December 2009 - developers - lists.mariadb.org

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2778)
by knielsen＠knielsen-hq.org 22 Dec '09

22 Dec '09

#At lp:maria 2778 knielsen(a)knielsen-hq.org 2009-12-22 [merge] Merge fix for PBXT running inside embedded server (MBug#439889). Also some small fixes to make the PBXT testsuite work in --embedded. renamed: mysql-test/suite/pbxt/t/load_unique_error1.inc => mysql-test/std_data/pbxt_load_unique_error1.inc modified: config/ac-macros/plugins.m4 mysql-test/suite/pbxt/r/pbxt_bugs.result mysql-test/suite/pbxt/t/pbxt_bugs.test mysql-test/suite/pbxt/t/pbxt_locking.test mysql-test/suite/pbxt/t/pbxt_transactions.test mysql-test/suite/pbxt/t/ps_1general.test sql/sql_plugin.cc storage/pbxt/plug.in storage/pbxt/src/Makefile.am storage/pbxt/src/ha_pbxt.cc storage/pbxt/src/table_xt.cc storage/pbxt/src/thread_xt.cc storage/pbxt/src/trace_xt.cc per-file messages: config/ac-macros/plugins.m4 MYSQL_PLUGIN_DEPENDS_ON_MYSQL_INTERNALS macro extended to support multiple files. mysql-test/std_data/pbxt_load_unique_error1.inc Move file to be accessible also for testing embedded server. mysql-test/suite/pbxt/r/pbxt_bugs.result Fix LOAD DATA LOCAL INFILE path so it works also for testing embedded server. mysql-test/suite/pbxt/t/pbxt_bugs.test Fix LOAD DATA LOCAL INFILE path so it works also for testing embedded server. mysql-test/suite/pbxt/t/pbxt_locking.test Disable for embedded, as it needs SHOW PROCESSLIST functionality not available there. mysql-test/suite/pbxt/t/pbxt_transactions.test Disable test for embedded, as it needs ability to connect from outside (mysqldump). mysql-test/suite/pbxt/t/ps_1general.test Fix replace_result for new mysql-test-run.pl. sql/sql_plugin.cc Remove hack that disables PBXT in embedded. storage/pbxt/plug.in Fix crash in PBXT in embedded server. storage/pbxt/src/Makefile.am Remove not needed CFLAGS/CXXFLAGS (they cause autotools to generate different object names, which in turn cause the MYSQL_PLUGIN_DEPENDS_ON_MYSQL_INTERNALS replacement of object files inside library files not to work). storage/pbxt/src/ha_pbxt.cc Ugly hack to allow more threads in embedded server (need a better fix I think). storage/pbxt/src/table_xt.cc Use stderr for logging not stdout (prevent spamming --embedded test suite output with stray messages). storage/pbxt/src/thread_xt.cc Use stderr for logging not stdout (prevent spamming --embedded test suite output with stray messages). storage/pbxt/src/trace_xt.cc Use stderr for logging not stdout (prevent spamming --embedded test suite output with stray messages). === modified file 'config/ac-macros/plugins.m4' --- a/config/ac-macros/plugins.m4 2009-04-25 10:05:32 +0000 +++ b/config/ac-macros/plugins.m4 2009-12-22 10:33:20 +0000 @@ -267,7 +267,6 @@ dnl we have to recompile these modules dnl to compile server parts with the different #defines dnl Normally it happens when we compile the embedded server dnl Thus one should mark such files in his handler using this macro -dnl (currently only one such a file per plugin is supported) dnl dnl --------------------------------------------------------------------------- @@ -463,11 +462,13 @@ dnl Although this is "pretty", it breaks mysql_plugin_defs="$mysql_plugin_defs, [builtin_]$2[_plugin]" [with_plugin_]$2=yes AC_MSG_RESULT([yes]) - m4_ifdef([$11],[ - condition_dependent_plugin_modules="$condition_dependent_plugin_modules m4_bregexp($11, [[^/]+$], [\&])" - condition_dependent_plugin_objects="$condition_dependent_plugin_objects m4_bregexp($11, [[^/]+\.], [\&o])" - condition_dependent_plugin_links="$condition_dependent_plugin_links $6/$11" - condition_dependent_plugin_includes="$condition_dependent_plugin_includes -I[\$(top_srcdir)]/$6/m4_bregexp($11, [^.+[/$]], [\&])" + m4_ifdef([$11], [ + m4_foreach([plugin], [$11], [ + condition_dependent_plugin_modules="$condition_dependent_plugin_modules m4_bregexp(plugin, [[^/]+$], [\&])" + condition_dependent_plugin_objects="$condition_dependent_plugin_objects m4_bregexp(plugin, [[^/]+\.], [\&o])" + condition_dependent_plugin_links="$condition_dependent_plugin_links $6/plugin" + condition_dependent_plugin_includes="$condition_dependent_plugin_includes -I[\$(top_srcdir)]/$6/m4_bregexp(plugin, [^.+[/$]], [\&])" + ]) ]) fi fi === renamed file 'mysql-test/suite/pbxt/t/load_unique_error1.inc' => 'mysql-test/std_data/pbxt_load_unique_error1.inc' === modified file 'mysql-test/suite/pbxt/r/pbxt_bugs.result' --- a/mysql-test/suite/pbxt/r/pbxt_bugs.result 2009-08-17 15:57:58 +0000 +++ b/mysql-test/suite/pbxt/r/pbxt_bugs.result 2009-12-22 10:33:20 +0000 @@ -1212,7 +1212,7 @@ c1 2147483647 DROP TABLE IF EXISTS t1; CREATE TABLE t1 (c1 INTEGER NOT NULL PRIMARY KEY, c2 VARCHAR(255)); -LOAD DATA LOCAL INFILE 'suite/pbxt/t/load_unique_error1.inc' REPLACE INTO TABLE t1 FIELDS TERMINATED BY ',' LINES TERMINATED BY '\n' (@c1,c2) SET c1 = @c1 % 2; +LOAD DATA LOCAL INFILE 'MYSQLTEST_VARDIR/std_data/pbxt_load_unique_error1.inc' REPLACE INTO TABLE t1 FIELDS TERMINATED BY ',' LINES TERMINATED BY '\n' (@c1,c2) SET c1 = @c1 % 2; SELECT * FROM t1 ORDER BY c1; c1 c2 0 opq === modified file 'mysql-test/suite/pbxt/t/pbxt_bugs.test' --- a/mysql-test/suite/pbxt/t/pbxt_bugs.test 2009-08-17 15:57:58 +0000 +++ b/mysql-test/suite/pbxt/t/pbxt_bugs.test 2009-12-22 10:33:20 +0000 @@ -921,7 +921,8 @@ SELECT c1 FROM t2; DROP TABLE IF EXISTS t1; --enable_warnings CREATE TABLE t1 (c1 INTEGER NOT NULL PRIMARY KEY, c2 VARCHAR(255)); -LOAD DATA LOCAL INFILE 'suite/pbxt/t/load_unique_error1.inc' REPLACE INTO TABLE t1 FIELDS TERMINATED BY ',' LINES TERMINATED BY '\n' (@c1,c2) SET c1 = @c1 % 2; +--replace_result $MYSQLTEST_VARDIR MYSQLTEST_VARDIR +eval LOAD DATA LOCAL INFILE '$MYSQLTEST_VARDIR/std_data/pbxt_load_unique_error1.inc' REPLACE INTO TABLE t1 FIELDS TERMINATED BY ',' LINES TERMINATED BY '\n' (@c1,c2) SET c1 = @c1 % 2; --sorted_result SELECT * FROM t1 ORDER BY c1; DROP TABLE t1; === modified file 'mysql-test/suite/pbxt/t/pbxt_locking.test' --- a/mysql-test/suite/pbxt/t/pbxt_locking.test 2009-04-02 10:03:14 +0000 +++ b/mysql-test/suite/pbxt/t/pbxt_locking.test 2009-12-22 10:33:20 +0000 @@ -1,6 +1,9 @@ # This test covers various aspects of PBXT locking mechanism, including # internal permanent/temporary row locking and MySQL locking +# SHOW PROCESSLIST has hardcoded "Writing to net" as state. +-- source include/not_embedded.inc + # TEST: select for update test drop table if exists t1; === modified file 'mysql-test/suite/pbxt/t/pbxt_transactions.test' --- a/mysql-test/suite/pbxt/t/pbxt_transactions.test 2009-04-02 10:03:14 +0000 +++ b/mysql-test/suite/pbxt/t/pbxt_transactions.test 2009-12-22 10:33:20 +0000 @@ -1,3 +1,6 @@ +# We cannot run mysqldump against embedded server. +-- source include/not_embedded.inc + --disable_warnings drop table if exists t1, t2, t3; --enable_warnings === modified file 'mysql-test/suite/pbxt/t/ps_1general.test' --- a/mysql-test/suite/pbxt/t/ps_1general.test 2009-04-02 10:03:14 +0000 +++ b/mysql-test/suite/pbxt/t/ps_1general.test 2009-12-22 10:33:20 +0000 @@ -582,7 +582,7 @@ prepare stmt1 from ' rename table t5 to create table t5 (a int) ; # rename must fail, t7 does not exist # Clean up the filename here because embedded server reports whole path ---replace_result $MYSQLTEST_VARDIR . master-data/ '' t7.frm t7 +--replace_result $MYSQLTEST_VARDIR . mysqld.1/data/ '' t7.frm t7 --error 1017 execute stmt1 ; create table t7 (a int) ; === modified file 'sql/sql_plugin.cc' --- a/sql/sql_plugin.cc 2009-12-03 11:19:05 +0000 +++ b/sql/sql_plugin.cc 2009-12-22 10:33:20 +0000 @@ -1168,22 +1168,7 @@ int plugin_init(int *argc, char **argv, !my_strnncoll(&my_charset_latin1, (const uchar*) plugin->name, 6, (const uchar*) "InnoDB", 6)) continue; -#ifdef EMBEDDED_LIBRARY - /* - MariaDB: disable PBXT in embedded server. We do this for two reasons - - PBXT currently doesn't work in embedded server (see - https://bugs.launchpad.net/maria/+bug/439889) - - Embedded server