1 Added some DB-access routines to help rsync keep extra filesystem info
2 about the files it is dealing with. This adds both the --db=CONFIG_FILE
3 option and the "db config" daemon parameter.
5 For the moment this only adds checksum caching when the --checksum option
6 is used. Future improvements may include:
8 - Updating of MD5 checksums when transferring any file, even w/o -c.
9 We should be able to extend this to work for MD4 checksums too if we
10 make the sender force checksum_seed to 0 when using a DB and having
11 the receiving side check to see if it got a 0 checksum_seed. (We
12 probably don't want to compute 2 MD4 checksums for the case where
13 the checksum_seed is non-zero.)
15 - Caching of path info that allows for the finding of files to use for
16 moving/linking/copying/alternate-basis-use.
18 - Extend DB support beyond MySQL and SQLite (PostgreSQL?).
20 To use this patch, run these commands for a successful build:
22 patch -p1 <patches/db.diff
23 ./configure (optional if already run)
26 diff --git a/Makefile.in b/Makefile.in
27 index feacb90..f9da8eb 100644
30 @@ -36,7 +36,7 @@ ZLIBOBJ=zlib/deflate.o zlib/inffast.o zlib/inflate.o zlib/inftrees.o \
31 OBJS1=flist.o rsync.o generator.o receiver.o cleanup.o sender.o exclude.o \
32 util.o main.o checksum.o match.o syscall.o log.o backup.o delete.o
33 OBJS2=options.o io.o compat.o hlink.o token.o uidlist.o socket.o hashtable.o \
34 - fileio.o batch.o clientname.o chmod.o acls.o xattrs.o
35 + fileio.o batch.o clientname.o chmod.o db.o acls.o xattrs.o
36 OBJS3=progress.o pipe.o
37 DAEMON_OBJ = params.o loadparm.o clientserver.o access.o connection.o authenticate.o
38 popt_OBJS=popt/findme.o popt/popt.o popt/poptconfig.o \
39 diff --git a/checksum.c b/checksum.c
40 index 811b5b6..a7617af 100644
45 extern int checksum_seed;
46 extern int protocol_version;
50 a simple 32 bit checksum that can be upadted from either end
51 @@ -98,10 +99,10 @@ void get_checksum2(char *buf, int32 len, char *sum)
55 -void file_checksum(char *fname, char *sum, OFF_T size)
56 +void file_checksum(const char *fname, STRUCT_STAT *st_p, char *sum)
58 struct map_struct *buf;
59 - OFF_T i, len = size;
60 + OFF_T i, len = st_p->st_size;
64 @@ -112,7 +113,7 @@ void file_checksum(char *fname, char *sum, OFF_T size)
68 - buf = map_file(fd, size, MAX_MAP_SIZE, CSUM_CHUNK);
69 + buf = map_file(fd, len, MAX_MAP_SIZE, CSUM_CHUNK);
71 if (protocol_version >= 30) {
73 @@ -146,6 +147,9 @@ void file_checksum(char *fname, char *sum, OFF_T size)
74 mdfour_result(&m, (uchar *)sum);
78 + db_set_checksum(fname, st_p, sum);
83 diff --git a/cleanup.c b/cleanup.c
84 index 19ef072..ca46868 100644
92 extern int keep_partial;
93 extern int got_xfer_error;
94 extern int output_needs_newline;
95 @@ -130,6 +131,12 @@ NORETURN void _exit_cleanup(int code, const char *file, int line)
105 if (cleanup_child_pid != -1) {
107 int pid = wait_process(cleanup_child_pid, &status, WNOHANG);
108 diff --git a/clientserver.c b/clientserver.c
109 index b6afe00..37cad54 100644
112 @@ -42,13 +42,16 @@ extern int numeric_ids;
113 extern int filesfrom_fd;
114 extern int remote_protocol;
115 extern int protocol_version;
116 +extern int always_checksum;
117 extern int io_timeout;
118 extern int no_detach;
120 extern int write_batch;
121 extern int default_af_hint;
122 extern int logfile_format_has_i;
123 extern int logfile_format_has_o_or_i;
124 extern mode_t orig_umask;
125 +extern char *db_config;
126 extern char *bind_address;
127 extern char *config_file;
128 extern char *logfile_format;
129 @@ -648,6 +651,9 @@ static int rsync_module(int f_in, int f_out, int i, const char *addr, const char
133 + if (*lp_db_config(i))
134 + db_read_config(FLOG, lp_db_config(i));
137 if (*lp_prexfer_exec(i) || *lp_postxfer_exec(i)) {
138 char *modname, *modpath, *hostaddr, *hostname, *username;
139 @@ -856,6 +862,10 @@ static int rsync_module(int f_in, int f_out, int i, const char *addr, const char
141 am_server = 1; /* Don't let someone try to be tricky. */
144 + if (!always_checksum)
147 if (lp_ignore_errors(module_id))
150 diff --git a/configure.in b/configure.in
151 index bc7d4a7..43d51ff 100644
154 @@ -312,7 +312,7 @@ AC_CHECK_HEADERS(sys/fcntl.h sys/select.h fcntl.h sys/time.h sys/unistd.h \
155 sys/un.h sys/attr.h mcheck.h arpa/inet.h arpa/nameser.h locale.h \
156 netdb.h malloc.h float.h limits.h iconv.h libcharset.h langinfo.h \
157 sys/acl.h acl/libacl.h attr/xattr.h sys/xattr.h sys/extattr.h \
158 - popt.h popt/popt.h)
159 + popt.h popt/popt.h mysql/mysql.h sqlite3.h)
162 AC_CACHE_CHECK([if makedev takes 3 args],rsync_cv_MAKEDEV_TAKES_3_ARGS,[
163 @@ -977,6 +977,29 @@ if test x"$enable_acl_support" = x"no" -o x"$enable_xattr_support" = x"no" -o x"
167 +AC_CHECK_PROG(MYSQL_CONFIG, mysql_config, 1, 0)
168 +if test x$MYSQL_CONFIG = x1; then
169 + AC_MSG_CHECKING(for mysql version >= 4)
170 + mysql_version=`mysql_config --version`
171 + mysql_major_version=`echo $mysql_version | sed 's/\..*//'`
172 + if test $mysql_major_version -lt 4; then
173 + AC_MSG_RESULT(no.. skipping MySQL)
177 + MYSQL_CFLAGS=`mysql_config --cflags`
178 + MYSQL_LIBS=`mysql_config --libs`
180 + CPPFLAGS="$CPPFLAGS $MYSQL_CFLAGS"
181 + LIBS="$MYSQL_LIBS $LIBS"
183 + AC_CHECK_LIB(mysqlclient, mysql_init)
187 +AC_CHECK_LIB(sqlite3, sqlite3_open)
188 +AC_CHECK_FUNCS(sqlite3_open_v2 sqlite3_prepare_v2)
191 ' checker'*|checker*)
