Re: rsync 3.0.0 apparently crashing when running a single-use "daemon" server process
On Tue, Mar 04, 2008 at 06:20:21PM +0100, Paul Slootman wrote: > The other issue in the bug report is something else, but as that happens > when using 3.0.0 going to the 2.6.8 version running at rsync.net It sounds like they're running something like the support/rrsync validation script, which reminded me that I hadn't updated it to handle the most modern use of the -e option to communicate some extra flags to the server. The git repository now has an updated rrsync script. ..wayne.. -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
Re: preserve ctimes of *unchanged* directories on receiver
On Tue, Mar 04, 2008 at 12:21:01PM +0100, Lasse Kliemann wrote: > 'rsync -a' updates the ctime on a directory even if no file in that directory > has changed. Earlier rsyncs always did an extra stat() prior to setting the time, so I've restored that behavior. The attached patch fixes this. ..wayne.. --- a/generator.c +++ b/generator.c @@ -1949,8 +1949,12 @@ static void touch_up_dirs(struct file_list *flist, int ndx) fname = f_name(file, NULL); if (!(file->mode & S_IWUSR)) do_chmod(fname, file->mode); - if (need_retouch_dir_times) - set_modtime(fname, file->modtime, file->mode); + if (need_retouch_dir_times) { + STRUCT_STAT st; + if (link_stat(fname, &st, 0) == 0 +&& cmp_time(st.st_mtime, file->modtime) != 0) + set_modtime(fname, file->modtime, file->mode); + } if (allowed_lull && !(counter % lull_mod)) maybe_send_keepalive(); else if (!(counter & 0xFF)) -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
DO NOT REPLY [Bug 5301] rsync 3.0.0 copyright dates not updated to 2008
https://bugzilla.samba.org/show_bug.cgi?id=5301 [EMAIL PROTECTED] changed: What|Removed |Added Status|NEW |RESOLVED Resolution||FIXED --- Comment #2 from [EMAIL PROTECTED] 2008-03-04 23:51 CST --- Thanks for pointing that out. I updated the comment headers, but not the --version output. I've updated my release script to check for such a situation so it won't happen again. -- Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the QA contact for the bug, or are watching the QA contact. -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
DO NOT REPLY [Bug 5297] itemize and merge tests fail
https://bugzilla.samba.org/show_bug.cgi?id=5297 [EMAIL PROTECTED] changed: What|Removed |Added Status|ASSIGNED|RESOLVED Resolution||FIXED --- Comment #3 from [EMAIL PROTECTED] 2008-03-04 23:50 CST --- The git repository has this fixed now. -- Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the QA contact for the bug, or are watching the QA contact. -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
DO NOT REPLY [Bug 5297] itemize and merge tests fail
https://bugzilla.samba.org/show_bug.cgi?id=5297 [EMAIL PROTECTED] changed: What|Removed |Added Status|NEW |ASSIGNED Summary|merge test fails|itemize and merge tests fail --- Comment #2 from [EMAIL PROTECTED] 2008-03-04 23:22 CST --- Test 11 chooses to anticipate either a hard-linked symlink ("hL") or a locally-copied symlink ("cL") depending on if it thinks that symlinks can be hard-linked or not. The test using "ln" on a dangling symlink should indicate if that is possible. It looks like that ln test reports the wrong result, so the itemize test is expecting "cL" when rsync returns "hL". Perhaps I should change that test to grep for CAN_HARDLINK_SYMLINK in the config.h file. As for the merge test, it appears that you have "cd reldir" output a path, and that output is causing the diff to fail. I'll code around such a possibility. -- Configure bugmail: https://bugzilla.samba.org/userprefs.cgi?tab=email --- You are receiving this mail because: --- You are the QA contact for the bug, or are watching the QA contact. -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
Re: Rsync sending complete file?