is supposed to be "leaner" and our current - understanding of that is "without PBXT". At the same time, we want - regular server to be with PBXT, and since we don't support compiling - embedded server with different options than the regular server, - the only way was to disable PBXT from here. - */ - if (!my_strnncoll(&my_charset_latin1, (const uchar*) plugin->name, - 4, (const uchar*) "PBXT", 4)) - continue; -#endif bzero(&tmp, sizeof(tmp)); tmp.plugin= plugin; tmp.name.str= (char *)plugin->name; === modified file 'storage/pbxt/plug.in' --- a/storage/pbxt/plug.in 2009-05-12 06:44:01 +0000 +++ b/storage/pbxt/plug.in 2009-12-09 21:39:23 +0000 @@ -5,3 +5,4 @@ MYSQL_PLUGIN_STATIC(pbxt, [src/libpbx MYSQL_PLUGIN_ACTIONS(pbxt, [ # AC_CONFIG_FILES(storage/pbxt/src/Makefile) ]) +MYSQL_PLUGIN_DEPENDS_ON_MYSQL_INTERNALS(pbxt, [[src/ha_pbxt.cc],[src/myxt_xt.cc],[src/discover_xt.cc]]) === modified file 'storage/pbxt/src/Makefile.am' --- a/storage/pbxt/src/Makefile.am 2009-11-24 10:55:06 +0000 +++ b/storage/pbxt/src/Makefile.am 2009-12-22 10:33:20 +0000 @@ -46,7 +46,5 @@ libpbxt_la_CFLAGS = $(AM_CFLAGS) -DMYSQ EXTRA_LIBRARIES = libpbxt.a noinst_LIBRARIES = libpbxt.a libpbxt_a_SOURCES = $(libpbxt_la_SOURCES) -libpbxt_a_CXXFLAGS = $(AM_CXXFLAGS) -libpbxt_a_CFLAGS = $(AM_CFLAGS) -std=c99 EXTRA_DIST = pbms_enabled.cc win_inttypes.h === modified file 'storage/pbxt/src/ha_pbxt.cc' --- a/storage/pbxt/src/ha_pbxt.cc 2009-11-27 15:37:02 +0000 +++ b/storage/pbxt/src/ha_pbxt.cc 2009-12-22 10:33:20 +0000 @@ -1175,9 +1175,13 @@ static int pbxt_init(void *p) * +1 Temporary thread (e.g. TempForClose, TempForEnd) */ #ifndef DRIZZLED +#ifdef EMBEDDED_LIBRARY + pbxt_max_threads = 100; +#else if (pbxt_max_threads == 0) pbxt_max_threads = max_connections + 7; #endif +#endif self = xt_init_threading(pbxt_max_threads); /* Create the main self: */ if (!self) goto error_3; === modified file 'storage/pbxt/src/table_xt.cc' --- a/storage/pbxt/src/table_xt.cc 2009-11-25 15:40:51 +0000 +++ b/storage/pbxt/src/table_xt.cc 2009-12-22 10:33:20 +0000 @@ -1297,7 +1297,7 @@ xtPublic void xt_create_table(XTThreadPt XTSortedListInfoRec li_undo; #ifdef TRACE_CREATE_TABLES - printf("CREATE %s\n", name->ps_path); + fprintf(stderr, "CREATE %s\n", name->ps_path); #endif enter_(); if (strlen(xt_last_name_of_path(name->ps_path)) > XT_TABLE_NAME_SIZE-1) @@ -1619,7 +1619,7 @@ xtPublic void xt_drop_table(XTThreadPtr enter_(); #ifdef TRACE_CREATE_TABLES - printf("DROP %s\n", tab_name->ps_path); + fprintf(stderr, "DROP %s\n", tab_name->ps_path); #endif table_pool = tab_lock_table(self, tab_name, FALSE, TRUE, TRUE, &tab); @@ -1777,7 +1777,7 @@ xtPublic void xt_check_table(XTThreadPtr u_llong ext_data_len = 0; #if defined(DUMP_CHECK_TABLE) || defined(CHECK_TABLE_STATS) - printf("\nCHECK TABLE: %s\n", tab->tab_name->ps_path); + fprintf(stderr, "\nCHECK TABLE: %s\n", tab->tab_name->ps_path); #endif xt_lock_mutex(self, &tab->tab_db->db_co_ext_lock); @@ -1787,38 +1787,38 @@ xtPublic void xt_check_table(XTThreadPtr pushr_(xt_unlock_mutex, &tab->tab_rec_lock); #ifdef CHECK_TABLE_STATS - printf("Record buffer size = %lu\n", (u_long) tab->tab_dic.dic_mysql_buf_size); - printf("Fixed length rec. len. = %lu\n", (u_long) tab->tab_dic.dic_mysql_rec_size); - printf("Handle data record size = %lu\n", (u_long) tab->tab_dic.dic_rec_size); - printf("Min/max header size = %d/%d\n", (int) offsetof(XTTabRecFix, rf_data), tab->tab_dic.dic_rec_fixed ? (int) offsetof(XTTabRecFix, rf_data) : (int) offsetof(XTTabRecExtDRec, re_data)); - printf("Min/avg/max record size = %llu/%llu/%llu\n", (u_llong) tab->tab_dic.dic_min_row_size, (u_llong) tab->tab_dic.dic_ave_row_size, (u_llong) tab->tab_dic.dic_max_row_size); + fprintf(stderr, "Record buffer size = %lu\n", (u_long) tab->tab_dic.dic_mysql_buf_size); + fprintf(stderr, "Fixed length rec. len. = %lu\n", (u_long) tab->tab_dic.dic_mysql_rec_size); + fprintf(stderr, "Handle data record size = %lu\n", (u_long) tab->tab_dic.dic_rec_size); + fprintf(stderr, "Min/max header size = %d/%d\n", (int) offsetof(XTTabRecFix, rf_data), tab->tab_dic.dic_rec_fixed ? (int) offsetof(XTTabRecFix, rf_data) : (int) offsetof(XTTabRecExtDRec, re_data)); + fprintf(stderr, "Min/avg/max record size = %llu/%llu/%llu\n", (u_llong) tab->tab_dic.dic_min_row_size, (u_llong) tab->tab_dic.dic_ave_row_size, (u_llong) tab->tab_dic.dic_max_row_size); if (tab->tab_dic.dic_def_ave_row_size) - printf("Avg row len set for tab = %lu\n", (u_long) tab->tab_dic.dic_def_ave_row_size); + fprintf(stderr, "Avg row len set for tab = %lu\n", (u_long) tab->tab_dic.dic_def_ave_row_size); else - printf("Avg row len set for tab = not specified\n"); - printf("Rows fixed length = %s\n", tab->tab_dic.dic_rec_fixed ? "YES" : "NO"); + fprintf(stderr, "Avg row len set for tab = not specified\n"); + fprintf(stderr, "Rows fixed length = %s\n", tab->tab_dic.dic_rec_fixed ? "YES" : "NO"); if (tab->tab_dic.dic_tab_flags & XT_TAB_FLAGS_TEMP_TAB) - printf("Table type = TEMP\n"); + fprintf(stderr, "Table type = TEMP\n"); if (tab->tab_dic.dic_def_ave_row_size) - printf("Maximum fixed size = %lu\n", (u_long) XT_TAB_MAX_FIX_REC_LENGTH_SPEC); + fprintf(stderr, "Maximum fixed size = %lu\n", (u_long) XT_TAB_MAX_FIX_REC_LENGTH_SPEC); else - printf("Maximum fixed size = %lu\n", (u_long) XT_TAB_MAX_FIX_REC_LENGTH); - printf("Minimum variable size = %lu\n", (u_long) XT_TAB_MIN_VAR_REC_LENGTH); - printf("Minimum auto-increment = %llu\n", (u_llong) tab->tab_dic.dic_min_auto_inc); - printf("Number of columns = %lu\n", (u_long) tab->tab_dic.dic_no_of_cols); - printf("Number of fixed columns = %lu\n", (u_long) tab->tab_dic.dic_fix_col_count); - printf("Columns req. for index = %lu\n", (u_long) tab->tab_dic.dic_ind_cols_req); + fprintf(stderr, "Maximum fixed size = %lu\n", (u_long) XT_TAB_MAX_FIX_REC_LENGTH); + fprintf(stderr, "Minimum variable size = %lu\n", (u_long) XT_TAB_MIN_VAR_REC_LENGTH); + fprintf(stderr, "Minimum auto-increment = %llu\n", (u_llong) tab->tab_dic.dic_min_auto_inc); + fprintf(stderr, "Number of columns = %lu\n", (u_long) tab->tab_dic.dic_no_of_cols); + fprintf(stderr, "Number of fixed columns = %lu\n", (u_long) tab->tab_dic.dic_fix_col_count); + fprintf(stderr, "Columns req. for index = %lu\n", (u_long) tab->tab_dic.dic_ind_cols_req); if (tab->tab_dic.dic_ind_rec_len) - printf("Rec len req. for index = %llu\n", (u_llong) tab->tab_dic.dic_ind_rec_len); - printf("Columns req. for blobs = %lu\n", (u_long) tab->tab_dic.dic_blob_cols_req); - printf("Number of blob columns = %lu\n", (u_long) tab->tab_dic.dic_blob_count); - printf("Number of indices = %lu\n", (u_long) tab->tab_dic.dic_key_count); + fprintf(stderr, "Rec len req. for index = %llu\n", (u_llong) tab->tab_dic.dic_ind_rec_len); + fprintf(stderr, "Columns req. for blobs = %lu\n", (u_long) tab->tab_dic.dic_blob_cols_req); + fprintf(stderr, "Number of blob columns = %lu\n", (u_long) tab->tab_dic.dic_blob_count); + fprintf(stderr, "Number of indices = %lu\n", (u_long) tab->tab_dic.dic_key_count); #endif #ifdef DUMP_CHECK_TABLE - printf("Records:-\n"); - printf("Free list: %llu (%llu)\n", (u_llong) tab->tab_rec_free_id, (u_llong) tab->tab_rec_fnum); - printf("EOF: %llu\n", (u_llong) tab->tab_rec_eof_id); + fprintf(stderr, "Records:-\n"); + fprintf(stderr, "Free list: %llu (%llu)\n", (u_llong) tab->tab_rec_free_id, (u_llong) tab->tab_rec_fnum); + fprintf(stderr, "EOF: %llu\n", (u_llong) tab->tab_rec_eof_id); #endif rec_size = XT_REC_EXT_HEADER_SIZE; @@ -1830,24 +1830,24 @@ xtPublic void xt_check_table(XTThreadPtr xt_throw(self); #ifdef DUMP_CHECK_TABLE - printf("%-4llu ", (u_llong) rec_id); + fprintf(stderr, "%-4llu ", (u_llong) rec_id); #endif switch (rec_buf->tr_rec_type_1 & XT_TAB_STATUS_MASK) { case XT_TAB_STATUS_FREED: #ifdef DUMP_CHECK_TABLE - printf("======== "); + fprintf(stderr, "======== "); #endif free_rec_count++; break; case XT_TAB_STATUS_DELETE: #ifdef DUMP_CHECK_TABLE - printf("delete "); + fprintf(stderr, "delete "); #endif