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/remote-option.diff
23 patch -p1 <patches/db.diff
24 ./configure (optional if already run)
27 diff --git a/Makefile.in b/Makefile.in
30 @@ -35,7 +35,7 @@ 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
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
44 extern int checksum_seed;
45 extern int protocol_version;
48 int csum_length = SHORT_SUM_LENGTH; /* initial value */
50 @@ -100,10 +101,10 @@ void get_checksum2(char *buf, int32 len, char *sum)
54 -void file_checksum(char *fname, char *sum, OFF_T size)
55 +void file_checksum(const char *fname, STRUCT_STAT *st_p, char *sum)
57 struct map_struct *buf;
58 - OFF_T i, len = size;
59 + OFF_T i, len = st_p->st_size;
63 @@ -114,7 +115,7 @@ void file_checksum(char *fname, char *sum, OFF_T size)
67 - buf = map_file(fd, size, MAX_MAP_SIZE, CSUM_CHUNK);
68 + buf = map_file(fd, len, MAX_MAP_SIZE, CSUM_CHUNK);
70 if (protocol_version >= 30) {
72 @@ -148,6 +149,9 @@ void file_checksum(char *fname, char *sum, OFF_T size)
73 mdfour_result(&m, (uchar *)sum);
77 + db_set_checksum(fname, st_p, sum);
82 diff --git a/cleanup.c b/cleanup.c
85 @@ -27,6 +27,7 @@ extern int am_daemon;
87 extern int keep_partial;
88 extern int got_xfer_error;
90 extern char *partial_dir;
91 extern char *logfile_name;
93 @@ -124,6 +125,12 @@ NORETURN void _exit_cleanup(int code, const char *file, int line)
103 if (cleanup_child_pid != -1) {
105 int pid = wait_process(cleanup_child_pid, &status, WNOHANG);
106 diff --git a/clientserver.c b/clientserver.c
109 @@ -42,13 +42,16 @@ extern int numeric_ids;
110 extern int filesfrom_fd;
111 extern int remote_protocol;
112 extern int protocol_version;
113 +extern int always_checksum;
114 extern int io_timeout;
115 extern int no_detach;
117 extern int write_batch;
118 extern int default_af_hint;
119 extern int logfile_format_has_i;
120 extern int logfile_format_has_o_or_i;
121 extern mode_t orig_umask;
122 +extern char *db_config;
123 extern char *bind_address;
124 extern char *sockopts;
125 extern char *config_file;
126 @@ -552,6 +555,9 @@ static int rsync_module(int f_in, int f_out, int i, char *addr, char *host)
130 + if (*lp_db_config(i))
131 + db_read_config(FLOG, lp_db_config(i));
134 if (*lp_prexfer_exec(i) || *lp_postxfer_exec(i)) {
135 char *modname, *modpath, *hostaddr, *hostname, *username;
136 @@ -768,6 +774,10 @@ static int rsync_module(int f_in, int f_out, int i, char *addr, char *host)
138 am_server = 1; /* Don't let someone try to be tricky. */
141 + if (!always_checksum)
144 if (lp_ignore_errors(module_id))
147 diff --git a/configure.in b/configure.in
150 @@ -314,7 +314,7 @@ AC_CHECK_HEADERS(sys/fcntl.h sys/select.h fcntl.h sys/time.h sys/unistd.h \
151 sys/un.h sys/attr.h mcheck.h arpa/inet.h arpa/nameser.h locale.h \
152 netdb.h malloc.h float.h limits.h iconv.h libcharset.h langinfo.h \
153 sys/acl.h acl/libacl.h attr/xattr.h sys/xattr.h sys/extattr.h \
154 - popt.h popt/popt.h)
155 + popt.h popt/popt.h mysql/mysql.h sqlite3.h)
158 AC_CACHE_CHECK([if makedev takes 3 args],rsync_cv_MAKEDEV_TAKES_3_ARGS,[
159 @@ -969,6 +969,29 @@ if test x"$enable_acl_support" = x"no" -o x"$enable_xattr_support" = x"no" -o x"
163 +AC_CHECK_PROG(MYSQL_CONFIG, mysql_config, 1, 0)
164 +if test x$MYSQL_CONFIG = x1; then
165 + AC_MSG_CHECKING(for mysql version >= 4)
166 + mysql_version=`mysql_config --version`
167 + mysql_major_version=`echo $mysql_version | sed 's/\..*//'`
168 + if test $mysql_major_version -lt 4; then
169 + AC_MSG_RESULT(no.. skipping MySQL)
173 + MYSQL_CFLAGS=`mysql_config --cflags`
174 + MYSQL_LIBS=`mysql_config --libs`
176 + CPPFLAGS="$CPPFLAGS $MYSQL_CFLAGS"
177 + LIBS="$MYSQL_LIBS $LIBS"
179 + AC_CHECK_LIB(mysqlclient, mysql_init)
183 +AC_CHECK_LIB(sqlite3, sqlite3_open)
184 +AC_CHECK_FUNCS(sqlite3_open_v2 sqlite3_prepare_v2)
187 ' checker'*|checker*)
188 AC_DEFINE(FORCE_FD_ZERO_MEMSET, 1, [Used to make "checker" understand that FD_ZERO() clears memory.])
189 diff --git a/db.c b/db.c
195 + * Routines to access extended file info via DB.
197 + * Copyright (C) 2008 Wayne Davison
199 + * This program is free software; you can redistribute it and/or modify
200 + * it under the terms of the GNU General Public License as published by
201 + * the Free Software Foundation; either version 3 of the License, or
202 + * (at your option) any later version.
204 + * This program is distributed in the hope that it will be useful,
205 + * but WITHOUT ANY WARRANTY; without even the implied warranty of
206 + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
207 + * GNU General Public License for more details.
209 + * You should have received a copy of the GNU General Public License along
210 + * with this program; if not, visit the http://fsf.org website.