192 AC_DEFINE(FORCE_FD_ZERO_MEMSET, 1, [Used to make "checker" understand that FD_ZERO() clears memory.])
193 diff --git a/db.c b/db.c
195 index 0000000..6855488
200 + * Routines to access extended file info via DB.
202 + * Copyright (C) 2008 Wayne Davison
204 + * This program is free software; you can redistribute it and/or modify
205 + * it under the terms of the GNU General Public License as published by
206 + * the Free Software Foundation; either version 3 of the License, or
207 + * (at your option) any later version.
209 + * This program is distributed in the hope that it will be useful,
210 + * but WITHOUT ANY WARRANTY; without even the implied warranty of
211 + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
212 + * GNU General Public License for more details.
214 + * You should have received a copy of the GNU General Public License along
215 + * with this program; if not, visit the http://fsf.org website.
221 +#if defined HAVE_MYSQL_MYSQL_H && defined HAVE_LIBMYSQLCLIENT
223 +#include <mysql/mysql.h>
224 +#include <mysql/errmsg.h>
227 +#if defined HAVE_SQLITE3_H && defined HAVE_LIBSQLITE3
229 +#include <sqlite3.h>
230 +#ifndef HAVE_SQLITE3_OPEN_V2
231 +#define sqlite3_open_v2(dbname, dbhptr, flags, vfs) \
232 + sqlite3_open(dbname, dbhptr)
234 +#ifndef HAVE_SQLITE3_PREPARE_V2
235 +#define sqlite3_prepare_v2 sqlite3_prepare
239 +extern int protocol_version;
240 +extern int checksum_len;
242 +#define DB_TYPE_NONE 0
243 +#define DB_TYPE_MYSQL 1
244 +#define DB_TYPE_SQLITE 2
246 +int use_db = DB_TYPE_NONE;
248 +static const char *dbhost = NULL, *dbuser = NULL, *dbpass = NULL, *dbname = NULL;
249 +static unsigned int dbport = 0;
264 +#define MAX_PREP_CNT 3
271 + sqlite3_stmt *sqlite;
274 +} statements[MAX_PREP_CNT];
277 +static enum logcode log_code;
280 +static unsigned int bind_disk_id;
281 +static unsigned long long bind_devno, bind_ino, bind_size, bind_mtime, bind_ctime;
282 +static char bind_sum[MAX_DIGEST_LEN];
284 +static char bind_thishost[256];
285 +static int bind_thishost_len;
287 +static unsigned int prior_disk_id = 0;
288 +static unsigned long long prior_devno = 0;
290 +int db_read_config(enum logcode code, const char *config_file)
292 + char buf[2048], *cp;
298 + bind_thishost_len = strlcpy(bind_thishost, "localhost", sizeof bind_thishost);
300 + if (!(fp = fopen(config_file, "r"))) {
301 + rsyserr(log_code, errno, "unable to open %s", config_file);
304 + while (fgets(buf, sizeof buf, fp)) {
306 + if ((cp = strchr(buf, '#')) == NULL
307 + && (cp = strchr(buf, '\r')) == NULL
308 + && (cp = strchr(buf, '\n')) == NULL)
309 + cp = buf + strlen(buf);
310 + while (cp != buf && isSpace(cp-1)) cp--;
316 + if (!(cp = strchr(buf, ':')))
320 + while (isSpace(cp)) cp++;
321 + if (strcasecmp(buf, "dbhost") == 0)
322 + dbhost = strdup(cp);
323 + else if (strcasecmp(buf, "dbuser") == 0)
324 + dbuser = strdup(cp);
325 + else if (strcasecmp(buf, "dbpass") == 0)
326 + dbpass = strdup(cp);
327 + else if (strcasecmp(buf, "dbname") == 0)
328 + dbname = strdup(cp);
329 + else if (strcasecmp(buf, "dbport") == 0)
331 + else if (strcasecmp(buf, "thishost") == 0)
332 + bind_thishost_len = strlcpy(bind_thishost, cp, sizeof bind_thishost);
333 + else if (strcasecmp(buf, "dbtype") == 0) {
335 + if (strcasecmp(cp, "mysql") == 0) {
336 + use_db = DB_TYPE_MYSQL;
341 + if (strcasecmp(cp, "sqlite") == 0) {
342 + use_db = DB_TYPE_SQLITE;
347 + "Unsupported dbtype on line #%d in %s.\n",
348 + lineno, config_file);
349 + use_db = DB_TYPE_NONE;
353 + rprintf(log_code, "Invalid line #%d in %s\n",
354 + lineno, config_file);
355 + use_db = DB_TYPE_NONE;
361 + if (bind_thishost_len >= (int)sizeof bind_thishost)
362 + bind_thishost_len = sizeof bind_thishost - 1;
364 + if (!use_db || !dbname) {
365 + rprintf(log_code, "Please specify at least dbtype and dbname in %s.\n", config_file);
366 + use_db = DB_TYPE_NONE;
370 + md_num = protocol_version >= 30 ? 5 : 4;
376 +static MYSQL_STMT *prepare_mysql(MYSQL_BIND *binds, int bind_cnt, const char *fmt, ...)