delete_rec_count++; break; case XT_TAB_STATUS_FIXED: #ifdef DUMP_CHECK_TABLE - printf("record-F "); + fprintf(stderr, "record-F "); #endif alloc_rec_count++; row_size = myxt_store_row_length(ot, (char *) ot->ot_row_rbuffer + XT_REC_FIX_HEADER_SIZE); @@ -1859,7 +1859,7 @@ xtPublic void xt_check_table(XTThreadPtr break; case XT_TAB_STATUS_VARIABLE: #ifdef DUMP_CHECK_TABLE - printf("record-V "); + fprintf(stderr, "record-V "); #endif alloc_rec_count++; row_size = myxt_load_row_length(ot, tab->tab_dic.dic_rec_size, ot->ot_row_rbuffer + XT_REC_FIX_HEADER_SIZE, NULL); @@ -1871,7 +1871,7 @@ xtPublic void xt_check_table(XTThreadPtr break; case XT_TAB_STATUS_EXT_DLOG: #ifdef DUMP_CHECK_TABLE - printf("record-X "); + fprintf(stderr, "record-X "); #endif alloc_rec_count++; ext_data_len += XT_GET_DISK_4(rec_buf->re_log_dat_siz_4); @@ -1885,9 +1885,9 @@ xtPublic void xt_check_table(XTThreadPtr } #ifdef DUMP_CHECK_TABLE if (rec_buf->tr_rec_type_1 & XT_TAB_STATUS_CLEANED_BIT) - printf("C"); + fprintf(stderr, "C"); else - printf(" "); + fprintf(stderr, " "); #endif prev_rec_id = XT_GET_DISK_4(rec_buf->tr_prev_rec_id_4); xn_id = XT_GET_DISK_4(rec_buf->tr_xact_id_4); @@ -1895,12 +1895,12 @@ xtPublic void xt_check_table(XTThreadPtr switch (rec_buf->tr_rec_type_1 & XT_TAB_STATUS_MASK) { case XT_TAB_STATUS_FREED: #ifdef DUMP_CHECK_TABLE - printf(" prev=%-3llu (xact=%-3llu row=%lu)\n", (u_llong) prev_rec_id, (u_llong) xn_id, (u_long) row_id); + fprintf(stderr, " prev=%-3llu (xact=%-3llu row=%lu)\n", (u_llong) prev_rec_id, (u_llong) xn_id, (u_long) row_id); #endif break; case XT_TAB_STATUS_EXT_DLOG: #ifdef DUMP_CHECK_TABLE - printf(" prev=%-3llu xact=%-3llu row=%lu Xlog=%lu Xoff=%llu Xsiz=%lu\n", (u_llong) prev_rec_id, (u_llong) xn_id, (u_long) row_id, (u_long) XT_GET_DISK_2(rec_buf->re_log_id_2), (u_llong) XT_GET_DISK_6(rec_buf->re_log_offs_6), (u_long) XT_GET_DISK_4(rec_buf->re_log_dat_siz_4)); + fprintf(stderr, " prev=%-3llu xact=%-3llu row=%lu Xlog=%lu Xoff=%llu Xsiz=%lu\n", (u_llong) prev_rec_id, (u_llong) xn_id, (u_long) row_id, (u_long) XT_GET_DISK_2(rec_buf->re_log_id_2), (u_llong) XT_GET_DISK_6(rec_buf->re_log_offs_6), (u_long) XT_GET_DISK_4(rec_buf->re_log_dat_siz_4)); #endif log_size = XT_GET_DISK_4(rec_buf->re_log_dat_siz_4); @@ -1922,7 +1922,7 @@ xtPublic void xt_check_table(XTThreadPtr break; default: #ifdef DUMP_CHECK_TABLE - printf(" prev=%-3llu xact=%-3llu row=%lu\n", (u_llong) prev_rec_id, (u_llong) xn_id, (u_long) row_id); + fprintf(stderr, " prev=%-3llu xact=%-3llu row=%lu\n", (u_llong) prev_rec_id, (u_llong) xn_id, (u_long) row_id); #endif break; } @@ -1931,16 +1931,16 @@ xtPublic void xt_check_table(XTThreadPtr #ifdef CHECK_TABLE_STATS if (!tab->tab_dic.dic_rec_fixed) - printf("Extendend data length = %llu\n", ext_data_len); + fprintf(stderr, "Extendend data length = %llu\n", ext_data_len); if (alloc_rec_count) { - printf("Minumum comp. rec. len. = %llu\n", (u_llong) min_comp_rec_len); - printf("Average comp. rec. len. = %llu\n", (u_llong) ((double) alloc_rec_bytes / (double) alloc_rec_count + (double) 0.5)); - printf("Maximum comp. rec. len. = %llu\n", (u_llong) max_comp_rec_len); - } - printf("Free record count = %llu\n", (u_llong) free_rec_count); - printf("Deleted record count = %llu\n", (u_llong) delete_rec_count); - printf("Allocated record count = %llu\n", (u_llong) alloc_rec_count); + fprintf(stderr, "Minumum comp. rec. len. = %llu\n", (u_llong) min_comp_rec_len); + fprintf(stderr, "Average comp. rec. len. = %llu\n", (u_llong) ((double) alloc_rec_bytes / (double) alloc_rec_count + (double) 0.5)); + fprintf(stderr, "Maximum comp. rec. len. = %llu\n", (u_llong) max_comp_rec_len); + } + fprintf(stderr, "Free record count = %llu\n", (u_llong) free_rec_count); + fprintf(stderr, "Deleted record count = %llu\n", (u_llong) delete_rec_count); + fprintf(stderr, "Allocated record count = %llu\n", (u_llong) alloc_rec_count); #endif if (tab->tab_rec_fnum != free_rec_count) xt_logf(XT_INFO, "Table %s: incorrect number of free blocks, %llu, should be: %llu\n", tab->tab_name, (u_llong) free_rec_count, (u_llong) tab->tab_rec_fnum); @@ -1978,9 +1978,9 @@ xtPublic void xt_check_table(XTThreadPtr pushr_(xt_unlock_mutex, &tab->tab_row_lock); #ifdef DUMP_CHECK_TABLE - printf("Rows:-\n"); - printf("Free list: %llu (%llu)\n", (u_llong) tab->tab_row_free_id, (u_llong) tab->tab_row_fnum); - printf("EOF: %llu\n", (u_llong) tab->tab_row_eof_id); + fprintf(stderr, "Rows:-\n"); + fprintf(stderr, "Free list: %llu (%llu)\n", (u_llong) tab->tab_row_free_id, (u_llong) tab->tab_row_fnum); + fprintf(stderr, "EOF: %llu\n", (u_llong) tab->tab_row_eof_id); #endif rec_id = 1; @@ -1988,13 +1988,13 @@ xtPublic void xt_check_table(XTThreadPtr if (!tab->tab_rows.xt_tc_read_4(ot->ot_row_file, rec_id, &ref_id, self)) xt_throw(self); #ifdef DUMP_CHECK_TABLE - printf("%-3llu ", (u_llong) rec_id); + fprintf(stderr, "%-3llu ", (u_llong) rec_id); #endif #ifdef DUMP_CHECK_TABLE if (ref_id == 0) - printf("====== 0\n"); + fprintf(stderr, "====== 0\n"); else - printf("in use %llu\n", (u_llong) ref_id); + fprintf(stderr, "in use %llu\n", (u_llong) ref_id); #endif rec_id++; } @@ -2026,7 +2026,7 @@ xtPublic void xt_rename_table(XTThreadPt memset(&dic, 0, sizeof(dic)); #ifdef TRACE_CREATE_TABLES - printf("RENAME %s --> %s\n", old_name->ps_path, new_name->ps_path); + fprintf(stderr, "RENAME %s --> %s\n", old_name->ps_path, new_name->ps_path); #endif if (strlen(xt_last_name_of_path(new_name->ps_path)) > XT_TABLE_NAME_SIZE-1) xt_throw_taberr(XT_CONTEXT, XT_ERR_NAME_TOO_LONG, new_name); @@ -2221,7 +2221,7 @@ xtPublic xtBool xt_flush_record_row(XTOp xt_tab_store_header(ot, &rec_head); #ifdef TRACE_FLUSH - printf("FLUSH rec/row %d %s\n", (int) tab->tab_bytes_to_flush, tab->tab_name->ps_path); + fprintf(stderr, "FLUSH rec/row %d %s\n", (int) tab->tab_bytes_to_flush, tab->tab_name->ps_path); fflush(stdout); #endif /* Write the table header: */ @@ -2276,7 +2276,7 @@ xtPublic xtBool xt_flush_record_row(XTOp xt_unlock_mutex_ns(&cp->cp_state_lock); #ifdef TRACE_FLUSH - printf("FLUSH --end-- %s\n", tab->tab_name->ps_path); + fprintf(stderr, "FLUSH --end-- %s\n", tab->tab_name->ps_path); fflush(stdout); #endif xt_unlock_mutex_ns(&tab->tab_rec_flush_lock); === modified file 'storage/pbxt/src/thread_xt.cc' --- a/storage/pbxt/src/thread_xt.cc 2009-11-24 10:55:06 +0000 +++ b/storage/pbxt/src/thread_xt.cc 2009-12-22 10:33:20 +0000 @@ -96,7 +96,7 @@ xtPublic xtBool xt_init_logging(void) { int err; - log_file = stdout; + log_file = stderr; log_level = XT_LOG_TRACE; err = xt_p_mutex_init_with_autoname(&log_mutex, NULL); if (err) { === modified file 'storage/pbxt/src/trace_xt.cc' --- a/storage/pbxt/src/trace_xt.cc 2009-08-17 11:12:36 +0000 +++ b/storage/pbxt/src/trace_xt.cc 2009-12-22 10:33:20 +0000 @@ -109,10 +109,10 @@ xtPublic void xt_print_trace(void) xt_lock_mutex_ns(&trace_mutex); if (trace_log_end > trace_log_offset+1) { trace_log_buffer[trace_log_end] = 0; - printf("%s", trace_log_buffer + trace_log_offset + 1); + fprintf(stderr, "%s", trace_log_buffer + trace_log_offset + 1); } trace_log_buffer[trace_log_offset] = 0; - printf("%s", trace_log_buffer); + fprintf(stderr, "%s", trace_log_buffer); trace_log_offset = 0; trace_log_end = 0; xt_unlock_mutex_ns(&trace_mutex); @@ -379,9 +379,9 @@ xtPublic void xt_dump_conn_tracking(void ptr = conn_info; for (int i=0; i<XT_TRACK_MAX_CONNS; i++) { if (ptr->ci_curr_xact_id || ptr->ci_prev_xact_id) { - printf("%3d curr=%d prev=%d prev-time=%ld\n", (int) ptr->cu_t_id, (int) ptr->ci_curr_xact_id, (int) ptr->ci_prev_xact_id, (long) ptr->ci_prev_xact_time); + fprintf(stderr, "%3d curr=%d prev=%d prev-time=%ld\n", (int) ptr->cu_t_id, (int) ptr->ci_curr_xact_id, (int) ptr->ci_prev_xact_id, (long) ptr->ci_prev_xact_time); if (i+1<XT_TRACK_MAX_CONNS) { - printf(" diff=%d\n", (int) (ptr+1)->ci_curr_xact_id - (int) ptr->ci_curr_xact_id); + fprintf(stderr, " diff=%d\n", (int) (ptr+1)->ci_curr_xact_id - (int) ptr->ci_curr_xact_id); } } ptr++;