216 +#if defined HAVE_MYSQL_MYSQL_H && defined HAVE_LIBMYSQLCLIENT
218 +#include <mysql/mysql.h>
219 +#include <mysql/errmsg.h>
222 +#if defined HAVE_SQLITE3_H && defined HAVE_LIBSQLITE3
224 +#include <sqlite3.h>
225 +#ifndef HAVE_SQLITE3_OPEN_V2
226 +#define sqlite3_open_v2(dbname, dbhptr, flags, vfs) \
227 + sqlite3_open(dbname, dbhptr)
229 +#ifndef HAVE_SQLITE3_PREPARE_V2
230 +#define sqlite3_prepare_v2 sqlite3_prepare
234 +extern int protocol_version;
235 +extern int checksum_len;
237 +#define DB_TYPE_NONE 0
238 +#define DB_TYPE_MYSQL 1
239 +#define DB_TYPE_SQLITE 2
241 +int use_db = DB_TYPE_NONE;
243 +static const char *dbhost = NULL, *dbuser = NULL, *dbpass = NULL, *dbname = NULL;
244 +static unsigned int dbport = 0;
259 +#define MAX_PREP_CNT 3
266 + sqlite3_stmt *sqlite;
269 +} statements[MAX_PREP_CNT];
272 +static enum logcode log_code;
275 +static unsigned int bind_disk_id;
276 +static unsigned long long bind_devno, bind_ino, bind_size, bind_mtime, bind_ctime;
277 +static char bind_sum[MAX_DIGEST_LEN];
279 +static char bind_thishost[256];
280 +static int bind_thishost_len;
282 +static unsigned int prior_disk_id = 0;
283 +static unsigned long long prior_devno = 0;
285 +int db_read_config(enum logcode code, const char *config_file)
287 + char buf[2048], *cp;
293 + bind_thishost_len = strlcpy(bind_thishost, "localhost", sizeof bind_thishost);
295 + if (!(fp = fopen(config_file, "r"))) {
296 + rsyserr(log_code, errno, "unable to open %s", config_file);
299 + while (fgets(buf, sizeof buf, fp)) {
301 + if ((cp = strchr(buf, '#')) == NULL
302 + && (cp = strchr(buf, '\r')) == NULL
303 + && (cp = strchr(buf, '\n')) == NULL)
304 + cp = buf + strlen(buf);
305 + while (cp != buf && isSpace(cp-1)) cp--;
311 + if (!(cp = strchr(buf, ':')))
315 + while (isSpace(cp)) cp++;
316 + if (strcasecmp(buf, "dbhost") == 0)
317 + dbhost = strdup(cp);
318 + else if (strcasecmp(buf, "dbuser") == 0)
319 + dbuser = strdup(cp);
320 + else if (strcasecmp(buf, "dbpass") == 0)
321 + dbpass = strdup(cp);
322 + else if (strcasecmp(buf, "dbname") == 0)
323 + dbname = strdup(cp);
324 + else if (strcasecmp(buf, "dbport") == 0)
326 + else if (strcasecmp(buf, "thishost") == 0)
327 + bind_thishost_len = strlcpy(bind_thishost, cp, sizeof bind_thishost);
328 + else if (strcasecmp(buf, "dbtype") == 0) {
330 + if (strcasecmp(cp, "mysql") == 0) {
331 + use_db = DB_TYPE_MYSQL;
336 + if (strcasecmp(cp, "sqlite") == 0) {
337 + use_db = DB_TYPE_SQLITE;
342 + "Unsupported dbtype on line #%d in %s.\n",
343 + lineno, config_file);
344 + use_db = DB_TYPE_NONE;
348 + rprintf(log_code, "Invalid line #%d in %s\n",
349 + lineno, config_file);
350 + use_db = DB_TYPE_NONE;
356 + if (bind_thishost_len >= (int)sizeof bind_thishost)
357 + bind_thishost_len = sizeof bind_thishost - 1;
359 + if (!use_db || !dbname) {
360 + rprintf(log_code, "Please specify at least dbtype and dbname in %s.\n", config_file);
361 + use_db = DB_TYPE_NONE;
365 + md_num = protocol_version >= 30 ? 5 : 4;
371 +static MYSQL_STMT *prepare_mysql(MYSQL_BIND *binds, int bind_cnt, const char *fmt, ...)