380 + int qlen, param_cnt;
381 + MYSQL_STMT *stmt = mysql_stmt_init(dbh.mysql);
384 + out_of_memory("prepare_mysql");
387 + qlen = vasprintf(&query, fmt, ap);
390 + out_of_memory("prepare_mysql");
392 + if (mysql_stmt_prepare(stmt, query, qlen) != 0) {
393 + rprintf(log_code, "Prepare failed: %s\n", mysql_stmt_error(stmt));
398 + if ((param_cnt = mysql_stmt_param_count(stmt)) != bind_cnt) {
399 + rprintf(log_code, "Parameters in statement = %d, bind vars = %d\n",
400 + param_cnt, bind_cnt);
404 + mysql_stmt_bind_param(stmt, binds);
411 +static int db_connect_mysql(void)
413 + MYSQL_BIND binds[10];
415 + if (!(dbh.mysql = mysql_init(NULL)))
416 + out_of_memory("db_read_config");
418 + if (!mysql_real_connect(dbh.mysql, dbhost, dbuser, dbpass, dbname, dbport, NULL, 0))
421 + memset(binds, 0, sizeof binds);
422 + binds[0].buffer_type = MYSQL_TYPE_LONGLONG;
423 + binds[0].buffer = &bind_devno;
424 + binds[1].buffer_type = MYSQL_TYPE_STRING;
425 + binds[1].buffer = &bind_thishost;
426 + binds[1].buffer_length = bind_thishost_len;
427 + statements[SEL_DEV].mysql = prepare_mysql(binds, 2,
430 + " WHERE devno = ? AND host = ? AND mounted = 1");
431 + if (!statements[SEL_DEV].mysql)
434 + memset(binds, 0, sizeof binds);
435 + binds[0].buffer_type = MYSQL_TYPE_LONG;
436 + binds[0].buffer = &bind_disk_id;
437 + binds[1].buffer_type = MYSQL_TYPE_LONGLONG;
438 + binds[1].buffer = &bind_ino;
439 + binds[2].buffer_type = MYSQL_TYPE_LONGLONG;
440 + binds[2].buffer = &bind_size;
441 + binds[3].buffer_type = MYSQL_TYPE_LONGLONG;
442 + binds[3].buffer = &bind_mtime;
443 + binds[4].buffer_type = MYSQL_TYPE_LONGLONG;
444 + binds[4].buffer = &bind_ctime;
445 + statements[SEL_SUM].mysql = prepare_mysql(binds, 5,
448 + " WHERE disk_id = ? AND ino = ? AND sum_type = %d"
449 + " AND size = ? AND mtime = ? AND ctime = ?",
451 + if (!statements[SEL_SUM].mysql)
454 + memset(binds, 0, sizeof binds);
455 + binds[0].buffer_type = MYSQL_TYPE_LONG;
456 + binds[0].buffer = &bind_disk_id;
457 + binds[1].buffer_type = MYSQL_TYPE_LONGLONG;
458 + binds[1].buffer = &bind_ino;
459 + binds[2].buffer_type = binds[6].buffer_type = MYSQL_TYPE_LONGLONG;
460 + binds[2].buffer = binds[6].buffer = &bind_size;
461 + binds[3].buffer_type = binds[7].buffer_type = MYSQL_TYPE_LONGLONG;
462 + binds[3].buffer = binds[7].buffer = &bind_mtime;
463 + binds[4].buffer_type = binds[8].buffer_type = MYSQL_TYPE_LONGLONG;
464 + binds[4].buffer = binds[8].buffer = &bind_ctime;
465 + binds[5].buffer_type = binds[9].buffer_type = MYSQL_TYPE_BLOB;
466 + binds[5].buffer = binds[9].buffer = &bind_sum;
467 + binds[5].buffer_length = binds[9].buffer_length = checksum_len;
468 + statements[REP_SUM].mysql = prepare_mysql(binds, 10,
469 + "INSERT INTO inode_map"
470 + " SET disk_id = ?, ino = ?, sum_type = %d,"
471 + " size = ?, mtime = ?, ctime = ?, checksum = ?"
472 + " ON DUPLICATE KEY"
473 + " UPDATE size = ?, mtime = ?, ctime = ?, checksum = ?",
475 + if (!statements[REP_SUM].mysql)
483 +static int db_connect_sqlite(void)
487 + if (sqlite3_open_v2(dbname, &dbh.sqlite, SQLITE_OPEN_READWRITE, NULL) != 0)
490 + sql = "SELECT disk_id"
492 + " WHERE devno = ? AND host = ? AND mounted = 1";
493 + if (sqlite3_prepare_v2(dbh.sqlite, sql, -1, &statements[SEL_DEV].sqlite, NULL) != 0)
499 + " WHERE disk_id = ? AND ino = ? AND sum_type = %d"
500 + " AND size = ? AND mtime = ? AND ctime = ?",
502 + || sqlite3_prepare_v2(dbh.sqlite, sql, -1, &statements[SEL_SUM].sqlite, NULL) != 0)
507 + "INSERT OR REPLACE INTO inode_map"
508 + " (disk_id, ino, sum_type, size, mtime, ctime, checksum)"
509 + " VALUES(?, ?, %d, ?, ?, ?, ?)",
511 + || sqlite3_prepare_v2(dbh.sqlite, sql, -1, &statements[REP_SUM].sqlite, NULL) != 0)
519 +int db_connect(void)
523 + case DB_TYPE_MYSQL:
524 + if (db_connect_mysql())
529 + case DB_TYPE_SQLITE:
530 + if (db_connect_sqlite())
536 + rprintf(log_code, "Unable to connect to DB\n");
538 + use_db = DB_TYPE_NONE;
543 +void db_disconnect(void)
550 + for (ndx = 0; ndx < MAX_PREP_CNT; ndx++) {
551 + if (statements[ndx].all) {
554 + case DB_TYPE_MYSQL:
555 + mysql_stmt_close(statements[ndx].mysql);
559 + case DB_TYPE_SQLITE:
560 + sqlite3_finalize(statements[ndx].sqlite);
564 + statements[ndx].all = NULL;
570 + case DB_TYPE_MYSQL:
571 + mysql_close(dbh.mysql);
575 + case DB_TYPE_SQLITE:
576 + sqlite3_close(dbh.sqlite);
585 +static MYSQL_STMT *exec_mysql(int ndx)
587 + MYSQL_STMT *stmt = statements[ndx].mysql;
590 + if ((rc = mysql_stmt_execute(stmt)) == CR_SERVER_LOST) {
592 + if (db_connect()) {
593 + stmt = statements[ndx].mysql;
594 + rc = mysql_stmt_execute(stmt);
598 + rprintf(log_code, "SQL execute failed: %s\n", mysql_stmt_error(stmt));
607 +static int fetch_mysql(MYSQL_BIND *binds, int bind_cnt, int ndx)