Ok, long story short I have tried and tried to get a solution that I feel comfortable with. I'm after an effective RAID1, but the cheap ass PCI-IDE-RAID card I had originally would refuse to see a HDD. I since got a nicer adaptec one, same problem so I now see a problem with HDD instead off PCI-card (which I'm a little surprised of to be honest). This lost my confidence in RAID, not RAID specifically but my available implementation. All I am trying to stop is HDD failure loosing EVERYTHING, i have had four HDD's go in the last year, a 120Gb, a 300Gb and two 80's. Almost everything was backed up, but still lost some stuff which annoys me as, being a geek, this shouldn't happen. So... I decided to go for Rsync. Upside - simple(ish), possible with nonidentical HDD's, can have more than one spare at a time (running a batch script safer than creating duplicate/backup RAID disks, also can keep one HDD offsite with parents etc), can be imported to another machine with no RAID Downside - has to be updated(prob once every other week/once a month)... manually, hot swapping not possible (not an issue) Bearing in mind down time more than acceptable, also I dont care if this takes a long time, I will probably be asleep, and I dont care about the MOBO-CPU, these are cheap surplus stuff from my employer. I would rather loose 10x mobo-cpu than one HDD. File status: As I said, this is to eliminate HDD failure, so speed not an issue. I have setup things here with one power hungry machine, lots of ram etc, this will host VM's. Another machine is low power usage (aimed for 50W ish, one stick of RAM, two HDD's and expensive earth thingy antec psu). The file server will be on almost 24/7... may even host simple FTP/web server etc. The other machine will be VM-GSX just with a few VM's, mainly a DC and SP1 XP machine (I am studying into security etc) so these need to be VM's to revert to when things go wrong. So copying files will be occasional (weekly at best), performance not an issue as long as its all usable, I will prob use Rsync hash's to really double check its done properly, erm, think thats all. I feel 'almost comfortable' with setup, but if you can think of anything I have missed would be more than grateful to hear them. Kind Regards Matt McCutchen-7 wrote: > > On Sun, 2008-03-02 at 03:05 -0800, Some user wrote: >> (PS - I'm after doing this as a lot of my files are 2gig VM files, and I >> don't want the odd small config change to mean lots if usage on second >> drive, would rather the CPU get pushed than the HDD, CPU more convenient >> to >> replace lol). > > If you could be more specific, I might be able to help you improve the > performance in the way you desire. Are you keeping just one copy of the > source files on the destination, or multiple copies over time? What > kind of "usage" of the destination disk to you want to reduce: disk I/O > traffic or space usage? > > Matt > > -- > To unsubscribe or change options: > https://lists.samba.org/mailman/listinfo/rsync > Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html > > -- View this message in context: http://www.nabble.com/Rsync-sending-complete-file--tp15780743p15839981.html Sent from the Samba - rsync mailing list archive at Nabble.com. -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
Re: Could you make the 2.6.9 source available via http again?
On Tue, Mar 04, 2008 at 06:25:12PM -0500, Matt McCutchen wrote: > Wayne, this brings up a good point. It would be nice if each source > package had a "persistent" URL that worked whether or not the version is > the latest Yeah, that's a good idea. I've organized the source (including the latest 3.0.0 release) into the "src" dir. The version-to-version diffs are in the "src-diffs" dir, and preview releases are in "src-previews". The old-* dirs still work (due to symlinks) and the latest files are hard-linked into the same top-level dir where they've always been. However, the web page now refers to old and new files alike via their src subdir, so that they will remain the same as new releases happen. ..wayne.. -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
Re: when rsync is called with -4 or -6, pass that on to ssh [PATCH]
On Tue, Mar 04, 2008 at 07:07:07PM +0100, Paul Slootman wrote: > If rsync is called with -4 or -6, that option is currently ignored when > ssh is used as the transport. That's because those options are only used when opening a socket. The user should configure ssh on their own. I'd suggest that they use the ~/.ssh/config file to set such options for particular hosts and/or all hosts. I may add some code to ssh to warn and continue if -4 or -6 is used with an external shell transport (to let the user know that it is meaningless). ..wayne.. -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
Re: Several changes missing from [HEAD] fileflags.diff
On Tue, Mar 04, 2008 at 10:40:26AM -0600, Mike Bombich wrote: > but this entire chunk of the that original diff file was lost: Generated files are not included in the repository or its patches. They are only included in the release tar. Those accessing the developer source are expected to have things like autoconf and yodl, so those changes to generated files get recreated by the normal build rules. ..wayne.. -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
Re: Could you make the 2.6.9 source available via http again?