375 + int qlen, param_cnt;
376 + MYSQL_STMT *stmt = mysql_stmt_init(dbh.mysql);
379 + out_of_memory("prepare_mysql");
382 + qlen = vasprintf(&query, fmt, ap);
385 + out_of_memory("prepare_mysql");
387 + if (mysql_stmt_prepare(stmt, query, qlen) != 0) {
388 + rprintf(log_code, "Prepare failed: %s\n", mysql_stmt_error(stmt));
393 + if ((param_cnt = mysql_stmt_param_count(stmt)) != bind_cnt) {
394 + rprintf(log_code, "Parameters in statement = %d, bind vars = %d\n",
395 + param_cnt, bind_cnt);
399 + mysql_stmt_bind_param(stmt, binds);
406 +static int db_connect_mysql(void)
408 + MYSQL_BIND binds[10];
410 + if (!(dbh.mysql = mysql_init(NULL)))
411 + out_of_memory("db_read_config");
413 + if (!mysql_real_connect(dbh.mysql, dbhost, dbuser, dbpass, dbname, dbport, NULL, 0))
416 + memset(binds, 0, sizeof binds);
417 + binds[0].buffer_type = MYSQL_TYPE_LONGLONG;
418 + binds[0].buffer = &bind_devno;
419 + binds[1].buffer_type = MYSQL_TYPE_STRING;
420 + binds[1].buffer = &bind_thishost;
421 + binds[1].buffer_length = bind_thishost_len;
422 + statements[SEL_DEV].mysql = prepare_mysql(binds, 2,
425 + " WHERE devno = ? AND host = ? AND mounted = 1");
426 + if (!statements[SEL_DEV].mysql)
429 + memset(binds, 0, sizeof binds);
430 + binds[0].buffer_type = MYSQL_TYPE_LONG;
431 + binds[0].buffer = &bind_disk_id;
432 + binds[1].buffer_type = MYSQL_TYPE_LONGLONG;
433 + binds[1].buffer = &bind_ino;
434 + binds[2].buffer_type = MYSQL_TYPE_LONGLONG;
435 + binds[2].buffer = &bind_size;
436 + binds[3].buffer_type = MYSQL_TYPE_LONGLONG;
437 + binds[3].buffer = &bind_mtime;
438 + binds[4].buffer_type = MYSQL_TYPE_LONGLONG;
439 + binds[4].buffer = &bind_ctime;
440 + statements[SEL_SUM].mysql = prepare_mysql(binds, 5,
443 + " WHERE disk_id = ? AND ino = ? AND sum_type = %d"
444 + " AND size = ? AND mtime = ? AND ctime = ?",
446 + if (!statements[SEL_SUM].mysql)
449 + memset(binds, 0, sizeof binds);
450 + binds[0].buffer_type = MYSQL_TYPE_LONG;
451 + binds[0].buffer = &bind_disk_id;
452 + binds[1].buffer_type = MYSQL_TYPE_LONGLONG;
453 + binds[1].buffer = &bind_ino;
454 + binds[2].buffer_type = binds[6].buffer_type = MYSQL_TYPE_LONGLONG;
455 + binds[2].buffer = binds[6].buffer = &bind_size;
456 + binds[3].buffer_type = binds[7].buffer_type = MYSQL_TYPE_LONGLONG;
457 + binds[3].buffer = binds[7].buffer = &bind_mtime;
458 + binds[4].buffer_type = binds[8].buffer_type = MYSQL_TYPE_LONGLONG;
459 + binds[4].buffer = binds[8].buffer = &bind_ctime;
460 + binds[5].buffer_type = binds[9].buffer_type = MYSQL_TYPE_BLOB;
461 + binds[5].buffer = binds[9].buffer = &bind_sum;
462 + binds[5].buffer_length = binds[9].buffer_length = checksum_len;
463 + statements[REP_SUM].mysql = prepare_mysql(binds, 10,
464 + "INSERT INTO inode_map"
465 + " SET disk_id = ?, ino = ?, sum_type = %d,"
466 + " size = ?, mtime = ?, ctime = ?, checksum = ?"
467 + " ON DUPLICATE KEY"
468 + " UPDATE size = ?, mtime = ?, ctime = ?, checksum = ?",
470 + if (!statements[REP_SUM].mysql)
478 +static int db_connect_sqlite(void)
482 + if (sqlite3_open_v2(dbname, &dbh.sqlite, SQLITE_OPEN_READWRITE, NULL) != 0)
485 + sql = "SELECT disk_id"
487 + " WHERE devno = ? AND host = ? AND mounted = 1";
488 + if (sqlite3_prepare_v2(dbh.sqlite, sql, -1, &statements[SEL_DEV].sqlite, NULL) != 0)
494 + " WHERE disk_id = ? AND ino = ? AND sum_type = %d"
495 + " AND size = ? AND mtime = ? AND ctime = ?",
497 + || sqlite3_prepare_v2(dbh.sqlite, sql, -1, &statements[SEL_SUM].sqlite, NULL) != 0)
502 + "INSERT OR REPLACE INTO inode_map"
503 + " (disk_id, ino, sum_type, size, mtime, ctime, checksum)"
504 + " VALUES(?, ?, %d, ?, ?, ?, ?)",
506 + || sqlite3_prepare_v2(dbh.sqlite, sql, -1, &statements[REP_SUM].sqlite, NULL) != 0)
514 +int db_connect(void)
518 + case DB_TYPE_MYSQL:
519 + if (db_connect_mysql())
524 + case DB_TYPE_SQLITE:
525 + if (db_connect_sqlite())
531 + rprintf(log_code, "Unable to connect to DB\n");
533 + use_db = DB_TYPE_NONE;
538 +void db_disconnect(void)
545 + for (ndx = 0; ndx < MAX_PREP_CNT; ndx++) {
546 + if (statements[ndx].all) {
549 + case DB_TYPE_MYSQL:
550 + mysql_stmt_close(statements[ndx].mysql);
554 + case DB_TYPE_SQLITE:
555 + sqlite3_finalize(statements[ndx].sqlite);
559 + statements[ndx].all = NULL;
565 + case DB_TYPE_MYSQL:
566 + mysql_close(dbh.mysql);
570 + case DB_TYPE_SQLITE:
571 + sqlite3_close(dbh.sqlite);
580 +static MYSQL_STMT *exec_mysql(int ndx)
582 + MYSQL_STMT *stmt = statements[ndx].mysql;
585 + if ((rc = mysql_stmt_execute(stmt)) == CR_SERVER_LOST) {
587 + if (db_connect()) {
588 + stmt = statements[ndx].mysql;
589 + rc = mysql_stmt_execute(stmt);
593 + rprintf(log_code, "SQL execute failed: %s\n", mysql_stmt_error(stmt));