1 0

[Maria-developers] bzr commit into Mariadb 5.2, with Maria 2.0:maria/5.2 branch (igor:2740)
by Igor Babaev 22 Dec '09

22 Dec '09

#At lp:maria/5.2 based on revid:igor@askmonty.org-20091221022615-kx5ieiu0okmiupuc 2740 Igor Babaev 2009-12-21 [merge] Merge from 5.2-dsmrr modified: mysql-test/r/innodb_mrr.result mysql-test/r/type_date.result === modified file 'mysql-test/r/innodb_mrr.result' --- a/mysql-test/r/innodb_mrr.result 2009-12-15 07:16:46 +0000 +++ b/mysql-test/r/innodb_mrr.result 2009-12-19 19:54:54 +0000 @@ -318,7 +318,7 @@ filler char(10), key(d), primary key (a, insert into t2 select A.a, B.a, B.a, A.a, 'filler' from t1 A, t1 B; explain select * from t2 force index (d) where d < 10; id select_type table type possible_keys key key_len ref rows Extra -1 SIMPLE t2 range d d 5 NULL 47 Using index condition; Using MRR +1 SIMPLE t2 range d d 5 NULL 53 Using index condition; Using MRR drop table t2; drop table t1; set @@read_rnd_buffer_size= @read_rnd_buffer_size_save; === modified file 'mysql-test/r/type_date.result' --- a/mysql-test/r/type_date.result 2009-01-13 14:04:28 +0000 +++ b/mysql-test/r/type_date.result 2009-12-19 19:54:54 +0000 @@ -207,6 +207,7 @@ id select_type table type possible_keys Warnings: Warning 1292 Incorrect date value: '0000-00-00' for column 'a' at row 1 Warning 1292 Incorrect date value: '0000-00-00' for column 'a' at row 1 +Warning 1265 Data truncated for column 'a' at row 1 SELECT * FROM t1 WHERE a = '0000-00-00'; a 0000-00-00 @@ -214,6 +215,7 @@ a Warnings: Warning 1292 Incorrect date value: '0000-00-00' for column 'a' at row 1 Warning 1292 Incorrect date value: '0000-00-00' for column 'a' at row 1 +Warning 1265 Data truncated for column 'a' at row 1 SELECT * FROM t2 WHERE a = '0000-00-00'; a 0000-00-00 @@ -245,6 +247,7 @@ id select_type table type possible_keys Warnings: Warning 1292 Incorrect date value: '1000-00-00' for column 'a' at row 1 Warning 1292 Incorrect date value: '1000-00-00' for column 'a' at row 1 +Warning 1265 Data truncated for column 'a' at row 1 SELECT * FROM t1 WHERE a = '1000-00-00'; a 1000-00-00 @@ -252,6 +255,7 @@ a Warnings: Warning 1292 Incorrect date value: '1000-00-00' for column 'a' at row 1 Warning 1292 Incorrect date value: '1000-00-00' for column 'a' at row 1 +Warning 1265 Data truncated for column 'a' at row 1 SELECT * FROM t2 WHERE a = '1000-00-00'; a 1000-00-00

2 1

Re: [Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2785)
by Adam M. Dutko 21 Dec '09

21 Dec '09

Kristian, Sorry for making it seem like I wasn't adressing the group but I thought it was Arjen posing the question about build hosts. On a related note, I just got a netra 1u system I can put opensolaris on and configure. Do you have notes on configuring buildbot in addition to those found on your live journal site that you want me to follow? I'll start working on getting it running. -Adam On Dec 21, 2009 11:29 AM, <knielsen(a)knielsen-hq.org> wrote: #At lp:maria 2785 knielsen(a)knielsen-hq.org 2009-12-21 Don't include the `innodb' suite in the list of default test suites. This suite is for the innodb plugin, which we do not use in MariaDB (we instead have XtraDB as default storage engine, which includes the innodb plugin). We have included the same tests in the main suite, so no point in running them twice (and most of them are disabled anyway due to the absense of the innodb plugin). modified: mysql-test/mysql-test-run.pl === modified file 'mysql-test/mysql-test-run.pl' --- a/mysql-test/mysql-test-run.pl 2009-12-06 17:34:54 +0000 +++ b/mysql-test/mysql-test-run.pl 2009-12-21 16:26:36 +0000 @@ -126,7 +126,7 @@ my $path_config_file; # The ge # executables will be used by the test suite. our $opt_vs_config = $ENV{'MTR_VS_CONFIG'}; -my $DEFAULT_SUITES= "main,binlog,federated,rpl,innodb,maria,parts"; +my $DEFAULT_SUITES= "main,binlog,federated,rpl,maria,parts"; my $opt_suites; our $opt_verbose= 0; # Verbose output, enable with --verbose _______________________________________________ Mailing list: https://launchpad.net/~maria-develop...

1 0

[Maria-developers] [Branch ~maria-captains/maria/5.1] Rev 2785: Don't include the `innodb' suite in the list of default test suites.
by noreply＠launchpad.net 21 Dec '09

21 Dec '09

------------------------------------------------------------ revno: 2785 committer: knielsen(a)knielsen-hq.org branch nick: mariadb-5.1 timestamp: Mon 2009-12-21 17:26:36 +0100 message: Don't include the `innodb' suite in the list of default test suites. This suite is for the innodb plugin, which we do not use in MariaDB (we instead have XtraDB as default storage engine, which includes the innodb plugin). We have included the same tests in the main suite, so no point in running them twice (and most of them are disabled anyway due to the absense of the innodb plugin). modified: mysql-test/mysql-test-run.pl -- lp:maria https://code.launchpad.net/~maria-captains/maria/5.1 Your team Maria developers is subscribed to branch lp:maria. To unsubscribe from this branch go to https://code.launchpad.net/~maria-captains/maria/5.1/+edit-subscription.

1 0

[Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2785)
by knielsen＠knielsen-hq.org 21 Dec '09

21 Dec '09

#At lp:maria 2785 knielsen(a)knielsen-hq.org 2009-12-21 Don't include the `innodb' suite in the list of default test suites. This suite is for the innodb plugin, which we do not use in MariaDB (we instead have XtraDB as default storage engine, which includes the innodb plugin). We have included the same tests in the main suite, so no point in running them twice (and most of them are disabled anyway due to the absense of the innodb plugin). modified: mysql-test/mysql-test-run.pl === modified file 'mysql-test/mysql-test-run.pl' --- a/mysql-test/mysql-test-run.pl 2009-12-06 17:34:54 +0000 +++ b/mysql-test/mysql-test-run.pl 2009-12-21 16:26:36 +0000 @@ -126,7 +126,7 @@ my $path_config_file; # The ge # executables will be used by the test suite. our $opt_vs_config = $ENV{'MTR_VS_CONFIG'}; -my $DEFAULT_SUITES= "main,binlog,federated,rpl,innodb,maria,parts"; +my $DEFAULT_SUITES= "main,binlog,federated,rpl,maria,parts"; my $opt_suites; our $opt_verbose= 0; # Verbose output, enable with --verbose

1 0

Re: [Maria-developers] bzr commit into MariaDB 5.1, with Maria 1.5:maria branch (knielsen:2785)
by Adam M. Dutko 21 Dec '09

21 Dec '09

Arjen, Do you have a matrix of distros/platforms you'd like to see added to the build pool? I have some more compute power that I might be able to add to the build pool. -Adam On Dec 18, 2009 6:02 PM, "Michael Widenius" <monty(a)askmonty.org> wrote: Hi! >>>>> "Kristian" == Kristian Nielsen <knielsen(a)knielsen-hq.org> writes: Kristian> Arjen Lentz <arjen(a)openquery.com> writes: >> Which other external libraries are currently... Kristian> These are the ones I know of: Kristian> zlib Kristian> libedit Kristian> libreadline Kristian> yassl Kristian> libevent Kristian> I think most (all?) of them have a ./configure option to use the system one Kristian> instead of the bundled one. Kristian> I guess the main problem with boost is the sheer size. The largest of the Kristian> above is yassl, with 182 files and <2 MB total size. Of the above, the following should be easy to remove: zlib libedit libreadline yassl is patched (or will be patched) libevent is probably patched; If not, it shouldn't be here. Regards, Monty _______________________________________________ Mailing list: https://launchpad.net/~maria-develope...