On Tue, 2008-03-04 at 22:56 +, Tom May wrote: > I noticed you've made rsync 3.0.0 available at > http://www.samba.org/ftp/rsync/rsync-3.0.0.tar.gz, which is great! > > However, until it's upgraded (if ever), the buildroot system I'm using to > build > embedded linux tries to download rsync 2.6.9 from > http://www.samba.org/ftp/rsync/rsync-2.6.9.tar.gz, which fails since this file > is no longer available -- even though it's linked from > http://www.samba.org/rsync/! > > Could I convince you to make 2.6.9 available again at the above URL? > > I realize I could get 2.6.9 by reverse patching from 3.0.0, and could likely > get > it from git, but that's not what the buildroot automation wants. Tom, the 2.6.9 source package is available at: http://rsync.samba.org/ftp/rsync/old-versions/rsync-2.6.9.tar.gz Wayne, this brings up a good point. It would be nice if each source package had a "persistent" URL that worked whether or not the version is the latest, e.g.: http://rsync.samba.org/ftp/rsync/versions/rsync-2.6.9.tar.gz You could move "old-versions" to "versions" (with a symlink "old-versions" -> "versions" for compatibility), pile the 3.0.0 files into "versions", and create symlinks to them in http://rsync.samba.org/ftp/rsync/ . Another approach is to use some mod_rewrite magic to redirect requests for nonexistent files in http://rsync.samba.org/ftp/rsync/ to the appropriate subdirectory. Matt -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
Could you make the 2.6.9 source available via http again?
I noticed you've made rsync 3.0.0 available at http://www.samba.org/ftp/rsync/rsync-3.0.0.tar.gz, which is great! However, until it's upgraded (if ever), the buildroot system I'm using to build embedded linux tries to download rsync 2.6.9 from http://www.samba.org/ftp/rsync/rsync-2.6.9.tar.gz, which fails since this file is no longer available -- even though it's linked from http://www.samba.org/rsync/! Could I convince you to make 2.6.9 available again at the above URL? I realize I could get 2.6.9 by reverse patching from 3.0.0, and could likely get it from git, but that's not what the buildroot automation wants. Thanks! .tom -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
when rsync is called with -4 or -6, pass that on to ssh [PATCH]
If rsync is called with -4 or -6, that option is currently ignored when ssh is used as the transport. The attached patch is a stab at passing the option on to the ssh process. It checks that the remote shell program is something that looks like ssh. Perhaps that could be done cleaner... Paul Slootman --- a/main.c2008-03-01 21:01:41.0 +0100 +++ b/main.c2008-03-04 18:55:10.933488013 +0100 @@ -82,6 +82,7 @@ #ifdef ICONV_OPTION extern iconv_t ic_send; #endif +extern int default_af_hint; uid_t our_uid; int local_server = 0; @@ -381,6 +382,23 @@ *t++ = '\0'; } +#ifdef AF_INET +if (default_af_hint == AF_INET) { +if (strncmp(cmd, "ssh", 3) == 0 || strstr(cmd, "/ssh") != NULL) { +/* we're using ssh so we can add a -4 option */ +args[argc++] = "-4"; +} +} +#endif +#ifdef AF_INET6 +if (default_af_hint == AF_INET6) { +if (strncmp(cmd, "ssh", 3) == 0 || strstr(cmd, "/ssh") != NULL) { +/* we're using ssh so we can add a -6 option */ +args[argc++] = "-6"; +} +} +#endif + /* check to see if we've already been given '-l user' in * the remote-shell command */ for (i = 0; i < argc-1; i++) { -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
RSync on Netware
Hi, I am using probably your port of rsync in Novell Netware environment (v. 2.6.6, I think), it helps us very much. Do you plan to make port of version 3, too? Kind regards, Karel Rys, CZ -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
Re: rsync 3.0.0 apparently crashing when running a single-use "daemon" server process