602 +static int fetch_mysql(MYSQL_BIND *binds, int bind_cnt, int ndx)
604 + unsigned long length[32];
605 + my_bool is_null[32], error[32];
610 + exit_cleanup(RERR_UNSUPPORTED);
612 + if ((stmt = exec_mysql(ndx)) == NULL)
615 + for (i = 0; i < bind_cnt; i++) {
616 + binds[i].is_null = &is_null[i];
617 + binds[i].length = &length[i];
618 + binds[i].error = &error[i];
620 + mysql_stmt_bind_result(stmt, binds);
622 + if ((rc = mysql_stmt_fetch(stmt)) != 0) {
623 + if (rc != MYSQL_NO_DATA) {
624 + rprintf(log_code, "SELECT fetch failed: %s\n",
625 + mysql_stmt_error(stmt));
627 + mysql_stmt_free_result(stmt);
631 + mysql_stmt_free_result(stmt);
633 + return is_null[0] ? 0 : 1;
637 +static void get_disk_id(unsigned long long devno)
641 + case DB_TYPE_MYSQL: {
642 + MYSQL_BIND binds[1];
644 + bind_devno = devno; /* The one variable SEL_DEV input value. */
646 + /* Bind where to put the output. */
647 + binds[0].buffer_type = MYSQL_TYPE_LONG;
648 + binds[0].buffer = &prior_disk_id;
649 + if (!fetch_mysql(binds, 1, SEL_DEV))
655 + case DB_TYPE_SQLITE: {
656 + sqlite3_stmt *stmt = statements[SEL_DEV].sqlite;
657 + sqlite3_bind_int64(stmt, 1, devno);
658 + sqlite3_bind_text(stmt, 2, bind_thishost, bind_thishost_len, SQLITE_STATIC);
659 + if (sqlite3_step(stmt) == SQLITE_ROW)
660 + prior_disk_id = sqlite3_column_int(stmt, 0);
663 + sqlite3_reset(stmt);
669 + prior_devno = devno;
672 +int db_get_checksum(UNUSED(const char *fname), const STRUCT_STAT *st_p, char *sum)
674 + if (prior_devno != st_p->st_dev)
675 + get_disk_id(st_p->st_dev);
676 + if (prior_disk_id == 0)
681 + case DB_TYPE_MYSQL: {
682 + MYSQL_BIND binds[1];
684 + bind_disk_id = prior_disk_id;
685 + bind_ino = st_p->st_ino;
686 + bind_size = st_p->st_size;
687 + bind_mtime = st_p->st_mtime;
688 + bind_ctime = st_p->st_ctime;
690 + binds[0].buffer_type = MYSQL_TYPE_BLOB;
691 + binds[0].buffer = sum;
692 + binds[0].buffer_length = checksum_len;
693 + return fetch_mysql(binds, 1, SEL_SUM);
697 + case DB_TYPE_SQLITE: {
698 + sqlite3_stmt *stmt = statements[SEL_SUM].sqlite;
699 + sqlite3_bind_int(stmt, 1, prior_disk_id);
700 + sqlite3_bind_int64(stmt, 2, st_p->st_ino);
701 + sqlite3_bind_int64(stmt, 3, st_p->st_size);
702 + sqlite3_bind_int64(stmt, 4, st_p->st_mtime);
703 + sqlite3_bind_int64(stmt, 5, st_p->st_ctime);
704 + if (sqlite3_step(stmt) == SQLITE_ROW) {
705 + int len = sqlite3_column_bytes(stmt, 0);
706 + if (len > MAX_DIGEST_LEN)
707 + len = MAX_DIGEST_LEN;
708 + memcpy(sum, sqlite3_column_blob(stmt, 0), len);
709 + sqlite3_reset(stmt);
712 + sqlite3_reset(stmt);
721 +int db_set_checksum(UNUSED(const char *fname), const STRUCT_STAT *st_p, const char *sum)
723 + if (prior_devno != st_p->st_dev)
724 + get_disk_id(st_p->st_dev);
725 + if (prior_disk_id == 0)
730 + case DB_TYPE_MYSQL: {
731 + bind_disk_id = prior_disk_id;
732 + bind_ino = st_p->st_ino;
733 + bind_size = st_p->st_size;
734 + bind_mtime = st_p->st_mtime;
735 + bind_ctime = st_p->st_ctime;
736 + memcpy(bind_sum, sum, checksum_len);
738 + return exec_mysql(REP_SUM) != NULL;
742 + case DB_TYPE_SQLITE: {
744 + sqlite3_stmt *stmt = statements[REP_SUM].sqlite;
745 + sqlite3_bind_int(stmt, 1, prior_disk_id);
746 + sqlite3_bind_int64(stmt, 2, st_p->st_ino);
747 + sqlite3_bind_int64(stmt, 3, st_p->st_size);
748 + sqlite3_bind_int64(stmt, 4, st_p->st_mtime);
749 + sqlite3_bind_int64(stmt, 5, st_p->st_ctime);
750 + sqlite3_bind_blob(stmt, 6, sum, checksum_len, SQLITE_TRANSIENT);
751 + rc = sqlite3_step(stmt);
752 + sqlite3_reset(stmt);
753 + return rc == SQLITE_DONE;
760 diff --git a/flist.c b/flist.c
763 @@ -54,6 +54,7 @@ extern int preserve_devices;
764 extern int preserve_specials;
768 extern int eol_nulls;
769 extern int relative_paths;
770 extern int implied_dirs;
771 @@ -1250,14 +1251,16 @@ struct file_struct *make_file(const char *fname, struct file_list *flist,
772 memcpy(bp + basename_len, linkname, linkname_len);
775 - if (always_checksum && am_sender && S_ISREG(st.st_mode))
776 - file_checksum(thisname, tmp_sum, st.st_size);
779 F_PATHNAME(file) = pathname;
781 F_DEPTH(file) = extra_len / EXTRA_LEN;
783 + if (always_checksum && am_sender && S_ISREG(st.st_mode)) {
784 + if (!use_db || !db_get_checksum(thisname, &st, tmp_sum))
785 + file_checksum(thisname, &st, tmp_sum);
788 /* This code is only used by the receiver when it is building
789 * a list of files for a delete pass. */
790 if (keep_dirlinks && linkname_len && flist) {
791 @@ -1877,6 +1880,9 @@ struct file_list *send_file_list(int f, int argc, char *argv[])
792 | (eol_nulls || reading_remotely ? RL_EOL_NULLS : 0);
793 int implied_dot_dir = 0;
798 rprintf(FLOG, "building file list\n");
799 if (show_filelist_p())
800 start_filelist_progress("building file list");