2 1

[Maria-developers] Updated (by Guest): mysqlbinlog: remove undesired effects of format description binlog statement (50)
by worklog-noreply＠askmonty.org 21 Dec '09

21 Dec '09

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: mysqlbinlog: remove undesired effects of format description binlog statement CREATION DATE..: Mon, 17 Aug 2009, 14:12 SUPERVISOR.....: Monty IMPLEMENTOR....: Knielsen COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 50 (http://askmonty.org/worklog/?tid=50) VERSION........: Server-9.x STATUS.........: Complete PRIORITY.......: 60 WORKED HOURS...: 16 ESTIMATE.......: 4 (hours remain) ORIG. ESTIMATE.: 10 PROGRESS NOTES: -=-=(Guest - Mon, 21 Dec 2009, 15:40)=-=- Status updated. --- /tmp/wklog.50.old.26984 2009-12-21 13:40:53.000000000 +0000 +++ /tmp/wklog.50.new.26984 2009-12-21 13:40:53.000000000 +0000 @@ -1 +1 @@ -Code-Review +Complete -=-=(Knielsen - Mon, 05 Oct 2009, 14:38)=-=- Follow-up with MySQL developers on patch Worked 2 hours and estimate 4 hours remain (original estimate unchanged). -=-=(Knielsen - Tue, 29 Sep 2009, 10:12)=-=- Category updated. --- /tmp/wklog.50.old.3293 2009-09-29 10:12:26.000000000 +0300 +++ /tmp/wklog.50.new.3293 2009-09-29 10:12:26.000000000 +0300 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Knielsen - Tue, 29 Sep 2009, 10:12)=-=- Status updated. --- /tmp/wklog.50.old.3293 2009-09-29 10:12:26.000000000 +0300 +++ /tmp/wklog.50.new.3293 2009-09-29 10:12:26.000000000 +0300 @@ -1 +1 @@ -Un-Assigned +Code-Review -=-=(Knielsen - Mon, 28 Sep 2009, 16:00)=-=- Updated patch against MySQL 5.1.39 Fixes a couple of issues with the patch. Discussions with MySQL developer. Worked 4 hours and estimate 6 hours remain (original estimate unchanged). -=-=(Knielsen - Wed, 09 Sep 2009, 18:31)=-=- Discussed with MySQL developers on Bug#46640. Committed a suggested fix: https://lists.launchpad.net/maria-developers/msg00926.html Worked 8 hours and estimate 10 hours remain (original estimate increased by 18 hours). -=-=(Guest - Wed, 09 Sep 2009, 11:50)=-=- High Level Description modified. --- /tmp/wklog.50.old.11584 2009-09-09 11:50:12.000000000 +0300 +++ /tmp/wklog.50.new.11584 2009-09-09 11:50:12.000000000 +0300 @@ -5,7 +5,11 @@ This will cause an error when one tires to apply mysqlbinlog output to a 5.0.x server. -2. When one applies "format description binlog statement" at the slave, it -will change the slave's server_id when applied. +2. When one executes a BINLOG statement (containing "format description binlog +statement" or other events), this will cause subsequent statements run in +the same session to be binlogged with the server id of the last event +executed, not with the real server id. This can cause problems like infinite +recursive application in a circular replication topology. This WL is to fix these issues. + -=-=(Knielsen - Tue, 08 Sep 2009, 15:07)=-=- Low Level Design modified. --- /tmp/wklog.50.old.19208 2009-09-08 15:07:09.000000000 +0300 +++ /tmp/wklog.50.new.19208 2009-09-08 15:07:09.000000000 +0300 @@ -1 +1,34 @@ +I think the fix for point 2 is to replace the call to +apply_event_and_update_pos() in mysql_client_binlog_statement() with a direct +call to ev->apply_event(). However, I need to check two things to be sure this +is correct: + +1. The existing code does + + if (!ev->when) ev->when= my_time(0) + +Need to understand if this is needed/correct or not. + +2. The existing code does ev->update_pos(). I think this is redundant for +BINLOG statement (as it uses a fake rli structure), but I need to check to +make sure. + +Once this is done, point 1 may no longer be needed. The user can use +--base64-output=never when applying the output to a 5.0 server, and omit that +option when applying the output to a 5.1 server. There should be no need to +omit the format description event in the output when other BINLOG statements +for row events are present, as it no longer changes the server id. In fact the +format description event is required to be able to execute other BINLOG +statements, as its purpose is to define the binary format of the events +contained in these statements. + +Alternatively, we could implement that if the format description event of the +source binlog has server version < 5.1, and --base64-output=auto (default), +then the format description event is omitted (and should any BINLOG statement +be needed, unlikely as it is from a 5.0 server, we will need to throw an +error). + +The binlog format version for 5.0 and 5.1 is actually the same, hence the need +to look at server version to guess if the format description event can be +omitted. -=-=(Knielsen - Tue, 08 Sep 2009, 14:29)=-=- High-Level Specification modified. --- /tmp/wklog.50.old.17531 2009-09-08 14:29:30.000000000 +0300 +++ /tmp/wklog.50.new.17531 2009-09-08 14:29:30.000000000 +0300 @@ -12,3 +12,37 @@ One question that one needs to sort out before disabling server_id change is why it was put there in the first place? Should it be always removed? +First, the problem is not the format description log event. In fact all log +events have their own value of server_id, and every event in the BINLOG +statement is executed with its own id as server_id. + +It seems it was introduced deliberately in the patch for Bug#32407. This patch +makes sql thread and BINLOG statement share code path for executing the binary +event, even though the bug is really about something different (outputting +format description event to allow proper execution of other BINLOG +statements). + +Nevertheless, I think using the server_id in the events of a BINLOG statement +is wrong: + +1. It is different behaviour from executing normal statement-based events not +written using BINLOG. + +2. It causes the possibility for infinite cycle of replication in a cyclic +replication setup, if the server ids in BINLOG are different from all other +server ids in the cycle. + +3. The functionality of applying events with original server id from the event +is still available by pointing the slave thread to the binlog. The +functionality of applying row-based binlog events with server id of the server +executing the BINLOG statements is not otherwise available. + +In fact most of the code from the slave thread that is now also run when +executing BINLOG statements is redundant, or does the wrong thing (like +updating binlog position). + +Maybe there is a confusion on the purpose of the BINLOG statement. Is it to be +able to apply row-based events manually to a server (mysqlbinlog|mysql)? Or is +it to be able to simulate the effect of the slave sql thread? I think the +former is the most important one. + -=-=(Knielsen - Mon, 07 Sep 2009, 10:06)=-=- Research worklog, bugs, and existing server code. Worked 2 hours and estimate 0 hours remain (original estimate increased by 2 hours). ------------------------------------------------------------ -=-=(View All Progress Notes, 12 total)=-=- http://askmonty.org/worklog/index.pl?tid=50&nolimit=1 DESCRIPTION: According to complaints in BUG#46640: 1. mysqlbinlog will emit a 5.0-incompatible "format description binlog statement" even when reading a binary log that was produced by 5.0. This will cause an error when one tires to apply mysqlbinlog output to a 5.0.x server. 2. When one executes a BINLOG statement (containing "format description binlog statement" or other events), this will cause subsequent statements run in the same session to be binlogged with the server id of the last event executed, not with the real server id. This can cause problems like infinite recursive application in a circular replication topology. This WL is to fix these issues. HIGH-LEVEL SPECIFICATION: First item ---------- AFAIU what needs to be done is: 1. record a source server version (it is in the first binlog event). 2. don't emit a BINLOG statement if the recorded version number is 5.0.x or below. and we'll get a 5.0-applicable binlog. Second item ----------- One question that one needs to sort out before disabling server_id change is why it was put there in the first place? Should it be always removed? First, the problem is not the format description log event. In fact all log events have their own value of server_id, and every event in the BINLOG statement is executed with its own id as server_id. It seems it was introduced deliberately in the patch for Bug#32407. This patch makes sql thread and BINLOG statement share code path for executing the binary event, even though the bug is really about something different (outputting format description event to allow proper execution of other BINLOG statements). Nevertheless, I think using the server_id in the events of a BINLOG statement is wrong: 1. It is different behaviour from executing normal statement-based events not written using BINLOG. 2. It causes the possibility for infinite cycle of replication in a cyclic replication setup, if the server ids in BINLOG are different from all other server ids in the cycle. 3. The functionality of applying events with original server id from the event is still available by pointing the slave thread to the binlog. The functionality of applying row-based binlog events with server id of the server executing the BINLOG statements is not otherwise available. In fact most of the code from the slave thread that is now also run when executing BINLOG statements is redundant, or does the wrong thing (like updating binlog position). Maybe there is a confusion on the purpose of the BINLOG statement. Is it to be able to apply row-based events manually to a server (mysqlbinlog|mysql)? Or is it to be able to simulate the effect of the slave sql thread? I think the former is the most important one. LOW-LEVEL DESIGN: I think the fix for point 2 is to replace the call to apply_event_and_update_pos() in mysql_client_binlog_statement() with a direct call to ev->apply_event(). However, I need to check two things to be sure this is correct: 1. The existing code does if (!ev->when) ev->when= my_time(0) Need to understand if this is needed/correct or not. 2. The existing code does ev->update_pos(). I think this is redundant for BINLOG statement (as it uses a fake rli structure), but I need to check to make sure. Once this is done, point 1 may no longer be needed. The user can use --base64-output=never when applying the output to a 5.0 server, and omit that option when applying the output to a 5.1 server. There should be no need to omit the format description event in the output when other BINLOG statements for row events are present, as it no longer changes the server id. In fact the format description event is required to be able to execute other BINLOG statements, as its purpose is to define the binary format of the events contained in these statements. Alternatively, we could implement that if the format description event of the source binlog has server version < 5.1, and --base64-output=auto (default), then the format description event is omitted (and should any BINLOG statement be needed, unlikely as it is from a 5.0 server, we will need to throw an error). The binlog format version for 5.0 and 5.1 is actually the same, hence the need to look at server version to guess if the format description event can be omitted. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Guest): mysqlbinlog: remove undesired effects of format description binlog statement (50)
by worklog-noreply＠askmonty.org 21 Dec '09

