| | 1 | NEWS for rsync 2.6.1 (26 Apr 2004) |
| | 2 | Protocol: 28 (changed) |
| | 3 | Changes since 2.6.0: |
| | 4 | |
| | 5 | SECURITY FIXES: |
| | 6 | |
| | 7 | - Paths sent to an rsync daemon are more thoroughly sanitized when |
| | 8 | chroot is not used. If you're running a non-read-only rsync |
| | 9 | daemon with chroot disabled, *please upgrade*, ESPECIALLY if the |
| | 10 | user privs you run rsync under is anything above "nobody". |
| | 11 | |
| | 12 | ENHANCEMENTS: |
| | 13 | |
| | 14 | - Lower memory use, more optimal transfer of data over the socket, |
| | 15 | and lower CPU usage (see the INTERNAL section for details). |
| | 16 | |
| | 17 | - The RSYNC_PROXY environment variable can now contain a |
| | 18 | "USER:PASS@" prefix before the "HOST:PORT" information. |
| | 19 | (Bardur Arantsson) |
| | 20 | |
| | 21 | - The --progress output now mentions how far along in the transfer |
| | 22 | we are, including both a count of files transferred and a |
| | 23 | percentage of the total file-count that we've processed. It also |
| | 24 | shows better current-rate-of-transfer and remaining-transfer-time |
| | 25 | values. |
| | 26 | |
| | 27 | - The configure script now accepts --with-rsyncd-conf=PATH to |
| | 28 | override the default value of the /etc/rsyncd.conf file. |
| | 29 | |
| | 30 | - Added a couple extra diffs in the "patches" dir, removed the ones |
| | 31 | that got applied, and rebuilt the rest. |
| | 32 | |
| | 33 | - Documentation changes now attempt to describe some often mis- |
| | 34 | understood features more clearly. |
| | 35 | |
| | 36 | BUG FIXES: |
| | 37 | |
| | 38 | - When -x (--one-file-system) is combined with -L (--copy-links) or |
| | 39 | --copy-unsafe-links, no symlinked files are skipped, even if the |
| | 40 | referent file is on a different filesystem. |
| | 41 | |
| | 42 | - The --link-dest code now works properly for a non-root user when |
| | 43 | (1) the UIDs of the source and destination differ and -o was |
| | 44 | specified, or (2) when the group of the source can't be used on |
| | 45 | the destination and -g was specified. |
| | 46 | |
| | 47 | - Fixed a bug in the handling of -H (hard-links) that might cause |
| | 48 | the expanded PATH/NAME value of the current item to get |
| | 49 | overwritten (due to an expanded-name caching bug). |
| | 50 | |
| | 51 | - We now reset the "new data has been sent" flag at the start of |
| | 52 | each file we send. This makes sure that an interrupted transfer |
| | 53 | with the --partial option set doesn't keep a shorter temp file |
| | 54 | than the current basis file when no new data has been transfered |
| | 55 | over the wire for that file. |
| | 56 | |
| | 57 | - Fixed a byte-order problem in --batch-mode on big-endian machines. |
| | 58 | (Jay Fenlason) |
| | 59 | |
| | 60 | - Fixed configure bug when running "./configure --disable-ipv6". |
| | 61 | |
| | 62 | - Fixed "make test" bug when build dir is not the source dir. |
| | 63 | |
| | 64 | - When using --cvs-exclude, the exclude items we get from a |
| | 65 | per-directory's .cvsignore file once again only affect that one |
| | 66 | directory (not all following directories too). The items are also |
| | 67 | now properly word-split and parsed without any +/- prefix parsing. |
| | 68 | |
| | 69 | - When specifying the USER@HOST: prefix for a file, the USER part |
| | 70 | can now contain an '@', if needed (i.e. the last '@' is used to |
| | 71 | find the HOST, not the first). |
| | 72 | |
| | 73 | - Fixed some bugs in the handling of group IDs for non-root users: |
| | 74 | (1) It properly handles a group that the sender didn't have a name |
| | 75 | for (it would previously skip changing the group on any files in |
| | 76 | that group). (2) If --numeric-ids is used, rsync no longer |
| | 77 | attempts to set groups that the user doesn't have the permission |
| | 78 | to set. |
| | 79 | |
| | 80 | - Fixed the "refuse options" setting in the rsyncd.conf file. |
| | 81 | |
| | 82 | - Improved the -x (--one-file-system) flag's handling of any mount- |
| | 83 | point directories we encounter. It is both more optimal (in that |
| | 84 | it no longer does a useless scan of the contents of the mount- |
| | 85 | point dirs) and also fixes a bug where a remapped mount of the |
| | 86 | original filesystem could get discovered in a subdir we should be |
| | 87 | ignoring. |
| | 88 | |
| | 89 | - Rsync no longer discards a double-slash at the start of a filename |
| | 90 | when trying to open the file. It also no longer constructs names |
| | 91 | that start with a double slash (unless the user supplied them). |
| | 92 | |
| | 93 | - Path-specifying options to a daemon should now work the same with |
| | 94 | or without chroot turned on. Previously, such a option (such as |
| | 95 | --link-dest) would get its absolute path munged into a relative |
| | 96 | one if chroot was not on, making that setting fairly useless. |
| | 97 | Rsync now transforms the path into one that is based on the |
| | 98 | module's base dir when chroot is not enabled. |