801 diff --git a/generator.c b/generator.c
804 @@ -58,6 +58,7 @@ extern int update_only;
805 extern int ignore_existing;
806 extern int ignore_non_existing;
809 extern int append_mode;
810 extern int make_backups;
811 extern int csum_length;
812 @@ -721,7 +722,8 @@ int unchanged_file(char *fn, struct file_struct *file, STRUCT_STAT *st)
813 of the file time to determine whether to sync */
814 if (always_checksum > 0 && S_ISREG(st->st_mode)) {
815 char sum[MAX_DIGEST_LEN];
816 - file_checksum(fn, sum, st->st_size);
817 + if (!use_db || !db_get_checksum(fn, st, sum))
818 + file_checksum(fn, st, sum);
819 return memcmp(sum, F_SUM(file), checksum_len) == 0;
822 @@ -2204,6 +2206,9 @@ void generate_files(int f_out, const char *local_name)
826 + if (use_db && always_checksum)
829 /* Since we often fill up the outgoing socket and then just sit around
830 * waiting for the other 2 processes to do their thing, we don't want
831 * to exit on a timeout. If the data stops flowing, the receiver will
832 diff --git a/loadparm.c b/loadparm.c
835 @@ -126,6 +126,7 @@ typedef struct
843 @@ -177,6 +178,7 @@ static service sDefault =
844 /* auth_users; */ NULL,
847 + /* db_config; */ NULL,
848 /* dont_compress; */ DEFAULT_DONT_COMPRESS,
850 /* exclude_from; */ NULL,
851 @@ -307,6 +309,7 @@ static struct parm_struct parm_table[] =
852 {"auth users", P_STRING, P_LOCAL, &sDefault.auth_users, NULL,0},
853 {"charset", P_STRING, P_LOCAL, &sDefault.charset, NULL,0},
854 {"comment", P_STRING, P_LOCAL, &sDefault.comment, NULL,0},
855 + {"db config", P_STRING, P_LOCAL, &sDefault.db_config, NULL,0},
856 {"dont compress", P_STRING, P_LOCAL, &sDefault.dont_compress, NULL,0},
857 {"exclude from", P_STRING, P_LOCAL, &sDefault.exclude_from, NULL,0},
858 {"exclude", P_STRING, P_LOCAL, &sDefault.exclude, NULL,0},
859 @@ -400,6 +403,7 @@ FN_GLOBAL_INTEGER(lp_rsync_port, &Globals.rsync_port)
860 FN_LOCAL_STRING(lp_auth_users, auth_users)
861 FN_LOCAL_STRING(lp_charset, charset)
862 FN_LOCAL_STRING(lp_comment, comment)
863 +FN_LOCAL_STRING(lp_db_config, db_config)
864 FN_LOCAL_STRING(lp_dont_compress, dont_compress)
865 FN_LOCAL_STRING(lp_exclude, exclude)
866 FN_LOCAL_STRING(lp_exclude_from, exclude_from)
867 diff --git a/main.c b/main.c
870 @@ -49,6 +49,7 @@ extern int copy_unsafe_links;
871 extern int keep_dirlinks;
872 extern int preserve_hard_links;
873 extern int protocol_version;
874 +extern int always_checksum;
875 extern int file_total;
877 extern int xfer_dirs;
878 @@ -73,6 +74,7 @@ extern char *partial_dir;
879 extern char *dest_option;
880 extern char *basis_dir[];
881 extern char *rsync_path;
882 +extern char *db_config;
883 extern char *shell_cmd;
884 extern char *batch_name;
885 extern char *password_file;
886 @@ -1491,6 +1493,9 @@ int main(int argc,char *argv[])
887 exit_cleanup(RERR_SYNTAX);
890 + if (db_config && always_checksum)
891 + db_read_config(FERROR, db_config);
894 set_nonblocking(STDIN_FILENO);
895 set_nonblocking(STDOUT_FILENO);
896 diff --git a/options.c b/options.c
899 @@ -92,6 +92,7 @@ int use_qsort = 0;
900 char *files_from = NULL;
901 int filesfrom_fd = -1;
902 char *filesfrom_host = NULL;
903 +char *db_config = NULL;
905 int protect_args = 0;
906 int human_readable = 0;
907 @@ -229,6 +230,7 @@ static void print_rsync_version(enum logcode f)
908 char const *links = "no ";
909 char const *iconv = "no ";
910 char const *ipv6 = "no ";
911 + char const *db = "no ";
912 STRUCT_STAT *dumstat;
914 #if SUBPROTOCOL_VERSION != 0
915 @@ -261,6 +263,11 @@ static void print_rsync_version(enum logcode f)
916 #if defined HAVE_LUTIMES && defined HAVE_UTIMES
919 +#if defined HAVE_MYSQL_MYSQL_H && defined HAVE_LIBMYSQLCLIENT
921 +#elif defined HAVE_SQLITE3_H && defined HAVE_LIBSQLITE3
925 rprintf(f, "%s version %s protocol version %d%s\n",
926 RSYNC_NAME, RSYNC_VERSION, PROTOCOL_VERSION, subprotocol);
927 @@ -274,8 +281,8 @@ static void print_rsync_version(enum logcode f)
928 (int)(sizeof (int64) * 8));
929 rprintf(f, " %ssocketpairs, %shardlinks, %ssymlinks, %sIPv6, batchfiles, %sinplace,\n",
930 got_socketpair, hardlinks, links, ipv6, have_inplace);
931 - rprintf(f, " %sappend, %sACLs, %sxattrs, %siconv, %ssymtimes\n",
932 - have_inplace, acls, xattrs, iconv, symtimes);
933 + rprintf(f, " %sappend, %sACLs, %sxattrs, %siconv, %ssymtimes, %sdb\n",
934 + have_inplace, acls, xattrs, iconv, symtimes, db);
936 #ifdef MAINTAINER_MODE
937 rprintf(f, "Panic Action: \"%s\"\n", get_panic_action());
938 @@ -321,6 +328,7 @@ void usage(enum logcode F)
939 rprintf(F," -q, --quiet suppress non-error messages\n");
940 rprintf(F," --no-motd suppress daemon-mode MOTD (see manpage caveat)\n");
941 rprintf(F," -c, --checksum skip based on checksum, not mod-time & size\n");