On Mon 03 Mar 2008, Wayne Davison wrote: > On Mon, Mar 03, 2008 at 01:50:28PM -0800, Wayne Davison wrote: > > I'll fix the NULL and look to get a test for this added. > > The git repository has both the fix and the extended test. > > Attached is just the NULL-pointer fix. Confirmed that it now works for that case. The other issue in the bug report is something else, but as that happens when using 3.0.0 going to the 2.6.8 version running at rsync.net, it's pretty impossible to debug. Apparently the remote rsync simply dies immediately when presented with the 3.0.0 client connection. As reproducing the setup with 3.0.0 and 2.6.8 works fine for me, I'm going to mark it unreproducible, it's probably a problem at rsync.net. > Thanks for your help! I think I should be thanking you for the quick fix :) Paul Slootman -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
Several changes missing from [HEAD] fileflags.diff
Looking at http://rsync.samba.org/ftp/rsync/patches/fileflags.diff -- It looks like the changes from fileflags-fixes.diff patch were applied to the patch from http://rsync.samba.org/ftp/rsync/rsync-patches-3.0.0.tar.gz , but this entire chunk of the that original diff file was lost: diff -up a/config.h.in b/config.h.in --- a/config.h.in +++ b/config.h.in @@ -64,6 +64,9 @@ /* Define to 1 if vsprintf has a C99-compatible return value */ #undef HAVE_C99_VSNPRINTF +/* Define to 1 if you have the `chflags' function. */ +#undef HAVE_CHFLAGS + /* Define to 1 if you have the `chmod' function. */ #undef HAVE_CHMOD diff -up a/configure.sh b/configure.sh --- a/configure.sh +++ b/configure.sh @@ -14789,12 +14789,13 @@ fi + for ac_func in waitpid wait4 getcwd strdup chown chmod lchmod mknod mkfifo \ fchmod fstat ftruncate strchr readlink link utime utimes lutimes strftime \ memmove lchown vsnprintf snprintf vasprintf asprintf setsid glob strpbrk \ strlcat strlcpy strtol mallinfo getgroups setgroups geteuid getegid \ setlocale setmode open64 lseek64 mkstemp64 mtrace va_copy __va_copy \ -strerror putenv iconv_open locale_charset nl_langinfo getxattr \ +chflags strerror putenv iconv_open locale_charset nl_langinfo getxattr \ extattr_get_link sigaction sigprocmask setattrlist do as_ac_var=`echo "ac_cv_func_$ac_func" | $as_tr_sh` diff -up a/proto.h b/proto.h --- a/proto.h +++ b/proto.h @@ -269,6 +269,8 @@ int read_ndx_and_attrs(int f_in, int *if void free_sums(struct sum_struct *s); mode_t dest_mode(mode_t flist_mode, mode_t stat_mode, int dflt_perms, int exists); +int make_mutable(const char *fname, mode_t mode, uint32 fileflags, uint32 iflags); +int undo_make_mutable(const char *fname, uint32 fileflags); int set_file_attrs(const char *fname, struct file_struct *file, stat_x *sxp, const char *fnamecmp, int flags); RETSIGTYPE sig_int(UNUSED(int val)); @@ -293,11 +295,12 @@ int sock_exec(const char *prog); int do_unlink(const char *fname); int do_symlink(const char *fname1, const char *fname2); int do_link(const char *fname1, const char *fname2); -int do_lchown(const char *path, uid_t owner, gid_t group); +int do_lchown(const char *path, uid_t owner, gid_t group, mode_t mode, uint32 fileflags); int do_mknod(const char *pathname, mode_t mode, dev_t dev); int do_rmdir(const char *pathname); int do_open(const char *pathname, int flags, mode_t mode); -int do_chmod(const char *path, mode_t mode); +int do_chmod(const char *path, mode_t mode, uint32 fileflags); +int do_chflags(const char *path, uint32 fileflags); int do_rename(const char *fname1, const char *fname2); void trim_trailing_slashes(char *name); int do_mkdir(char *fname, mode_t mode); @@ -326,7 +329,7 @@ int fd_pair(int fd[2]); void print_child_argv(const char *prefix, char **cmd); NORETURN void out_of_memory(const char *str); NORETURN void overflow_exit(const char *str); -int set_modtime(const char *fname, time_t modtime, mode_t mode); +int set_modtime(const char *fname, time_t modtime, mode_t mode, uint32 fileflags); int mkdir_defmode(char *fname); int create_directory_path(char *fname); int full_write(int desc, const char *ptr, size_t len); diff -up a/rsync.1 b/rsync.1 --- a/rsync.1 +++ b/rsync.1 @@ -413,6 +413,7 @@ to the detailed description below for a \-K, \-\-keep\-dirlinks treat symlinked dir on receiver as dir \-H, \-\-hard\-linkspreserve hard links \-p, \-\-perms preserve permissions + \-\-fileflags preserve file-flags (aka chflags) \-E, \-\-executability preserve executability \-\-chmod=CHMOD affect file and/or directory permissions \-A, \-\-acls preserve ACLs (implies \-p) @@ -444,7 +445,10 @@ to the detailed description below for a \-\-delete\-after receiver deletes after transfer, not before \-\-delete\-excluded also delete excluded files from dest dirs \-\-ignore\-errors delete even if there are I/O errors - \-\-force force deletion of dirs even if not empty + \-\-force\-delete force deletion of dirs even if not empty + \-\-force\-change affect user/system immutable files/ dirs + \-\-force\-uchange affect user-immutable files/dirs + \-\-force\-schange affect system-immutable files/dirs \-\-max\-delete=NUMdon't delete more than NUM files \-\-max\-size=SIZE don't transfer any file larger than SIZE \-\-min\-size=SIZE don't transfer any file smaller than SIZE @@ -631,7 +635,8 @@ specified, in which case \fB\-r\fP is no .IP Note that \fB\-a\fP \fBdoes not preserve hardlinks\fP, because finding multiply-linked files is expensive. You must separately -specify \fB\-H\fP. +specify \fB\-H\fP. Note also that for backward compatibility, \fB\- a\fP +curren
preserve ctimes of *unchanged* directories on receiver
'rsync -a' updates the ctime on a directory even if no file in that directory has changed. A kind of workaround is to use '-O', but then the mtimes of directories are not preserved. (Usage example where this is important: maintain a copy of filesystem A in filesystem B, and use filesystem B as the source for incremental backups (e.g., with star). rsync is run before an incremental backup is made. Because of the above described ctime problem, every incremental backup will contain *all* directories, even if filesystem A was not changed since the last sync and backup.) pgpn2ojRbgBeU.pgp Description: PGP signature -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html
crypting remote data
I don't know if this is the propper place to ask for new funcionality, but after searching the website i didn't find another place to do it. I'm very sorry if I`m asking this in the wrong place. It would be greatful to have crypting funcionality added to rsync, appart from using ssh to crypt transmissions. I know this is not the original purpose for rsync, but anyway it's a fact that many people need (at least i need, I hope I'm not alone) additional security features for offsite backups, like having the data crypted at remote, but not at source, so the problem is crypting the data and then syncronizing it. I've programmed a wrapper in java that do it, but it's very time and resources consuming. The idea was to copy file by file in a temp dir, crypt the file, rsync it, and delete the file. I go from top to down sending the source (subdirs). About sending the data file by file, the question is many times we have very large files and/or very large dirs, so imposing the user to have the same amount of free space that he/she wants to send it's not a good idea I guess. It isn't a bad solution, but neither a good one. Perfect should be having support to it from rsync. It is just an idea to improve rsync. Thank you very much in advance and again sorry if this wasn't the list to post a request. -- To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/rsync Before posting, read: http://www.catb.org/~esr/faqs/smart-questions.html