21 Dec '09

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: mysqlbinlog: remove undesired effects of format description binlog statement CREATION DATE..: Mon, 17 Aug 2009, 14:12 SUPERVISOR.....: Monty IMPLEMENTOR....: Knielsen COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 50 (http://askmonty.org/worklog/?tid=50) VERSION........: Server-9.x STATUS.........: Complete PRIORITY.......: 60 WORKED HOURS...: 16 ESTIMATE.......: 4 (hours remain) ORIG. ESTIMATE.: 10 PROGRESS NOTES: -=-=(Guest - Mon, 21 Dec 2009, 15:40)=-=- Status updated. --- /tmp/wklog.50.old.26984 2009-12-21 13:40:53.000000000 +0000 +++ /tmp/wklog.50.new.26984 2009-12-21 13:40:53.000000000 +0000 @@ -1 +1 @@ -Code-Review +Complete -=-=(Knielsen - Mon, 05 Oct 2009, 14:38)=-=- Follow-up with MySQL developers on patch Worked 2 hours and estimate 4 hours remain (original estimate unchanged). -=-=(Knielsen - Tue, 29 Sep 2009, 10:12)=-=- Category updated. --- /tmp/wklog.50.old.3293 2009-09-29 10:12:26.000000000 +0300 +++ /tmp/wklog.50.new.3293 2009-09-29 10:12:26.000000000 +0300 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Knielsen - Tue, 29 Sep 2009, 10:12)=-=- Status updated. --- /tmp/wklog.50.old.3293 2009-09-29 10:12:26.000000000 +0300 +++ /tmp/wklog.50.new.3293 2009-09-29 10:12:26.000000000 +0300 @@ -1 +1 @@ -Un-Assigned +Code-Review -=-=(Knielsen - Mon, 28 Sep 2009, 16:00)=-=- Updated patch against MySQL 5.1.39 Fixes a couple of issues with the patch. Discussions with MySQL developer. Worked 4 hours and estimate 6 hours remain (original estimate unchanged). -=-=(Knielsen - Wed, 09 Sep 2009, 18:31)=-=- Discussed with MySQL developers on Bug#46640. Committed a suggested fix: https://lists.launchpad.net/maria-developers/msg00926.html Worked 8 hours and estimate 10 hours remain (original estimate increased by 18 hours). -=-=(Guest - Wed, 09 Sep 2009, 11:50)=-=- High Level Description modified. --- /tmp/wklog.50.old.11584 2009-09-09 11:50:12.000000000 +0300 +++ /tmp/wklog.50.new.11584 2009-09-09 11:50:12.000000000 +0300 @@ -5,7 +5,11 @@ This will cause an error when one tires to apply mysqlbinlog output to a 5.0.x server. -2. When one applies "format description binlog statement" at the slave, it -will change the slave's server_id when applied. +2. When one executes a BINLOG statement (containing "format description binlog +statement" or other events), this will cause subsequent statements run in +the same session to be binlogged with the server id of the last event +executed, not with the real server id. This can cause problems like infinite +recursive application in a circular replication topology. This WL is to fix these issues. + -=-=(Knielsen - Tue, 08 Sep 2009, 15:07)=-=- Low Level Design modified. --- /tmp/wklog.50.old.19208 2009-09-08 15:07:09.000000000 +0300 +++ /tmp/wklog.50.new.19208 2009-09-08 15:07:09.000000000 +0300 @@ -1 +1,34 @@ +I think the fix for point 2 is to replace the call to +apply_event_and_update_pos() in mysql_client_binlog_statement() with a direct +call to ev->apply_event(). However, I need to check two things to be sure this +is correct: + +1. The existing code does + + if (!ev->when) ev->when= my_time(0) + +Need to understand if this is needed/correct or not. + +2. The existing code does ev->update_pos(). I think this is redundant for +BINLOG statement (as it uses a fake rli structure), but I need to check to +make sure. + +Once this is done, point 1 may no longer be needed. The user can use +--base64-output=never when applying the output to a 5.0 server, and omit that +option when applying the output to a 5.1 server. There should be no need to +omit the format description event in the output when other BINLOG statements +for row events are present, as it no longer changes the server id. In fact the +format description event is required to be able to execute other BINLOG +statements, as its purpose is to define the binary format of the events +contained in these statements. + +Alternatively, we could implement that if the format description event of the +source binlog has server version < 5.1, and --base64-output=auto (default), +then the format description event is omitted (and should any BINLOG statement +be needed, unlikely as it is from a 5.0 server, we will need to throw an +error). + +The binlog format version for 5.0 and 5.1 is actually the same, hence the need +to look at server version to guess if the format description event can be +omitted. -=-=(Knielsen - Tue, 08 Sep 2009, 14:29)=-=- High-Level Specification modified. --- /tmp/wklog.50.old.17531 2009-09-08 14:29:30.000000000 +0300 +++ /tmp/wklog.50.new.17531 2009-09-08 14:29:30.000000000 +0300 @@ -12,3 +12,37 @@ One question that one needs to sort out before disabling server_id change is why it was put there in the first place? Should it be always removed? +First, the problem is not the format description log event. In fact all log +events have their own value of server_id, and every event in the BINLOG +statement is executed with its own id as server_id. + +It seems it was introduced deliberately in the patch for Bug#32407. This patch +makes sql thread and BINLOG statement share code path for executing the binary +event, even though the bug is really about something different (outputting +format description event to allow proper execution of other BINLOG +statements). + +Nevertheless, I think using the server_id in the events of a BINLOG statement +is wrong: + +1. It is different behaviour from executing normal statement-based events not +written using BINLOG. + +2. It causes the possibility for infinite cycle of replication in a cyclic +replication setup, if the server ids in BINLOG are different from all other +server ids in the cycle. + +3. The functionality of applying events with original server id from the event +is still available by pointing the slave thread to the binlog. The +functionality of applying row-based binlog events with server id of the server +executing the BINLOG statements is not otherwise available. + +In fact most of the code from the slave thread that is now also run when +executing BINLOG statements is redundant, or does the wrong thing (like +updating binlog position). + +Maybe there is a confusion on the purpose of the BINLOG statement. Is it to be +able to apply row-based events manually to a server (mysqlbinlog|mysql)? Or is +it to be able to simulate the effect of the slave sql thread? I think the +former is the most important one. + -=-=(Knielsen - Mon, 07 Sep 2009, 10:06)=-=- Research worklog, bugs, and existing server code. Worked 2 hours and estimate 0 hours remain (original estimate increased by 2 hours). ------------------------------------------------------------ -=-=(View All Progress Notes, 12 total)=-=- http://askmonty.org/worklog/index.pl?tid=50&nolimit=1 DESCRIPTION: According to complaints in BUG#46640: 1. mysqlbinlog will emit a 5.0-incompatible "format description binlog statement" even when reading a binary log that was produced by 5.0. This will cause an error when one tires to apply mysqlbinlog output to a 5.0.x server. 2. When one executes a BINLOG statement (containing "format description binlog statement" or other events), this will cause subsequent statements run in the same session to be binlogged with the server id of the last event executed, not with the real server id. This can cause problems like infinite recursive application in a circular replication topology. This WL is to fix these issues. HIGH-LEVEL SPECIFICATION: First item ---------- AFAIU what needs to be done is: 1. record a source server version (it is in the first binlog event). 2. don't emit a BINLOG statement if the recorded version number is 5.0.x or below. and we'll get a 5.0-applicable binlog. Second item ----------- One question that one needs to sort out before disabling server_id change is why it was put there in the first place? Should it be always removed? First, the problem is not the format description log event. In fact all log events have their own value of server_id, and every event in the BINLOG statement is executed with its own id as server_id. It seems it was introduced deliberately in the patch for Bug#32407. This patch makes sql thread and BINLOG statement share code path for executing the binary event, even though the bug is really about something different (outputting format description event to allow proper execution of other BINLOG statements). Nevertheless, I think using the server_id in the events of a BINLOG statement is wrong: 1. It is different behaviour from executing normal statement-based events not written using BINLOG. 2. It causes the possibility for infinite cycle of replication in a cyclic replication setup, if the server ids in BINLOG are different from all other server ids in the cycle. 3. The functionality of applying events with original server id from the event is still available by pointing the slave thread to the binlog. The functionality of applying row-based binlog events with server id of the server executing the BINLOG statements is not otherwise available. In fact most of the code from the slave thread that is now also run when executing BINLOG statements is redundant, or does the wrong thing (like updating binlog position). Maybe there is a confusion on the purpose of the BINLOG statement. Is it to be able to apply row-based events manually to a server (mysqlbinlog|mysql)? Or is it to be able to simulate the effect of the slave sql thread? I think the former is the most important one. LOW-LEVEL DESIGN: I think the fix for point 2 is to replace the call to apply_event_and_update_pos() in mysql_client_binlog_statement() with a direct call to ev->apply_event(). However, I need to check two things to be sure this is correct: 1. The existing code does if (!ev->when) ev->when= my_time(0) Need to understand if this is needed/correct or not. 2. The existing code does ev->update_pos(). I think this is redundant for BINLOG statement (as it uses a fake rli structure), but I need to check to make sure. Once this is done, point 1 may no longer be needed. The user can use --base64-output=never when applying the output to a 5.0 server, and omit that option when applying the output to a 5.1 server. There should be no need to omit the format description event in the output when other BINLOG statements for row events are present, as it no longer changes the server id. In fact the format description event is required to be able to execute other BINLOG statements, as its purpose is to define the binary format of the events contained in these statements. Alternatively, we could implement that if the format description event of the source binlog has server version < 5.1, and --base64-output=auto (default), then the format description event is omitted (and should any BINLOG statement be needed, unlikely as it is from a 5.0 server, we will need to throw an error). The binlog format version for 5.0 and 5.1 is actually the same, hence the need to look at server version to guess if the format description event can be omitted. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Guest): mysqlbinlog: remove undesired effects of format description binlog statement (50)
by worklog-noreply＠askmonty.org 21 Dec '09