609 + unsigned long length[32];
610 + my_bool is_null[32], error[32];
615 + exit_cleanup(RERR_UNSUPPORTED);
617 + if ((stmt = exec_mysql(ndx)) == NULL)
620 + for (i = 0; i < bind_cnt; i++) {
621 + binds[i].is_null = &is_null[i];
622 + binds[i].length = &length[i];
623 + binds[i].error = &error[i];
625 + mysql_stmt_bind_result(stmt, binds);
627 + if ((rc = mysql_stmt_fetch(stmt)) != 0) {
628 + if (rc != MYSQL_NO_DATA) {
629 + rprintf(log_code, "SELECT fetch failed: %s\n",
630 + mysql_stmt_error(stmt));
632 + mysql_stmt_free_result(stmt);
636 + mysql_stmt_free_result(stmt);
638 + return is_null[0] ? 0 : 1;
642 +static void get_disk_id(unsigned long long devno)
646 + case DB_TYPE_MYSQL: {
647 + MYSQL_BIND binds[1];
649 + bind_devno = devno; /* The one variable SEL_DEV input value. */
651 + /* Bind where to put the output. */
652 + binds[0].buffer_type = MYSQL_TYPE_LONG;
653 + binds[0].buffer = &prior_disk_id;
654 + if (!fetch_mysql(binds, 1, SEL_DEV))
660 + case DB_TYPE_SQLITE: {
661 + sqlite3_stmt *stmt = statements[SEL_DEV].sqlite;
662 + sqlite3_bind_int64(stmt, 1, devno);
663 + sqlite3_bind_text(stmt, 2, bind_thishost, bind_thishost_len, SQLITE_STATIC);
664 + if (sqlite3_step(stmt) == SQLITE_ROW)
665 + prior_disk_id = sqlite3_column_int(stmt, 0);
668 + sqlite3_reset(stmt);
674 + prior_devno = devno;
677 +int db_get_checksum(UNUSED(const char *fname), const STRUCT_STAT *st_p, char *sum)
679 + if (prior_devno != st_p->st_dev)
680 + get_disk_id(st_p->st_dev);
681 + if (prior_disk_id == 0)
686 + case DB_TYPE_MYSQL: {
687 + MYSQL_BIND binds[1];
689 + bind_disk_id = prior_disk_id;
690 + bind_ino = st_p->st_ino;
691 + bind_size = st_p->st_size;
692 + bind_mtime = st_p->st_mtime;
693 + bind_ctime = st_p->st_ctime;
695 + binds[0].buffer_type = MYSQL_TYPE_BLOB;
696 + binds[0].buffer = sum;
697 + binds[0].buffer_length = checksum_len;
698 + return fetch_mysql(binds, 1, SEL_SUM);
702 + case DB_TYPE_SQLITE: {
703 + sqlite3_stmt *stmt = statements[SEL_SUM].sqlite;
704 + sqlite3_bind_int(stmt, 1, prior_disk_id);
705 + sqlite3_bind_int64(stmt, 2, st_p->st_ino);
706 + sqlite3_bind_int64(stmt, 3, st_p->st_size);
707 + sqlite3_bind_int64(stmt, 4, st_p->st_mtime);
708 + sqlite3_bind_int64(stmt, 5, st_p->st_ctime);
709 + if (sqlite3_step(stmt) == SQLITE_ROW) {
710 + int len = sqlite3_column_bytes(stmt, 0);
711 + if (len > MAX_DIGEST_LEN)
712 + len = MAX_DIGEST_LEN;
713 + memcpy(sum, sqlite3_column_blob(stmt, 0), len);
714 + sqlite3_reset(stmt);
717 + sqlite3_reset(stmt);
726 +int db_set_checksum(UNUSED(const char *fname), const STRUCT_STAT *st_p, const char *sum)
728 + if (prior_devno != st_p->st_dev)
729 + get_disk_id(st_p->st_dev);
730 + if (prior_disk_id == 0)
735 + case DB_TYPE_MYSQL: {
736 + bind_disk_id = prior_disk_id;
737 + bind_ino = st_p->st_ino;
738 + bind_size = st_p->st_size;
739 + bind_mtime = st_p->st_mtime;
740 + bind_ctime = st_p->st_ctime;
741 + memcpy(bind_sum, sum, checksum_len);
743 + return exec_mysql(REP_SUM) != NULL;
747 + case DB_TYPE_SQLITE: {
749 + sqlite3_stmt *stmt = statements[REP_SUM].sqlite;
750 + sqlite3_bind_int(stmt, 1, prior_disk_id);
751 + sqlite3_bind_int64(stmt, 2, st_p->st_ino);
752 + sqlite3_bind_int64(stmt, 3, st_p->st_size);
753 + sqlite3_bind_int64(stmt, 4, st_p->st_mtime);
754 + sqlite3_bind_int64(stmt, 5, st_p->st_ctime);
755 + sqlite3_bind_blob(stmt, 6, sum, checksum_len, SQLITE_TRANSIENT);
756 + rc = sqlite3_step(stmt);
757 + sqlite3_reset(stmt);
758 + return rc == SQLITE_DONE;
765 diff --git a/flist.c b/flist.c
766 index 09b4fc5..8d280e9 100644
769 @@ -54,6 +54,7 @@ extern int preserve_specials;
770 extern int missing_args;
774 extern int eol_nulls;
775 extern int relative_paths;
776 extern int implied_dirs;
777 @@ -1267,11 +1268,8 @@ struct file_struct *make_file(const char *fname, struct file_list *flist,
778 extra_len += EXTRA_LEN;
781 - if (always_checksum && am_sender && S_ISREG(st.st_mode)) {
782 - file_checksum(thisname, tmp_sum, st.st_size);
783 - if (sender_keeps_checksum)
784 - extra_len += SUM_EXTRA_CNT * EXTRA_LEN;
786 + if (sender_keeps_checksum && S_ISREG(st.st_mode))
787 + extra_len += SUM_EXTRA_CNT * EXTRA_LEN;
789 #if EXTRA_ROUNDING > 0
790 if (extra_len & (EXTRA_ROUNDING * EXTRA_LEN))
791 @@ -1347,8 +1345,12 @@ struct file_struct *make_file(const char *fname, struct file_list *flist,
795 - if (sender_keeps_checksum && S_ISREG(st.st_mode))
796 - memcpy(F_SUM(file), tmp_sum, checksum_len);
797 + if (always_checksum && am_sender && S_ISREG(st.st_mode)) {
798 + if (!use_db || !db_get_checksum(thisname, &st, tmp_sum))
799 + file_checksum(thisname, &st, tmp_sum);
800 + if (sender_keeps_checksum)
801 + memcpy(F_SUM(file), tmp_sum, checksum_len);
805 F_NDX(file) = stats.num_dirs;