942 + rprintf(F," --db=CONFIG_FILE specify a CONFIG_FILE for DB checksums\n");
943 rprintf(F," -a, --archive archive mode; equals -rlptgoD (no -H,-A,-X)\n");
944 rprintf(F," --no-OPTION turn off an implied OPTION (e.g. --no-D)\n");
945 rprintf(F," -r, --recursive recurse into directories\n");
946 @@ -579,6 +587,7 @@ static struct poptOption long_options[] = {
947 {"checksum", 'c', POPT_ARG_VAL, &always_checksum, 1, 0, 0 },
948 {"no-checksum", 0, POPT_ARG_VAL, &always_checksum, 0, 0, 0 },
949 {"no-c", 0, POPT_ARG_VAL, &always_checksum, 0, 0, 0 },
950 + {"db", 0, POPT_ARG_STRING, &db_config, 0, 0, 0 },
951 {"block-size", 'B', POPT_ARG_LONG, &block_size, 0, 0, 0 },
952 {"compare-dest", 0, POPT_ARG_STRING, 0, OPT_COMPARE_DEST, 0, 0 },
953 {"copy-dest", 0, POPT_ARG_STRING, 0, OPT_COPY_DEST, 0, 0 },
954 diff --git a/pipe.c b/pipe.c
957 @@ -26,6 +26,9 @@ extern int am_sender;
958 extern int am_server;
959 extern int blocking_io;
960 extern int filesfrom_fd;
961 +extern int always_checksum;
963 +extern char *db_config;
964 extern mode_t orig_umask;
965 extern char *logfile_name;
966 extern int remote_option_cnt;
967 @@ -141,6 +144,9 @@ pid_t local_child(int argc, char **argv, int *f_in, int *f_out,
974 if (remote_option_cnt) {
975 int rc = remote_option_cnt + 1;
976 const char **rv = remote_options;
977 @@ -148,6 +154,8 @@ pid_t local_child(int argc, char **argv, int *f_in, int *f_out,
979 exit_cleanup(RERR_SYNTAX);
981 + if (db_config && always_checksum)
982 + db_read_config(FERROR, db_config);
985 if (dup2(to_child_pipe[0], STDIN_FILENO) < 0 ||
986 diff --git a/rsync.yo b/rsync.yo
989 @@ -317,6 +317,7 @@ to the detailed description below for a complete description. verb(
990 -q, --quiet suppress non-error messages
991 --no-motd suppress daemon-mode MOTD (see caveat)
992 -c, --checksum skip based on checksum, not mod-time & size
993 + --db=CONFIG_FILE specify a CONFIG_FILE for DB checksums
994 -a, --archive archive mode; equals -rlptgoD (no -H,-A,-X)
995 --no-OPTION turn off an implied OPTION (e.g. --no-D)
996 -r, --recursive recurse into directories
997 @@ -533,6 +534,47 @@ checksum that is generated as the file is transferred, but that
998 automatic after-the-transfer verification has nothing to do with this
999 option's before-the-transfer "Does this file need to be updated?" check.
1001 +dit(bf(--db=CONFIG_FILE)) This option specifies a CONFIG_FILE to read
1002 +that holds connection details for a database of checksum information.
1003 +When combined with the bf(--checksum) (bf(-c)) option, rsync will try to
1004 +use cached checksum information from the DB, and will update it if it is
1007 +The currently supported DB choices are MySQL and SQLite. For example, a
1008 +MySQL configuration might look like this:
1010 +verb( dbtype: mysql
1016 + thishost: hostname )
1018 +And a SQLite configuration might look like this:
1020 +verb( dbtype: SQLite
1021 + dbname: /var/cache/rsync/sum.db )
1023 +This option only affects one side of a transfer. See the
1024 +bf(--remote-option) option for a way to specify the option for both
1025 +sides of the transfer (with each side reading the config file from
1026 +their local filesystem). For example:
1028 +verb( rsync -avc {-M,}--db=/etc/rsyncdb.conf src/ host:dest/ )
1030 +See the perl script "rsyncdb" in the support directory of the source code
1031 +(which may also be installed in /usr/bin) for a way to create the tables,
1032 +populate the mounted-disk information, check files against their checksums,
1033 +and update both the MD4 and MD5 checksums for files at the same time (since
1034 +an rsync copy will only update one or the other).
1036 +You can use a single MySQL DB for all your hosts if you give each one
1037 +their own "thishost" name and setup their device-mapping data. Or feel
1038 +free to use separate databases, separate servers, etc. See the rsync
1039 +daemon's "db config" parameter for how to configure a daemon to use a DB
1040 +(since a client cannot control this parameter on a daemon).
1042 dit(bf(-a, --archive)) This is equivalent to bf(-rlptgoD). It is a quick
1043 way of saying you want recursion and want to preserve almost
1044 everything (with -H being a notable omission).
1045 diff --git a/rsyncd.conf.yo b/rsyncd.conf.yo
1046 --- a/rsyncd.conf.yo
1047 +++ b/rsyncd.conf.yo
1048 @@ -270,6 +270,18 @@ is daemon. This setting has no effect if the "log file" setting is a
1049 non-empty string (either set in the per-modules settings, or inherited
1050 from the global settings).
1052 +dit(bf(db config)) This parameter specifies a config file to read that
1053 +holds connection details for a database of checksum information.
