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 IMPORTANT ------------------------------------------------------------
79 --dry-run is insufficiently dry
81 Mark Santcroos points out that -n fails to list files which have
82 only metadata changes, though it probably should.
84 There may be a Debian bug about this as well.
89 If the platform doesn't support it, then don't even try.
91 If running as non-root, then don't fail, just give a warning.
92 (There was a thread about this a while ago?)
94 http://lists.samba.org/pipermail/rsync/2001-August/thread.html
95 http://lists.samba.org/pipermail/rsync/2001-September/thread.html
100 Avoids traversal. Better option than a pile of --include statements
101 for people who want to generate the file list using a find(1)
105 File list structure in memory
107 Rather than one big array, perhaps have a tree in memory mirroring
110 This might make sorting much faster! (I'm not sure it's a big CPU
113 It might also reduce memory use in storing repeated directory names
114 -- again I'm not sure this is a problem.
118 Traverse just one directory at a time. Tridge says it's possible.
120 At the moment rsync reads the whole file list into memory at the
121 start, which makes us use a lot of memory and also not pipeline
122 network access as much as we could.
125 Handling duplicate names
127 We need to be careful of duplicate names getting into the file list.
128 See clean_flist(). This could happen if multiple arguments include
131 I think duplicates are only a problem if they're both flowing
132 through the pipeline at the same time. For example we might have
133 updated the first occurrence after reading the checksums for the
134 second. So possibly we just need to make sure that we don't have
135 both in the pipeline at the same time.
137 Possibly if we did one directory at a time that would be sufficient.
139 Alternatively we could pre-process the arguments to make sure no
140 duplicates will ever be inserted. There could be some bad cases
141 when we're collapsing symlinks.
143 We could have a hash table.
145 The root of the problem is that we do not want more than one file
146 list entry referring to the same file. At first glance there are
147 several ways this could happen: symlinks, hardlinks, and repeated
148 names on the command line.
150 If names are repeated on the command line, they may be present in
151 different forms, perhaps by traversing directory paths in different
152 ways, traversing paths including symlinks. Also we need to allow
153 for expansion of globs by rsync.
155 At the moment, clean_flist() requires having the entire file list in
156 memory. Duplicate names are detected just by a string comparison.
158 We don't need to worry about hard links causing duplicates because
159 files are never updated in place. Similarly for symlinks.
161 I think even if we're using a different symlink mode we don't need
164 Unless we're really clever this will introduce a protocol
165 incompatibility, so we need to be able to accept the old format as
171 At exit, show how much memory was used for the file list, etc.
173 Also we do a wierd exponential-growth allocation in flist.c. I'm
174 not sure this makes sense with modern mallocs. At any rate it will
175 make us allocate a huge amount of memory for large file lists.
180 At the moment hardlink handling is very expensive, so it's off by
181 default. It does not need to be so.
183 Since most of the solutions are rather intertwined with the file
184 list it is probably better to fix that first, although fixing
185 hardlinks is possibly simpler.
187 We can rule out hardlinked directories since they will probably
188 screw us up in all kinds of ways. They simply should not be used.
190 At the moment rsync only cares about hardlinks to regular files. I
191 guess you could also use them for sockets, devices and other beasts,
192 but I have not seen them.
194 When trying to reproduce hard links, we only need to worry about
195 files that have more than one name (nlinks>1 && !S_ISDIR).
197 The basic point of this is to discover alternate names that refer to
198 the same file. All operations, including creating the file and
199 writing modifications to it need only to be done for the first name.
200 For all later names, we just create the link and then leave it
203 If hard links are to be preserved:
205 Before the generator/receiver fork, the list of files is received
206 from the sender (recv_file_list), and a table for detecting hard
209 The generator looks for hard links within the file list and does
210 not send checksums for them, though it does send other metadata.
212 The sender sends the device number and inode with file entries, so
213 that files are uniquely identified.
215 The receiver goes through and creates hard links (do_hard_links)
216 after all data has been written, but before directory permissions
219 At the moment device and inum are sent as 4-byte integers, which
220 will probably cause problems on large filesystems. On Linux the
221 kernel uses 64-bit ino_t's internally, and people will soon have
222 filesystems big enough to use them. We ought to follow NFS4 in
223 using 64-bit device and inode identification, perhaps with a
224 protocol version bump.