806 @@ -2010,6 +2012,9 @@ struct file_list *send_file_list(int f, int argc, char *argv[])
807 | (eol_nulls || reading_remotely ? RL_EOL_NULLS : 0);
808 int implied_dot_dir = 0;
813 rprintf(FLOG, "building file list\n");
814 if (show_filelist_p())
815 start_filelist_progress("building file list");
816 diff --git a/generator.c b/generator.c
817 index 12007a1..315463f 100644
820 @@ -60,6 +60,7 @@ extern int human_readable;
821 extern int ignore_existing;
822 extern int ignore_non_existing;
825 extern int append_mode;
826 extern int make_backups;
827 extern int csum_length;
828 @@ -531,7 +532,8 @@ int unchanged_file(char *fn, struct file_struct *file, STRUCT_STAT *st)
829 of the file time to determine whether to sync */
830 if (always_checksum > 0 && S_ISREG(st->st_mode)) {
831 char sum[MAX_DIGEST_LEN];
832 - file_checksum(fn, sum, st->st_size);
833 + if (!use_db || !db_get_checksum(fn, st, sum))
834 + file_checksum(fn, st, sum);
835 return memcmp(sum, F_SUM(file), checksum_len) == 0;
838 @@ -2022,6 +2024,9 @@ void generate_files(int f_out, const char *local_name)
842 + if (use_db && always_checksum)
845 /* Since we often fill up the outgoing socket and then just sit around
846 * waiting for the other 2 processes to do their thing, we don't want
847 * to exit on a timeout. If the data stops flowing, the receiver will
848 diff --git a/loadparm.c b/loadparm.c
849 index 8e48e6d..e6eaec8 100644
852 @@ -107,6 +107,7 @@ typedef struct {
860 @@ -181,6 +182,7 @@ static const all_vars Defaults = {
861 /* auth_users; */ NULL,
864 + /* db_config; */ NULL,
865 /* dont_compress; */ DEFAULT_DONT_COMPRESS,
867 /* exclude_from; */ NULL,
868 @@ -316,6 +318,7 @@ static struct parm_struct parm_table[] =
869 {"auth users", P_STRING, P_LOCAL, &Vars.l.auth_users, NULL,0},
870 {"charset", P_STRING, P_LOCAL, &Vars.l.charset, NULL,0},
871 {"comment", P_STRING, P_LOCAL, &Vars.l.comment, NULL,0},
872 + {"db config", P_STRING, P_LOCAL, &Vars.l.db_config, NULL,0},
873 {"dont compress", P_STRING, P_LOCAL, &Vars.l.dont_compress, NULL,0},
874 {"exclude from", P_STRING, P_LOCAL, &Vars.l.exclude_from, NULL,0},
875 {"exclude", P_STRING, P_LOCAL, &Vars.l.exclude, NULL,0},
876 @@ -396,6 +399,7 @@ FN_GLOBAL_INTEGER(lp_rsync_port, &Vars.g.rsync_port)
877 FN_LOCAL_STRING(lp_auth_users, auth_users)
878 FN_LOCAL_STRING(lp_charset, charset)
879 FN_LOCAL_STRING(lp_comment, comment)
880 +FN_LOCAL_STRING(lp_db_config, db_config)
881 FN_LOCAL_STRING(lp_dont_compress, dont_compress)
882 FN_LOCAL_STRING(lp_exclude, exclude)
883 FN_LOCAL_STRING(lp_exclude_from, exclude_from)
884 diff --git a/main.c b/main.c
885 index 2ef2f47..ac88e12 100644
888 @@ -49,6 +49,7 @@ extern int copy_unsafe_links;
889 extern int keep_dirlinks;
890 extern int preserve_hard_links;
891 extern int protocol_version;
892 +extern int always_checksum;
893 extern int file_total;
895 extern int xfer_dirs;
896 @@ -74,6 +75,7 @@ extern char *filesfrom_host;
897 extern char *partial_dir;
898 extern char *dest_option;
899 extern char *rsync_path;
900 +extern char *db_config;
901 extern char *shell_cmd;
902 extern char *batch_name;
903 extern char *password_file;
904 @@ -1565,6 +1567,9 @@ int main(int argc,char *argv[])
905 exit_cleanup(RERR_SYNTAX);
908 + if (db_config && always_checksum)
909 + db_read_config(FERROR, db_config);
912 set_nonblocking(STDIN_FILENO);
913 set_nonblocking(STDOUT_FILENO);
914 diff --git a/options.c b/options.c
915 index e7c6c61..d47cb7c 100644
918 @@ -92,6 +92,7 @@ int use_qsort = 0;
919 char *files_from = NULL;
920 int filesfrom_fd = -1;
921 char *filesfrom_host = NULL;
922 +char *db_config = NULL;
924 int protect_args = 0;
925 int human_readable = 1;
926 @@ -566,6 +567,7 @@ static void print_rsync_version(enum logcode f)
927 char const *links = "no ";
928 char const *iconv = "no ";
929 char const *ipv6 = "no ";
930 + char const *db = "no ";
931 STRUCT_STAT *dumstat;
933 #if SUBPROTOCOL_VERSION != 0
934 @@ -599,6 +601,11 @@ static void print_rsync_version(enum logcode f)
935 #if defined HAVE_LUTIMES && defined HAVE_UTIMES
938 +#if defined HAVE_MYSQL_MYSQL_H && defined HAVE_LIBMYSQLCLIENT
940 +#elif defined HAVE_SQLITE3_H && defined HAVE_LIBSQLITE3
944 rprintf(f, "%s version %s protocol version %d%s\n",
945 RSYNC_NAME, RSYNC_VERSION, PROTOCOL_VERSION, subprotocol);
946 @@ -612,8 +619,8 @@ static void print_rsync_version(enum logcode f)
947 (int)(sizeof (int64) * 8));
948 rprintf(f, " %ssocketpairs, %shardlinks, %ssymlinks, %sIPv6, batchfiles, %sinplace,\n",
949 got_socketpair, hardlinks, links, ipv6, have_inplace);
950 - rprintf(f, " %sappend, %sACLs, %sxattrs, %siconv, %ssymtimes\n",
951 - have_inplace, acls, xattrs, iconv, symtimes);
952 + rprintf(f, " %sappend, %sACLs, %sxattrs, %siconv, %ssymtimes, %sdb\n",