21 Dec '09

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: mysqlbinlog: remove undesired effects of format description binlog statement CREATION DATE..: Mon, 17 Aug 2009, 14:12 SUPERVISOR.....: Monty IMPLEMENTOR....: Knielsen COPIES TO......: CATEGORY.......: Server-Sprint TASK ID........: 50 (http://askmonty.org/worklog/?tid=50) VERSION........: Server-9.x STATUS.........: Complete PRIORITY.......: 60 WORKED HOURS...: 16 ESTIMATE.......: 4 (hours remain) ORIG. ESTIMATE.: 10 PROGRESS NOTES: -=-=(Guest - Mon, 21 Dec 2009, 15:40)=-=- Status updated. --- /tmp/wklog.50.old.26984 2009-12-21 13:40:53.000000000 +0000 +++ /tmp/wklog.50.new.26984 2009-12-21 13:40:53.000000000 +0000 @@ -1 +1 @@ -Code-Review +Complete -=-=(Knielsen - Mon, 05 Oct 2009, 14:38)=-=- Follow-up with MySQL developers on patch Worked 2 hours and estimate 4 hours remain (original estimate unchanged). -=-=(Knielsen - Tue, 29 Sep 2009, 10:12)=-=- Category updated. --- /tmp/wklog.50.old.3293 2009-09-29 10:12:26.000000000 +0300 +++ /tmp/wklog.50.new.3293 2009-09-29 10:12:26.000000000 +0300 @@ -1 +1 @@ -Server-RawIdeaBin +Server-Sprint -=-=(Knielsen - Tue, 29 Sep 2009, 10:12)=-=- Status updated. --- /tmp/wklog.50.old.3293 2009-09-29 10:12:26.000000000 +0300 +++ /tmp/wklog.50.new.3293 2009-09-29 10:12:26.000000000 +0300 @@ -1 +1 @@ -Un-Assigned +Code-Review -=-=(Knielsen - Mon, 28 Sep 2009, 16:00)=-=- Updated patch against MySQL 5.1.39 Fixes a couple of issues with the patch. Discussions with MySQL developer. Worked 4 hours and estimate 6 hours remain (original estimate unchanged). -=-=(Knielsen - Wed, 09 Sep 2009, 18:31)=-=- Discussed with MySQL developers on Bug#46640. Committed a suggested fix: https://lists.launchpad.net/maria-developers/msg00926.html Worked 8 hours and estimate 10 hours remain (original estimate increased by 18 hours). -=-=(Guest - Wed, 09 Sep 2009, 11:50)=-=- High Level Description modified. --- /tmp/wklog.50.old.11584 2009-09-09 11:50:12.000000000 +0300 +++ /tmp/wklog.50.new.11584 2009-09-09 11:50:12.000000000 +0300 @@ -5,7 +5,11 @@ This will cause an error when one tires to apply mysqlbinlog output to a 5.0.x server. -2. When one applies "format description binlog statement" at the slave, it -will change the slave's server_id when applied. +2. When one executes a BINLOG statement (containing "format description binlog +statement" or other events), this will cause subsequent statements run in +the same session to be binlogged with the server id of the last event +executed, not with the real server id. This can cause problems like infinite +recursive application in a circular replication topology. This WL is to fix these issues. + -=-=(Knielsen - Tue, 08 Sep 2009, 15:07)=-=- Low Level Design modified. --- /tmp/wklog.50.old.19208 2009-09-08 15:07:09.000000000 +0300 +++ /tmp/wklog.50.new.19208 2009-09-08 15:07:09.000000000 +0300 @@ -1 +1,34 @@ +I think the fix for point 2 is to replace the call to +apply_event_and_update_pos() in mysql_client_binlog_statement() with a direct +call to ev->apply_event(). However, I need to check two things to be sure this +is correct: + +1. The existing code does + + if (!ev->when) ev->when= my_time(0) + +Need to understand if this is needed/correct or not. + +2. The existing code does ev->update_pos(). I think this is redundant for +BINLOG statement (as it uses a fake rli structure), but I need to check to +make sure. + +Once this is done, point 1 may no longer be needed. The user can use +--base64-output=never when applying the output to a 5.0 server, and omit that +option when applying the output to a 5.1 server. There should be no need to +omit the format description event in the output when other BINLOG statements +for row events are present, as it no longer changes the server id. In fact the +format description event is required to be able to execute other BINLOG +statements, as its purpose is to define the binary format of the events +contained in these statements. + +Alternatively, we could implement that if the format description event of the +source binlog has server version < 5.1, and --base64-output=auto (default), +then the format description event is omitted (and should any BINLOG statement +be needed, unlikely as it is from a 5.0 server, we will need to throw an +error). + +The binlog format version for 5.0 and 5.1 is actually the same, hence the need +to look at server version to guess if the format description event can be +omitted. -=-=(Knielsen - Tue, 08 Sep 2009, 14:29)=-=- High-Level Specification modified. --- /tmp/wklog.50.old.17531 2009-09-08 14:29:30.000000000 +0300 +++ /tmp/wklog.50.new.17531 2009-09-08 14:29:30.000000000 +0300 @@ -12,3 +12,37 @@ One question that one needs to sort out before disabling server_id change is why it was put there in the first place? Should it be always removed? +First, the problem is not the format description log event. In fact all log +events have their own value of server_id, and every event in the BINLOG +statement is executed with its own id as server_id. + +It seems it was introduced deliberately in the patch for Bug#32407. This patch +makes sql thread and BINLOG statement share code path for executing the binary +event, even though the bug is really about something different (outputting +format description event to allow proper execution of other BINLOG +statements). + +Nevertheless, I think using the server_id in the events of a BINLOG statement +is wrong: + +1. It is different behaviour from executing normal statement-based events not +written using BINLOG. + +2. It causes the possibility for infinite cycle of replication in a cyclic +replication setup, if the server ids in BINLOG are different from all other +server ids in the cycle. + +3. The functionality of applying events with original server id from the event +is still available by pointing the slave thread to the binlog. The +functionality of applying row-based binlog events with server id of the server +executing the BINLOG statements is not otherwise available. + +In fact most of the code from the slave thread that is now also run when +executing BINLOG statements is redundant, or does the wrong thing (like +updating binlog position). + +Maybe there is a confusion on the purpose of the BINLOG statement. Is it to be +able to apply row-based events manually to a server (mysqlbinlog|mysql)? Or is +it to be able to simulate the effect of the slave sql thread? I think the +former is the most important one. + -=-=(Knielsen - Mon, 07 Sep 2009, 10:06)=-=- Research worklog, bugs, and existing server code. Worked 2 hours and estimate 0 hours remain (original estimate increased by 2 hours). ------------------------------------------------------------ -=-=(View All Progress Notes, 12 total)=-=- http://askmonty.org/worklog/index.pl?tid=50&nolimit=1 DESCRIPTION: According to complaints in BUG#46640: 1. mysqlbinlog will emit a 5.0-incompatible "format description binlog statement" even when reading a binary log that was produced by 5.0. This will cause an error when one tires to apply mysqlbinlog output to a 5.0.x server. 2. When one executes a BINLOG statement (containing "format description binlog statement" or other events), this will cause subsequent statements run in the same session to be binlogged with the server id of the last event executed, not with the real server id. This can cause problems like infinite recursive application in a circular replication topology. This WL is to fix these issues. HIGH-LEVEL SPECIFICATION: First item ---------- AFAIU what needs to be done is: 1. record a source server version (it is in the first binlog event). 2. don't emit a BINLOG statement if the recorded version number is 5.0.x or below. and we'll get a 5.0-applicable binlog. Second item ----------- One question that one needs to sort out before disabling server_id change is why it was put there in the first place? Should it be always removed? First, the problem is not the format description log event. In fact all log events have their own value of server_id, and every event in the BINLOG statement is executed with its own id as server_id. It seems it was introduced deliberately in the patch for Bug#32407. This patch makes sql thread and BINLOG statement share code path for executing the binary event, even though the bug is really about something different (outputting format description event to allow proper execution of other BINLOG statements). Nevertheless, I think using the server_id in the events of a BINLOG statement is wrong: 1. It is different behaviour from executing normal statement-based events not written using BINLOG. 2. It causes the possibility for infinite cycle of replication in a cyclic replication setup, if the server ids in BINLOG are different from all other server ids in the cycle. 3. The functionality of applying events with original server id from the event is still available by pointing the slave thread to the binlog. The functionality of applying row-based binlog events with server id of the server executing the BINLOG statements is not otherwise available. In fact most of the code from the slave thread that is now also run when executing BINLOG statements is redundant, or does the wrong thing (like updating binlog position). Maybe there is a confusion on the purpose of the BINLOG statement. Is it to be able to apply row-based events manually to a server (mysqlbinlog|mysql)? Or is it to be able to simulate the effect of the slave sql thread? I think the former is the most important one. LOW-LEVEL DESIGN: I think the fix for point 2 is to replace the call to apply_event_and_update_pos() in mysql_client_binlog_statement() with a direct call to ev->apply_event(). However, I need to check two things to be sure this is correct: 1. The existing code does if (!ev->when) ev->when= my_time(0) Need to understand if this is needed/correct or not. 2. The existing code does ev->update_pos(). I think this is redundant for BINLOG statement (as it uses a fake rli structure), but I need to check to make sure. Once this is done, point 1 may no longer be needed. The user can use --base64-output=never when applying the output to a 5.0 server, and omit that option when applying the output to a 5.1 server. There should be no need to omit the format description event in the output when other BINLOG statements for row events are present, as it no longer changes the server id. In fact the format description event is required to be able to execute other BINLOG statements, as its purpose is to define the binary format of the events contained in these statements. Alternatively, we could implement that if the format description event of the source binlog has server version < 5.1, and --base64-output=auto (default), then the format description event is omitted (and should any BINLOG statement be needed, unlikely as it is from a 5.0 server, we will need to throw an error). The binlog format version for 5.0 and 5.1 is actually the same, hence the need to look at server version to guess if the format description event can be omitted. ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