226 Once we've seen all the names for a particular file, we no longer
227 need to think about it and we can deallocate the memory.
229 We can also have the case where there are links to a file that are
230 not in the tree being transferred. There's nothing we can do about
231 that. Because we rename the destination into place after writing,
232 any hardlinks to the old file are always going to be orphaned. In
233 fact that is almost necessary because otherwise we'd get really
234 confused if we were generating checksums for one name of a file and
237 At the moment the code seems to make a whole second copy of the file
238 list, which seems unnecessary.
240 We should have a test case that exercises hard links. Since it
241 might be hard to compare ./tls output where the inodes change we
242 might need a little program to check whether several names refer to
247 Implement suggestions from http://www.kame.net/newsletter/19980604/
248 and ftp://ftp.iij.ad.jp/pub/RFC/rfc2553.txt
250 If a host has multiple addresses, then listen try to connect to all
251 in order until we get through. (getaddrinfo may return multiple
252 addresses.) This is kind of implemented already.
254 Possibly also when starting as a server we may need to listen on
255 multiple passive addresses. This might be a bit harder, because we
256 may need to select on all of them. Hm.
258 Define a syntax for IPv6 literal addresses. Since they include
259 colons, they tend to break most naming systems, including ours.
260 Based on the HTTP IPv6 syntax, I think we should use
262 rsync://[::1]/foo/bar
265 which should just take a small change to the parser code.
270 If we hang or get SIGINT, then explain where we were up to. Perhaps
271 have a static buffer that contains the current function name, or
272 some kind of description of what we were trying to do. This is a
273 little easier on people than needing to run strace/truss.
275 "The dungeon collapses! You are killed." Rather than "unexpected
276 eof" give a message that is more detailed if possible and also more
279 If we get an error writing to a socket, then we should perhaps
280 continue trying to read to see if an error message comes across
281 explaining why the socket is closed. I'm not sure if this would
282 work, but it would certainly make our messages more helpful.
284 What happens if a directory is missing -x attributes. Do we lose
285 our load? (Debian #28416) Probably fixed now, but a test case
291 Device major/minor numbers should be at least 32 bits each. See
292 http://lists.samba.org/pipermail/rsync/2001-November/005357.html
294 Transfer ACLs. Need to think of a standard representation.
295 Probably better not to even try to convert between NT and POSIX.
296 Possibly can share some code with Samba.
300 With the current common --include '*/' --exclude '*' pattern, people
301 can end up with many empty directories. We might avoid this by
302 lazily creating such directories.
307 Perhaps don't use our own zlib.
311 - will automatically be up to date with bugfixes in zlib
313 - can leave it out for small rsync on e.g. recovery disks
315 - can use a shared library
317 - avoids people breaking rsync by trying to do this themselves and
320 Should we ship zlib for systems that don't have it, or require
321 people to install it separately?
323 Apparently this will make us incompatible with versions of rsync
324 that use the patched version of rsync. Probably the simplest way to
325 do this is to just disable gzip (with a warning) when talking to old
331 Perhaps flush stdout after each filename, so that people trying to
332 monitor progress in a log file can do so more easily. See
333 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=48108
335 At the connections that just get a list of modules are not logged,
338 If a child of the rsync daemon dies with a signal, we should notice
339 that when we reap it and log a message.
341 Keep stderr and stdout properly separated (Debian #23626)
343 Use a separate function for reporting errors; prefix it with
344 "rsync:" or "rsync(remote)", or perhaps even "rsync(local
350 There are already some patches to do this.
354 Allow RSYNC_PROXY to be http://user:pass@proxy.foo:3128/, and do
355 HTTP Basic Proxy-Authentication.
357 Multiple schemes are possible, up to and including the insanity that
358 is NTLM, but Basic probably covers most cases.
362 Add --with-socks, and then perhaps a command-line option to put them
363 on or off. This might be more reliable than LD_PRELOAD hacks.