953 + have_inplace, acls, xattrs, iconv, symtimes, db);
955 #ifdef MAINTAINER_MODE
956 rprintf(f, "Panic Action: \"%s\"\n", get_panic_action());
957 @@ -661,6 +668,7 @@ void usage(enum logcode F)
958 rprintf(F," -q, --quiet suppress non-error messages\n");
959 rprintf(F," --no-motd suppress daemon-mode MOTD (see manpage caveat)\n");
960 rprintf(F," -c, --checksum skip based on checksum, not mod-time & size\n");
961 + rprintf(F," --db=CONFIG_FILE specify a CONFIG_FILE for DB checksums\n");
962 rprintf(F," -a, --archive archive mode; equals -rlptgoD (no -H,-A,-X)\n");
963 rprintf(F," --no-OPTION turn off an implied OPTION (e.g. --no-D)\n");
964 rprintf(F," -r, --recursive recurse into directories\n");
965 @@ -933,6 +941,7 @@ static struct poptOption long_options[] = {
966 {"checksum", 'c', POPT_ARG_VAL, &always_checksum, 1, 0, 0 },
967 {"no-checksum", 0, POPT_ARG_VAL, &always_checksum, 0, 0, 0 },
968 {"no-c", 0, POPT_ARG_VAL, &always_checksum, 0, 0, 0 },
969 + {"db", 0, POPT_ARG_STRING, &db_config, 0, 0, 0 },
970 {"block-size", 'B', POPT_ARG_LONG, &block_size, 0, 0, 0 },
971 {"compare-dest", 0, POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 },
972 {"copy-dest", 0, POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 },
973 diff --git a/pipe.c b/pipe.c
974 index a33117c..2f6aa21 100644
977 @@ -27,6 +27,9 @@ extern int am_server;
978 extern int blocking_io;
979 extern int filesfrom_fd;
980 extern int munge_symlinks;
981 +extern int always_checksum;
983 +extern char *db_config;
984 extern mode_t orig_umask;
985 extern char *logfile_name;
986 extern int remote_option_cnt;
987 @@ -143,6 +146,9 @@ pid_t local_child(int argc, char **argv, int *f_in, int *f_out,
994 if (remote_option_cnt) {
995 int rc = remote_option_cnt + 1;
996 const char **rv = remote_options;
997 @@ -150,6 +156,8 @@ pid_t local_child(int argc, char **argv, int *f_in, int *f_out,
999 exit_cleanup(RERR_SYNTAX);
1001 + if (db_config && always_checksum)
1002 + db_read_config(FERROR, db_config);
1005 if (dup2(to_child_pipe[0], STDIN_FILENO) < 0 ||
1006 diff --git a/rsync.yo b/rsync.yo
1007 index 941f7a5..1b81189 100644
1010 @@ -323,6 +323,7 @@ to the detailed description below for a complete description. verb(
1011 -q, --quiet suppress non-error messages
1012 --no-motd suppress daemon-mode MOTD (see caveat)
1013 -c, --checksum skip based on checksum, not mod-time & size
1014 + --db=CONFIG_FILE specify a CONFIG_FILE for DB checksums
1015 -a, --archive archive mode; equals -rlptgoD (no -H,-A,-X)
1016 --no-OPTION turn off an implied OPTION (e.g. --no-D)
1017 -r, --recursive recurse into directories
1018 @@ -584,6 +585,47 @@ checksum that is generated as the file is transferred, but that
1019 automatic after-the-transfer verification has nothing to do with this
1020 option's before-the-transfer "Does this file need to be updated?" check.
1022 +dit(bf(--db=CONFIG_FILE)) This option specifies a CONFIG_FILE to read
1023 +that holds connection details for a database of checksum information.
1024 +When combined with the bf(--checksum) (bf(-c)) option, rsync will try to
1025 +use cached checksum information from the DB, and will update it if it is
1028 +The currently supported DB choices are MySQL and SQLite. For example, a
1029 +MySQL configuration might look like this:
1031 +verb( dbtype: mysql
1037 + thishost: hostname )
1039 +And a SQLite configuration might look like this:
1041 +verb( dbtype: SQLite
1042 + dbname: /var/cache/rsync/sum.db )
1044 +This option only affects one side of a transfer. See the
1045 +bf(--remote-option) option for a way to specify the option for both
1046 +sides of the transfer (with each side reading the config file from
1047 +their local filesystem). For example:
1049 +verb( rsync -avc {-M,}--db=/etc/rsyncdb.conf src/ host:dest/ )
1051 +See the perl script "rsyncdb" in the support directory of the source code
1052 +(which may also be installed in /usr/bin) for a way to create the tables,
1053 +populate the mounted-disk information, check files against their checksums,
1054 +and update both the MD4 and MD5 checksums for files at the same time (since
1055 +an rsync copy will only update one or the other).
1057 +You can use a single MySQL DB for all your hosts if you give each one
1058 +their own "thishost" name and setup their device-mapping data. Or feel
1059 +free to use separate databases, separate servers, etc. See the rsync
1060 +daemon's "db config" parameter for how to configure a daemon to use a DB
1061 +(since a client cannot control this parameter on a daemon).
1063 dit(bf(-a, --archive)) This is equivalent to bf(-rlptgoD). It is a quick
1064 way of saying you want recursion and want to preserve almost
1065 everything (with -H being a notable omission).