1055 +The config file will be read-in prior to any chroot restrictions, but
1056 +the connection occurs from inside the chroot. This means that you
1057 +should use a socket connection (e.g. 127.0.0.1 rather than localhost)
1058 +for a MySQL config from inside a chroot. For SQLite, the DB file must
1059 +be placed inside the chroot (though it can be placed outside the
1060 +transfer dir if you configured an inside-chroot path).
1062 +See the bf(--db=CONFIG_FILE) option for full details.
1064 dit(bf(max verbosity)) This parameter allows you to control
1065 the maximum amount of verbose information that you'll allow the daemon to
1066 generate (since the information goes into the log file). The default is 1,
1067 diff --git a/support/rsyncdb b/support/rsyncdb
1068 new file mode 100755
1070 +++ b/support/rsyncdb
1077 +use Cwd qw(abs_path cwd);
1081 +my $MOUNT_FILE = '/etc/mtab';
1083 +&Getopt::Long::Configure('bundling');
1084 +&usage if !&GetOptions(
1085 + 'db=s' => \( my $db_config ),
1086 + 'init' => \( my $init_db ),
1087 + 'mounts|m' => \( my $update_mounts ),
1088 + 'recurse|r' => \( my $recurse_opt ),
1089 + 'check|c' => \( my $check_opt ),
1090 + 'verbose|v+' => \( my $verbosity = 0 ),
1091 + 'help|h' => \( my $help_opt ),
1093 +&usage if $help_opt || !defined $db_config;
1096 +open(IN, '<', $db_config) or die "Unable to open $db_config: $!\n";
1100 + my($key, $val) = /^(\S+):\s*(.*)/ or die "Unable to parse line $. of $db_config\n";
1101 + $config{$key} = $val;
1105 +die "You must define at least dbtype and dbname in $db_config\n"
1106 + unless defined $config{'dbtype'} && defined $config{'dbname'};
1108 +my $sqlite = $config{'dbtype'} =~ /^sqlite$/i;
1110 +my $thishost = $config{'thishost'} || 'localhost';
1112 +my $connect = 'DBI:' . $config{'dbtype'} . ':';
1113 +$connect .= 'dbname=' . $config{'dbname'} if $sqlite;
1114 +$connect .= 'database=' . $config{'dbname'} if !$sqlite && !$init_db;
1115 +$connect .= ';host=' . $config{'dbhost'} if defined $config{'dbhost'};
1116 +$connect .= ';port=' . $config{'dbport'} if defined $config{'dbport'};
1118 +my $dbh = DBI->connect($connect, $config{'dbuser'}, $config{'dbpass'})
1119 + or die "DB connection failed\n";
1122 + $dbh->disconnect if defined $dbh;
1126 + my $unsigned = $sqlite ? '' : 'unsigned';
1127 + my $auto_increment = $sqlite ? 'AUTOINCREMENT' : 'AUTO_INCREMENT';
1128 + my $dbname = $config{'dbname'};
1131 + $dbh->do("CREATE DATABASE IF NOT EXISTS `$dbname`");
1132 + $dbh->do("USE `$dbname`");
1135 + print "Dropping old tables (if they exist) ...\n" if $verbosity;
1136 + $dbh->do("DROP TABLE IF EXISTS disk") or die $dbh->errstr;
1137 + $dbh->do("DROP TABLE IF EXISTS inode_map") or die $dbh->errstr;
1139 + print "Creating empty tables ...\n" if $verbosity;
1141 + CREATE TABLE disk (
1142 + disk_id integer $unsigned NOT NULL PRIMARY KEY $auto_increment,
1143 + devno bigint $unsigned NOT NULL,
1144 + host varchar(256) NOT NULL default 'localhost',
1145 + mounted tinyint NOT NULL default '1',
1146 + comment varchar(256) default NULL
1147 + )") or die $dbh->errstr;
1150 + CREATE TABLE inode_map (
1151 + disk_id integer $unsigned NOT NULL,
1152 + ino bigint $unsigned NOT NULL,
1153 + size bigint $unsigned NOT NULL,
1154 + mtime bigint NOT NULL,
1155 + ctime bigint NOT NULL,
1156 + sum_type tinyint NOT NULL default '0',
1157 + checksum binary(16) NOT NULL,
1158 + PRIMARY KEY (disk_id,ino,sum_type)
1159 + )") or die $dbh->errstr;
1161 + exit unless $update_mounts;
1164 +my $sel_disk_H = $dbh->prepare("
1165 + SELECT disk_id, devno, mounted, comment
1168 + ") or die $dbh->errstr;
1170 +my $ins_disk_H = $dbh->prepare("
1172 + (devno, host, mounted, comment)
1173 + VALUES(?, ?, ?, ?)
1174 + ") or die $dbh->errstr;
1176 +my $up_disk_H = $dbh->prepare("
1180 + ") or die $dbh->errstr;
1182 +my $row_id = $sqlite ? 'ROWID' : 'ID';
1183 +my $sel_lastid_H = $dbh->prepare("
1184 + SELECT LAST_INSERT_$row_id()
1185 + ") or die $dbh->errstr;
1187 +my $sel_sum_H = $dbh->prepare("
1188 + SELECT sum_type, checksum
1190 + WHERE disk_id = ? AND ino = ? AND size = ? AND mtime = ? AND ctime = ?
1191 + ") or die $dbh->errstr;
1193 +my $rep_sum_H = $dbh->prepare("
1194 + REPLACE INTO inode_map
1195 + (disk_id, ino, size, mtime, ctime, sum_type, checksum)
1196 + VALUES(?, ?, ?, ?, ?, ?, ?)