367 <Rasmus> mbp: hey, how about an rsync option that just gives you the
368 summary without the list of files? And perhaps gives more
369 information like the number of new files, number of changed,
371 <mbp> Rasmus: nice idea
372 <mbp> there is --stats
373 <mbp> but at the moment it's very tridge-oriented
374 <mbp> rather than user-friendly
375 <mbp> it would be nice to improve it
376 <mbp> that would also work well with --dryrun
380 Rather than storing the file list in memory, store it in a TDB.
382 This *might* make memory usage lower while building the file list.
384 Hashtable lookup will mean files are not transmitted in order,
387 This would neatly eliminate one of the major post-fork shared data
393 On 12 Mar 2002, Dave Dykstra <dwd@bell-labs.com> wrote:
394 > If we would add an option to do that functionality, I would vote for one
395 > that was more general which could mask off any set of permission bits and
396 > possibly add any set of bits. Perhaps a chmod-like syntax if it could be
397 > implemented simply.
399 I think that would be good too. For example, people uploading files
400 to a web server might like to say
402 rsync -avzP --chmod a+rX ./ sourcefrog.net:/home/www/sourcefrog/
404 Ideally the patch would implement as many of the gnu chmod semantics
405 as possible. I think the mode parser should be a separate function
406 that passes back something like (mask,set) description to the rest of
407 the program. For bonus points there would be a test case for the
415 Allow people to specify the diff command. (Might want to use wdiff,
418 Just diff the temporary file with the destination file, and delete
419 the tmp file rather than moving it into place.
421 Interaction with --partial.
423 Security interactions with daemon mode?
425 (Suggestion from david.e.sewell)
428 Incorrect timestamps (Debian #100295)
430 A bit hard to believe, but apparently it happens.
433 Check "refuse options works"
435 We need a test case for this...
437 Was this broken when we changed to popt?
442 Test whether this is actually faster than just using malloc(). If
443 it's not (anymore), throw it out.
447 PLATFORMS ------------------------------------------------------------
451 Don't detach, because this messes up --srvany.
453 http://sources.redhat.com/ml/cygwin/2001-08/msg00234.html
455 According to "Effective TCP/IP Programming" (??) close() on a socket
456 has incorrect behaviour on Windows -- it sends a RST packet to the
457 other side, which gives a "connection reset by peer" error. On that
458 platform we should probably do shutdown() instead. However, on Unix
459 we are correct to call close(), because shutdown() discards
462 DEVELOPMENT ----------------------------------------------------------
466 Build rsync with SPLINT to try to find security holes. Add
467 annotations as necessary. Keep track of the number of warnings
468 found initially, and see how many of them are real bugs, or real
469 security bugs. Knowing the percentage of likely hits would be
470 really interesting for other projects.
474 Something that just keeps running rsync continuously over a data set
475 likely to generate problems.
479 Run current rsync versions against significant past releases.
483 jra recommends Valgrind:
485 http://devel-home.kde.org/~sewardj/
487 TESTING --------------------------------------------------------------
491 Part of the regression suite should be making sure that we don't
492 break backwards compatibility: old clients vs new servers and so
493 on. Ideally we would test the cross product of versions.
495 It might be sufficient to test downloads from well-known public
496 rsync servers running different versions of rsync. This will give
497 some testing and also be the most common case for having different
498 versions and not being able to upgrade.
502 Sparse and non-sparse
506 Insert bytes, delete bytes, swap blocks, ...
508 configure option to enable dangerous tests
510 If tests are skipped, say why.
512 Test daemon feature to disallow particular options.
514 Pipe program that makes slow/jerky connections.
516 Versions of read() and write() that corrupt the stream, or abruptly fail
518 Separate makefile target to run rough tests -- or perhaps just run
522 DOCUMENTATION --------------------------------------------------------
526 Keep list of open issues and todos on the web site
528 Update web site from CVS
530 BUILD FARM -----------------------------------------------------------
534 AMDAHL UTS (Dave Dykstra)
536 Cygwin (on different versions of Win32?)
538 HP-UX variants (via HP?)
542 NICE -----------------------------------------------------------------
544 --no-detach and --no-fork options
546 Very useful for debugging. Also good when running under a
547 daemon-monitoring process that tries to restart the service when the
550 hang/timeout friendliness
554 Indicate whether files are new, updated, or deleted
556 At end of transfer, show how many files were or were not transferred
561 Change to using gettext(). Probably need to ship this for platforms
564 Solicit translations.
570 Write a small emulation of interactive ftp as a Pythonn program
571 that calls rsync. Commands such as "cd", "ls", "ls *.c" etc map
572 fairly directly into rsync commands: it just needs to remember the
573 current host, directory and so on. We can probably even do
574 completion of remote filenames.