1066 diff --git a/rsyncd.conf.yo b/rsyncd.conf.yo
1067 index d4978cd..68c588c 100644
1068 --- a/rsyncd.conf.yo
1069 +++ b/rsyncd.conf.yo
1070 @@ -281,6 +281,18 @@ is daemon. This setting has no effect if the "log file" setting is a
1071 non-empty string (either set in the per-modules settings, or inherited
1072 from the global settings).
1074 +dit(bf(db config)) This parameter specifies a config file to read that
1075 +holds connection details for a database of checksum information.
1077 +The config file will be read-in prior to any chroot restrictions, but
1078 +the connection occurs from inside the chroot. This means that you
1079 +should use a socket connection (e.g. 127.0.0.1 rather than localhost)
1080 +for a MySQL config from inside a chroot. For SQLite, the DB file must
1081 +be placed inside the chroot (though it can be placed outside the
1082 +transfer dir if you configured an inside-chroot path).
1084 +See the bf(--db=CONFIG_FILE) option for full details.
1086 dit(bf(max verbosity)) This parameter allows you to control
1087 the maximum amount of verbose information that you'll allow the daemon to
1088 generate (since the information goes into the log file). The default is 1,
1089 diff --git a/support/rsyncdb b/support/rsyncdb
1090 new file mode 100755
1091 index 0000000..801068c
1093 +++ b/support/rsyncdb
1100 +use Cwd qw(abs_path cwd);
1104 +my $MOUNT_FILE = '/etc/mtab';
1106 +&Getopt::Long::Configure('bundling');
1107 +&usage if !&GetOptions(
1108 + 'db=s' => \( my $db_config ),
1109 + 'init' => \( my $init_db ),
1110 + 'mounts|m' => \( my $update_mounts ),
1111 + 'recurse|r' => \( my $recurse_opt ),
1112 + 'check|c' => \( my $check_opt ),
1113 + 'verbose|v+' => \( my $verbosity = 0 ),
1114 + 'help|h' => \( my $help_opt ),
1116 +&usage if $help_opt || !defined $db_config;
1119 +open(IN, '<', $db_config) or die "Unable to open $db_config: $!\n";
1123 + my($key, $val) = /^(\S+):\s*(.*)/ or die "Unable to parse line $. of $db_config\n";
1124 + $config{$key} = $val;
1128 +die "You must define at least dbtype and dbname in $db_config\n"
1129 + unless defined $config{'dbtype'} && defined $config{'dbname'};
1131 +my $sqlite = $config{'dbtype'} =~ /^sqlite$/i;
1133 +my $thishost = $config{'thishost'} || 'localhost';
1135 +my $connect = 'DBI:' . $config{'dbtype'} . ':';
1136 +$connect .= 'dbname=' . $config{'dbname'} if $sqlite;
1137 +$connect .= 'database=' . $config{'dbname'} if !$sqlite && !$init_db;
1138 +$connect .= ';host=' . $config{'dbhost'} if defined $config{'dbhost'};
1139 +$connect .= ';port=' . $config{'dbport'} if defined $config{'dbport'};
1141 +my $dbh = DBI->connect($connect, $config{'dbuser'}, $config{'dbpass'})
1142 + or die "DB connection failed\n";
1145 + $dbh->disconnect if defined $dbh;
1149 + my $unsigned = $sqlite ? '' : 'unsigned';
1150 + my $auto_increment = $sqlite ? 'AUTOINCREMENT' : 'AUTO_INCREMENT';
1151 + my $dbname = $config{'dbname'};
1154 + $dbh->do("CREATE DATABASE IF NOT EXISTS `$dbname`");
1155 + $dbh->do("USE `$dbname`");
1158 + print "Dropping old tables (if they exist) ...\n" if $verbosity;
1159 + $dbh->do("DROP TABLE IF EXISTS disk") or die $dbh->errstr;
1160 + $dbh->do("DROP TABLE IF EXISTS inode_map") or die $dbh->errstr;
1162 + print "Creating empty tables ...\n" if $verbosity;
1164 + CREATE TABLE disk (
1165 + disk_id integer $unsigned NOT NULL PRIMARY KEY $auto_increment,
1166 + devno bigint $unsigned NOT NULL,
1167 + host varchar(256) NOT NULL default 'localhost',
1168 + mounted tinyint NOT NULL default '1',
1169 + comment varchar(256) default NULL
1170 + )") or die $dbh->errstr;
1173 + CREATE TABLE inode_map (
1174 + disk_id integer $unsigned NOT NULL,
1175 + ino bigint $unsigned NOT NULL,
1176 + size bigint $unsigned NOT NULL,
1177 + mtime bigint NOT NULL,
1178 + ctime bigint NOT NULL,
1179 + sum_type tinyint NOT NULL default '0',
1180 + checksum binary(16) NOT NULL,
1181 + PRIMARY KEY (disk_id,ino,sum_type)
1182 + )") or die $dbh->errstr;
1184 + exit unless $update_mounts;
1187 +my $sel_disk_H = $dbh->prepare("
1188 + SELECT disk_id, devno, mounted, comment
1191 + ") or die $dbh->errstr;
1193 +my $ins_disk_H = $dbh->prepare("
1195 + (devno, host, mounted, comment)
1196 + VALUES(?, ?, ?, ?)
1197 + ") or die $dbh->errstr;
1199 +my $up_disk_H = $dbh->prepare("
1203 + ") or die $dbh->errstr;
1205 +my $row_id = $sqlite ? 'ROWID' : 'ID';
1206 +my $sel_lastid_H = $dbh->prepare("
1207 + SELECT LAST_INSERT_$row_id()
1208 + ") or die $dbh->errstr;
1210 +my $sel_sum_H = $dbh->prepare("
1211 + SELECT sum_type, checksum
1213 + WHERE disk_id = ? AND ino = ? AND size = ? AND mtime = ? AND ctime = ?
1214 + ") or die $dbh->errstr;
1216 +my $rep_sum_H = $dbh->prepare("
1217 + REPLACE INTO inode_map
1218 + (disk_id, ino, size, mtime, ctime, sum_type, checksum)
1219 + VALUES(?, ?, ?, ?, ?, ?, ?)