| | 99 | |
| | 100 | - Fixed compilation problem on Tru64 Unix (having to do with |
| | 101 | sockaddr.sa_len and sockaddr.sin_len). |
| | 102 | |
| | 103 | - Fixed a compatibility problem interacting with older rsync |
| | 104 | versions that might send us an empty --suffix value without |
| | 105 | telling us that --backup-dir was specified. |
| | 106 | |
| | 107 | - The "hosts allow" option for a daemon-over-remote-shell process |
| | 108 | now has improved support for IPv6 addresses and a fix for systems |
| | 109 | that have a length field in their socket structs. |
| | 110 | |
| | 111 | - Fixed the ability to request an empty backup --suffix when sending |
| | 112 | files to an rsync daemon. |
| | 113 | |
| | 114 | INTERNAL: |
| | 115 | |
| | 116 | - Most of the I/O is now buffered, which results in a pretty large |
| | 117 | speedup when running under MS Windows. (Craig Barratt) |
| | 118 | |
| | 119 | - Optimizations to the name-handling/comparing code have made some |
| | 120 | significant reductions in user-CPU time for large file sets. |
| | 121 | |
| | 122 | - Some cleanup of the variable types make the code more consistent. |
| | 123 | |
| | 124 | - Reduced memory requirements of hard link preservation. |
| | 125 | (J.W. Schultz) |
| | 126 | |
| | 127 | - Implemented a new algorithm for hard-link handling that speeds up |
| | 128 | the code significantly. (J.W. Schultz and Wayne Davison) |
| | 129 | |
| | 130 | - The --hard-link option now uses the first existing file in the |
| | 131 | group of linked files as the basis for the transfer. This |
| | 132 | prevents the sub-optimal transfer of a file's data when a new |
| | 133 | hardlink is added on the sending side and it sorts alphabetically |
| | 134 | earlier in the list than the files that are already present on the |
| | 135 | receiving side. |
| | 136 | |
| | 137 | - Dropped support for protocol versions less than 20 (2.3.0 released |
| | 138 | 15 Mar 1999) and activated warnings for protocols less than 25 |
| | 139 | (2.5.0 released 23 Aug 2001). (Wayne Davison and J.W. Schultz, |
| | 140 | severally) |
| | 141 | |
| | 142 | - More optimal data transmission for --hard-links (protocol 28). |
| | 143 | |
| | 144 | - More optimal data transmission for --checksum (protocol 28). |
| | 145 | |
| | 146 | - Less memory is used when --checksum is specified. |
| | 147 | |
| | 148 | - Less memory is used in the file list (a per-file savings). |
| | 149 | |
| | 150 | - The generator is now better about not modifying the file list |
| | 151 | during the transfer in order to avoid a copy-on-write memory |
| | 152 | bifurcation (on systems where fork() uses shared memory). |
| | 153 | Previously, rsync's shared memory would slowly become unshared, |
| | 154 | resulting in real memory usage nearly doubling on the receiving |
| | 155 | side by the end of the transfer. Now, as long as permissions |
| | 156 | are being preserved, the shared memory should remain that way |
| | 157 | for the entire transfer. |
| | 158 | |
| | 159 | - Changed hardlink info and file_struct + strings to use allocation |
| | 160 | pools. This reduces memory use for large file-sets and permits |
| | 161 | freeing memory to the OS. (J.W. Schultz) |
| | 162 | |
| | 163 | - The 2 pipes used between the receiver and generator processes |
| | 164 | (which are forked on the same machine) were reduced to 1 pipe and |
| | 165 | the protocol improved so that (1) it is now impossible to have the |
| | 166 | "redo" pipe fill up and hang rsync, and (2) trailing messages from |
| | 167 | the receiver don't get lost on their way through the generator |
| | 168 | over to the sender (which mainly affected hard-link messages and |
| | 169 | verbose --stats output). |
| | 170 | |
| | 171 | - Improved the internal uid/gid code to be more portable and a |
| | 172 | little more optimized. |
| | 173 | |
| | 174 | - The device numbers sent when using --devices are now sent as |
| | 175 | separate major/minor values with 32-bit accuracy (protocol 28). |
| | 176 | Previously, the copied devices were sent as a single 32-bit |
| | 177 | number. This will make inter-operation of 64-bit binaries more |
| | 178 | compatible with their 32-bit brethren (with both ends of the |
| | 179 | connection are using protocol 28). Note that optimizations in the |
| | 180 | binary protocol for sending the device numbers often results in |
| | 181 | fewer bytes being used than before, even though more precision is |
| | 182 | now available. |
| | 183 | |
| | 184 | - Some cleanup of the exclude/include structures and its code made |
| | 185 | things clearer (internally), simpler, and more efficient. |
| | 186 | |
| | 187 | - The reading & writing of the file-list in batch-mode is now |
| | 188 | handled by the same code that sends & receives the list over the |
| | 189 | wire. This makes it much easier to maintain. (Note that the |
| | 190 | batch code is still considered to be experimental.) |
Matt McCutchen's Web Site / rsync / rsync.git / blame_incremental