up a definite space in the protocol -- we can either
469 set it to 0, or perhaps just leave it out.
473 Perhaps borrow an assembler MD4 from someone?
475 Make sure we call MD4 with properly-sized blocks whenever possible
476 to avoid copying into the residue region?
480 Test whether this is actually faster than just using malloc(). If
481 it's not (anymore), throw it out.
485 PLATFORMS ------------------------------------------------------------
489 Don't detach, because this messes up --srvany.
491 http://sources.redhat.com/ml/cygwin/2001-08/msg00234.html
493 According to "Effective TCP/IP Programming" (??) close() on a socket
494 has incorrect behaviour on Windows -- it sends a RST packet to the
495 other side, which gives a "connection reset by peer" error. On that
496 platform we should probably do shutdown() instead. However, on Unix
497 we are correct to call close(), because shutdown() discards
500 DEVELOPMENT ----------------------------------------------------------
504 Build rsync with SPLINT to try to find security holes. Add
505 annotations as necessary. Keep track of the number of warnings
506 found initially, and see how many of them are real bugs, or real
507 security bugs. Knowing the percentage of likely hits would be
508 really interesting for other projects.
512 Something that just keeps running rsync continuously over a data set
513 likely to generate problems.
517 Run current rsync versions against significant past releases.
521 jra recommends Valgrind:
523 http://devel-home.kde.org/~sewardj/
525 TESTING --------------------------------------------------------------
529 Part of the regression suite should be making sure that we don't
530 break backwards compatibility: old clients vs new servers and so
531 on. Ideally we would test the cross product of versions.
533 It might be sufficient to test downloads from well-known public
534 rsync servers running different versions of rsync. This will give
535 some testing and also be the most common case for having different
536 versions and not being able to upgrade.
540 Sparse and non-sparse
544 Insert bytes, delete bytes, swap blocks, ...
546 configure option to enable dangerous tests
548 If tests are skipped, say why.
550 Test daemon feature to disallow particular options.
552 Pipe program that makes slow/jerky connections.
554 Versions of read() and write() that corrupt the stream, or abruptly fail
556 Separate makefile target to run rough tests -- or perhaps just run
560 DOCUMENTATION --------------------------------------------------------
564 Keep list of open issues and todos on the web site
566 Update web site from CVS
568 BUILD FARM -----------------------------------------------------------
572 AMDAHL UTS (Dave Dykstra)
574 Cygwin (on different versions of Win32?)
576 HP-UX variants (via HP?)
581 LOGGING --------------------------------------------------------------
583 Perhaps flush stdout after each filename, so that people trying to
584 monitor progress in a log file can do so more easily. See
585 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=48108
587 At the connections that just get a list of modules are not logged,
590 If a child of the rsync daemon dies with a signal, we should notice
591 that when we reap it and log a message.
593 Keep stderr and stdout properly separated (Debian #23626)
595 Use a separate function for reporting errors; prefix it with
596 "rsync:" or "rsync(remote)", or perhaps even "rsync(local
601 Indicate whether files are new, updated, or deleted
603 At end of transfer, show how many files were or were not transferred
608 Explain *why* every file is transferred or not (e.g. "local mtime
609 123123 newer than 1283198")
614 Add an rsyncd.conf parameter to turn on debugging on the server.
618 NICE -----------------------------------------------------------------
620 --no-detach and --no-fork options
622 Very useful for debugging. Also good when running under a
623 daemon-monitoring process that tries to restart the service when the
626 hang/timeout friendliness
630 Change to using gettext(). Probably need to ship this for platforms
633 Solicit translations.
635 Does anyone care? Before we bother modifying the code, we ought to
636 get the manual translated first, because that's possibly more useful
637 and at any rate demonstrates desire.
642 Write a small emulation of interactive ftp as a Pythonn program
643 that calls rsync. Commands such as "cd", "ls", "ls *.c" etc map
644 fairly directly into rsync commands: it just needs to remember the
645 current host, directory and so on. We can probably even do
646 completion of remote filenames.