1220 + ") or die $dbh->errstr;
1223 +if ($update_mounts) {
1224 + open(IN, $MOUNT_FILE) or die "Unable to open $MOUNT_FILE: $!\n";
1226 + my($devname, $mnt) = (split)[0,1];
1227 + next unless $devname =~ m#^/dev#;
1228 + my($devno) = (stat($mnt))[0];
1229 + if (!defined $devno) {
1230 + warn "Unable to stat $mnt: $!\n";
1233 + $mounts{$devno} = "$devname on $mnt";
1239 +$sel_disk_H->execute($thishost);
1240 +while (my($disk_id, $devno, $mounted, $comment) = $sel_disk_H->fetchrow_array) {
1241 + if ($update_mounts) {
1242 + if (defined $mounts{$devno}) {
1243 + if ($comment ne $mounts{$devno}) {
1245 + print "Umounting $comment ($thishost:$devno)\n" if $verbosity;
1246 + $up_disk_H->execute(0, $disk_id);
1251 + print "Mounting $comment ($thishost:$devno)\n" if $verbosity;
1252 + $up_disk_H->execute(1, $disk_id);
1256 + print "Umounting $comment ($thishost:$devno)\n" if $verbosity;
1257 + $up_disk_H->execute(0, $disk_id);
1262 + next unless $mounted;
1264 + $disk_id{$devno} = $disk_id;
1266 +$sel_disk_H->finish;
1268 +if ($update_mounts) {
1269 + while (my($devno, $comment) = each %mounts) {
1270 + next if $disk_id{$devno};
1271 + print "Adding $comment ($thishost:$devno)\n" if $verbosity;
1272 + $ins_disk_H->execute($devno, $thishost, 1, $comment);
1273 + $sel_lastid_H->execute;
1274 + ($disk_id{$devno}) = $sel_lastid_H->fetchrow_array;
1275 + $sel_lastid_H->finish;
1280 +my $start_dir = cwd();
1283 +@dirs = '.' unless @dirs;
1285 + $_ = abs_path($_);
1292 +my $md4 = Digest::MD4->new;
1293 +my $md5 = Digest::MD5->new;
1296 + my $dir = shift @dirs;
1298 + if (!chdir($dir)) {
1299 + warn "Unable to chdir to $dir: $!\n";
1302 + if (!opendir(DP, '.')) {
1303 + warn "Unable to opendir $dir: $!\n";
1307 + my $reldir = $dir;
1308 + $reldir =~ s#^$start_dir(/|$)# $1 ? '' : '.' #eo;
1309 + print "$reldir ... \n" if $verbosity;
1312 + while (defined(my $fn = readdir(DP))) {
1313 + next if $fn =~ /^\.\.?$/ || -l $fn;
1315 + push(@subdirs, "$dir/$fn") unless $fn =~ /^(CVS|\.svn|\.git|\.bzr)$/;
1320 + my($dev,$ino,$size,$mtime,$ctime) = (stat(_))[0,1,7,9,10];
1321 + my $disk_id = $disk_id{$dev} or next;
1322 + $sel_sum_H->execute($disk_id,$ino,$size,$mtime,$ctime) or die $!;
1323 + my($sum4, $dbsum4, $sum5, $dbsum5);
1325 + while (my($sum_type, $checksum) = $sel_sum_H->fetchrow_array) {
1326 + if ($sum_type == 4) {
1327 + $dbsum4 = $checksum;
1329 + } elsif ($sum_type == 5) {
1330 + $dbsum5 = $checksum;
1334 + $sel_sum_H->finish;
1336 + next if !$check_opt && $dbsumcnt == 2;
1338 + if (!$check_opt || $dbsumcnt || $verbosity > 2) {
1339 + if (!open(IN, $fn)) {
1340 + print STDERR "Unable to read $fn: $!\n";
1345 + while (sysread(IN, $_, 64*1024)) {
1349 + $sum4 = $md4->digest;
1350 + $sum5 = $md5->digest;
1351 + print ' ', unpack('H*', $sum4), ' ', unpack('H*', $sum5) if $verbosity > 2;
1352 + print " $fn" if $verbosity > 1;
1353 + my($ino2,$size2,$mtime2,$ctime2) = (stat(IN))[1,7,9,10];
1354 + last if $ino == $ino2 && $size == $size2 && $mtime == $mtime2 && $ctime == $ctime2;
1359 + sysseek(IN, 0, 0);
1360 + print " REREADING\n" if $verbosity > 1;
1364 + } elsif ($verbosity > 1) {
1370 + if ($dbsumcnt == 0) {
1371 + $dif = ' --MISSING--';
1374 + if (!defined $dbsum4) {
1375 + $dif .= ' -NO-MD4-';
1376 + } elsif ($sum4 ne $dbsum4) {
1377 + $dif .= ' -MD4-CHANGED-';
1379 + if (!defined $dbsum5) {
1380 + $dif .= ' ---NO-MD5---';
1381 + } elsif ($sum5 ne $dbsum5) {
1382 + $dif .= ' -MD5-CHANGED-';
1385 + print " ====OK====\n" if $verbosity > 1;
1388 + $dif =~ s/MD4-CHANGED MD5-//;
1390 + if ($verbosity < 2) {
1391 + print $verbosity ? ' ' : "$reldir/";
1397 + print "\n" if $verbosity > 1;
1398 + $rep_sum_H->execute($disk_id, $ino, $size, $mtime, $ctime, 4, $sum4);
1399 + $rep_sum_H->execute($disk_id, $ino, $size, $mtime, $ctime, 5, $sum5);
1405 + unshift(@dirs, sort @subdirs) if $recurse_opt;
1413 +Usage: rsyncsums --db=CONFIG_FILE [OPTIONS] [DIRS]
1416 + --db=FILE Specify the config FILE to read for the DB info.
1417 + --init Create (recreate) needed tables (making them empty).
1418 + No DIR scanning, but can be combined with --mounts.
1419 + -m, --mounts Update mount info. Does no DIR scanning.
1420 + -r, --recurse Scan files in subdirectories too.
1421 + -c, --check Check if the checksums are right (doesn't update).
1422 + -v, --verbose Mention what we're doing. Repeat for more info.
1423 + -h, --help Display this help message.