1197 + ") or die $dbh->errstr;
1200 +if ($update_mounts) {
1201 + open(IN, $MOUNT_FILE) or die "Unable to open $MOUNT_FILE: $!\n";
1203 + my($devname, $mnt) = (split)[0,1];
1204 + next unless $devname =~ m#^/dev#;
1205 + my($devno) = (stat($mnt))[0];
1206 + if (!defined $devno) {
1207 + warn "Unable to stat $mnt: $!\n";
1210 + $mounts{$devno} = "$devname on $mnt";
1216 +$sel_disk_H->execute($thishost);
1217 +while (my($disk_id, $devno, $mounted, $comment) = $sel_disk_H->fetchrow_array) {
1218 + if ($update_mounts) {
1219 + if (defined $mounts{$devno}) {
1220 + if ($comment ne $mounts{$devno}) {
1222 + print "Umounting $comment ($thishost:$devno)\n" if $verbosity;
1223 + $up_disk_H->execute(0, $disk_id);
1228 + print "Mounting $comment ($thishost:$devno)\n" if $verbosity;
1229 + $up_disk_H->execute(1, $disk_id);
1233 + print "Umounting $comment ($thishost:$devno)\n" if $verbosity;
1234 + $up_disk_H->execute(0, $disk_id);
1239 + next unless $mounted;
1241 + $disk_id{$devno} = $disk_id;
1243 +$sel_disk_H->finish;
1245 +if ($update_mounts) {
1246 + while (my($devno, $comment) = each %mounts) {
1247 + next if $disk_id{$devno};
1248 + print "Adding $comment ($thishost:$devno)\n" if $verbosity;
1249 + $ins_disk_H->execute($devno, $thishost, 1, $comment);
1250 + $sel_lastid_H->execute;
1251 + ($disk_id{$devno}) = $sel_lastid_H->fetchrow_array;
1252 + $sel_lastid_H->finish;
1257 +my $start_dir = cwd();
1260 +@dirs = '.' unless @dirs;
1262 + $_ = abs_path($_);
1269 +my $md4 = Digest::MD4->new;
1270 +my $md5 = Digest::MD5->new;
1273 + my $dir = shift @dirs;
1275 + if (!chdir($dir)) {
1276 + warn "Unable to chdir to $dir: $!\n";
1279 + if (!opendir(DP, '.')) {
1280 + warn "Unable to opendir $dir: $!\n";
1284 + my $reldir = $dir;
1285 + $reldir =~ s#^$start_dir(/|$)# $1 ? '' : '.' #eo;
1286 + print "$reldir ... \n" if $verbosity;
1289 + while (defined(my $fn = readdir(DP))) {
1290 + next if $fn =~ /^\.\.?$/ || -l $fn;
1292 + push(@subdirs, "$dir/$fn") unless $fn =~ /^(CVS|\.svn|\.git|\.bzr)$/;
1297 + my($dev,$ino,$size,$mtime,$ctime) = (stat(_))[0,1,7,9,10];
1298 + my $disk_id = $disk_id{$dev} or next;
1299 + $sel_sum_H->execute($disk_id,$ino,$size,$mtime,$ctime) or die $!;
1300 + my($sum4, $dbsum4, $sum5, $dbsum5);
1302 + while (my($sum_type, $checksum) = $sel_sum_H->fetchrow_array) {
1303 + if ($sum_type == 4) {
1304 + $dbsum4 = $checksum;
1306 + } elsif ($sum_type == 5) {
1307 + $dbsum5 = $checksum;
1311 + $sel_sum_H->finish;
1313 + next if !$check_opt && $dbsumcnt == 2;
1315 + if (!$check_opt || $dbsumcnt || $verbosity > 2) {
1316 + if (!open(IN, $fn)) {
1317 + print STDERR "Unable to read $fn: $!\n";
1322 + while (sysread(IN, $_, 64*1024)) {
1326 + $sum4 = $md4->digest;
1327 + $sum5 = $md5->digest;
1328 + print ' ', unpack('H*', $sum4), ' ', unpack('H*', $sum5) if $verbosity > 2;
1329 + print " $fn" if $verbosity > 1;
1330 + my($ino2,$size2,$mtime2,$ctime2) = (stat(IN))[1,7,9,10];
1331 + last if $ino == $ino2 && $size == $size2 && $mtime == $mtime2 && $ctime == $ctime2;
1336 + sysseek(IN, 0, 0);
1337 + print " REREADING\n" if $verbosity > 1;
1341 + } elsif ($verbosity > 1) {
1347 + if ($dbsumcnt == 0) {
1348 + $dif = ' --MISSING--';
1351 + if (!defined $dbsum4) {
1352 + $dif .= ' -NO-MD4-';
1353 + } elsif ($sum4 ne $dbsum4) {
1354 + $dif .= ' -MD4-CHANGED-';
1356 + if (!defined $dbsum5) {
1357 + $dif .= ' ---NO-MD5---';
1358 + } elsif ($sum5 ne $dbsum5) {
1359 + $dif .= ' -MD5-CHANGED-';
1362 + print " ====OK====\n" if $verbosity > 1;
1365 + $dif =~ s/MD4-CHANGED MD5-//;
1367 + if ($verbosity < 2) {
1368 + print $verbosity ? ' ' : "$reldir/";
1374 + print "\n" if $verbosity > 1;
1375 + $rep_sum_H->execute($disk_id, $ino, $size, $mtime, $ctime, 4, $sum4);
1376 + $rep_sum_H->execute($disk_id, $ino, $size, $mtime, $ctime, 5, $sum5);
1382 + unshift(@dirs, sort @subdirs) if $recurse_opt;
1390 +Usage: rsyncsums --db=CONFIG_FILE [OPTIONS] [DIRS]
1393 + --db=FILE Specify the config FILE to read for the DB info.
1394 + --init Create (recreate) needed tables (making them empty).
1395 + No DIR scanning, but can be combined with --mounts.
1396 + -m, --mounts Update mount info. Does no DIR scanning.
1397 + -r, --recurse Scan files in subdirectories too.
1398 + -c, --check Check if the checksums are right (doesn't update).
1399 + -v, --verbose Mention what we're doing. Repeat for more info.
1400 + -h, --help Display this help message.