1 0

[Maria-developers] Updated (by Alexi): Store in binlog text of statements that caused RBR events (47)
by worklog-noreply＠askmonty.org 21 Dec '09

21 Dec '09

----------------------------------------------------------------------- WORKLOG TASK -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- TASK...........: Store in binlog text of statements that caused RBR events CREATION DATE..: Sat, 15 Aug 2009, 23:48 SUPERVISOR.....: Monty IMPLEMENTOR....: COPIES TO......: Knielsen CATEGORY.......: Server-RawIdeaBin TASK ID........: 47 (http://askmonty.org/worklog/?tid=47) VERSION........: Server-9.x STATUS.........: Un-Assigned PRIORITY.......: 60 WORKED HOURS...: 20 ESTIMATE.......: 35 (hours remain) ORIG. ESTIMATE.: 35 PROGRESS NOTES: -=-=(Alexi - Sat, 19 Dec 2009, 15:58)=-=- Low Level Design modified. --- /tmp/wklog.47.old.15437 2009-12-19 13:58:12.000000000 +0000 +++ /tmp/wklog.47.new.15437 2009-12-19 13:58:12.000000000 +0000 @@ -1 +1,337 @@ +Content +~~~~~~~ + 1. Annotate_rows event number + 2. Outline of Annotate_rows event behavior + 3. How Master writes Annotate_rows events to the binary log + 4. How slave treats replicate-annotate-rows-events option + 5. How slave IO thread requests Annotate_rows events + 6. How master executes the request + 7. How slave SQL thread processes Annotate_rows events + 8. General remarks + +1. Annotate_rows event number +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +To avoid possible event numbers conflict with MySQL/Sun, we leave a gap +between the last MySQL event number and the Annotate_rows event number: + + enum Log_event_type + { ... + INCIDENT_EVENT= 26, + // New MySQL event numbers are to be added here + MYSQL_EVENTS_END, + + MARIA_EVENTS_BEGIN= 51, + // New Maria event numbers start from here + ANNOTATE_ROWS_EVENT= 51, + + ENUM_END_EVENT + }; + +together with the corresponding extension of 'post_header_len' array in the +Format description event. (This extension does not affect the compatibility +of the binary log). Here is how Format description event looks like with +this extension: + + ************************ + FORMAT_DESCRIPTION_EVENT + ************************ + 00000004 | A1 A0 2C 4B | time_when = 1261215905 + 00000008 | 0F | event_type = 15 + 00000009 | 64 00 00 00 | server_id = 100 + 0000000D | 7F 00 00 00 | event_len = 127 + 00000011 | 83 00 00 00 | log_pos = 00000083 + 00000015 | 01 00 | flags = LOG_EVENT_BINLOG_IN_USE_F + ------------------------ + 00000017 | 04 00 | binlog_ver = 4 + 00000019 | 35 2E 32 2E | server_ver = 5.2.0-MariaDB-alpha-debug-log + ..... ... + 0000004B | A1 A0 2C 4B | time_created = 1261215905 + 0000004F | 13 | common_header_len = 19 + ------------------------ + post_header_len + ------------------------ + 00000050 | 38 | 56 - START_EVENT_V3 [1] + ..... ... + 00000069 | 02 | 2 - INCIDENT_EVENT [26] + 0000006A | 00 | 0 - RESERVED [27] + ..... ... + 00000081 | 00 | 0 - RESERVED [50] + 00000082 | 00 | 0 - ANNOTATE_RBR_EVENT [51] + ************************ + +2. Outline of Annotate_rows event behavior +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +Each Annotate_rows_log_event object has two private members describing the +corresponding query: + + char *m_query_txt; + uint m_query_len; + +When the object is created for writing to a binary log, this query is taken +from 'thd' (for short, below we omit the 'Annotate_rows_log_event::' prefix +as well as other implementation details): + + Annotate_rows_log_event(THD *thd) + { + m_query_txt = thd->query(); + m_query_len = thd->query_length(); + } + +When the object is read from a binary log, the query is taken from the buffer +containing the binary log representation of the event (this buffer is allocated +in Log_event object from which all Log events are derived): + + Annotate_rows_log_event(char *buf, uint event_len, + Format_description_log_event *desc) + { + m_query_len = event_len - desc->common_header_len; + m_query_txt = buf + desc->common_header_len; + } + +The events are written to the binary log by the Log_event::write() member +which calls virtual write_data_header() and write_data_body() members +("data header" and "post header" are synonym in replication terminology). +In our case, data header is empty and data body is just the query: + + bool write_data_body(IO_CACHE *file) + { + return my_b_safe_write(file, (uchar*) m_query_txt, m_query_len); + } + +Printing the event is just printing the query: + + void Annotate_rows_log_event::print(FILE *file, PRINT_EVENT_INFO *pinfo) + { + my_b_printf(&pinfo->head_cache, "\tQuery: `%s`\n", m_query_txt); + } + +3. How Master writes Annotate_rows events to the binary log +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +The event is written to the binary log just before the group of Table_map +events which precede corresponding Rows events (one query may generate +several Table map events in the binary log, but the corresponding +Annotate_rows event must be written only once before the first Table map +event; hence the boolean variable 'with_annotate' below): + + int write_locked_table_maps(THD *thd) + { ... + bool with_annotate= thd->variables.binlog_annotate_rows_events; + ... + for (uint i= 0; i < ... <number of tables> ...; ++i) + { ... + thd->binlog_write_table_map(table, ..., with_annotate); + with_annotate= 0; // write Annotate_event not more than once + ... + } + ... + } + + int THD::binlog_write_table_map(TABLE *table, ..., bool with_annotate) + { ... + Table_map_log_event the_event(...); + ... + if (with_annotate) + { + Annotate_rows_log_event anno(this); + mysql_bin_log.write(&anno); + } + + mysql_bin_log.write(&the_event); + ... + } + +4. How slave treats replicate-annotate-rows-events option +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +The replicate-annotate-rows-events option is treated just as the session +value of the binlog_annotate_rows_events variable for the slave IO and +SQL threads. This setting is done during initialization of these threads: + + pthread_handler_t handle_slave_io(void *arg) + { + THD *thd= new THD; + ... + init_slave_thread(thd, SLAVE_THD_IO); + ... + } + + pthread_handler_t handle_slave_sql(void *arg) + { + THD *thd= new THD; + ... + init_slave_thread(thd, SLAVE_THD_SQL); + ... + } + + int init_slave_thread(THD* thd, SLAVE_THD_TYPE thd_type) + { ... + thd->variables.binlog_annotate_rows_events= + opt_replicate_annotate_rows_events; + ... + } + +5. How slave IO thread requests Annotate_rows events +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +When requesting an event, the slave should inform the master whether +it should send Annotate_rows events or not. To that end we add a new +BINLOG_SEND_ANNOTATE_ROWS_EVENT flag used when requesting an event: + + #define BINLOG_DUMP_NON_BLOCK 1 + #define BINLOG_SEND_ANNOTATE_ROWS_EVENT 2 + + pthread_handler_t handle_slave_io(void *arg) + { ... + request_dump(mysql, ...); + ... + } + + int request_dump(MYSQL* mysql, ...) + { ... + if (opt_log_slave_updates && + mi->io_thd->variables.binlog_annotate_rows_events) + binlog_flags|= BINLOG_SEND_ANNOTATE_ROWS_EVENT; + ... + int2store(buf + 4, binlog_flags); + ... + simple_command(mysql, COM_BINLOG_DUMP, buf, ...); + ... + } + +6. How master executes the request +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ + + case COM_BINLOG_DUMP: + { ... + flags= uint2korr(packet + 4); + ... + mysql_binlog_send(thd, ..., flags); + ... + } + + void mysql_binlog_send(THD* thd, ..., ushort flags) + { ... + Log_event::read_log_event(&log, packet, ...); + ... + if ((*packet)[EVENT_TYPE_OFFSET + 1] != ANNOTATE_ROWS_EVENT || + flags & BINLOG_SEND_ANNOTATE_ROWS_EVENT || + thd->server_id == 0 /* slave == mysqlbinlog */ ) + { + my_net_write(net, packet->ptr(), packet->length()); + } + ... + } + +7. How slave SQL thread processes Annotate_rows events +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +The slave processes each recieved event by "applying" it, i.e. by +calling the Log_event::apply_event() function which in turn calls +the virtual do_apply_event() member specific for each type of the +event. + + int exec_relay_log_event(THD* thd, Relay_log_info* rli) + { ... + Log_event *ev = next_event(rli); + ... + apply_event_and_update_pos(ev, ...); + + if (ev->get_type_code() != FORMAT_DESCRIPTION_EVENT) + delete ev; + ... + } + + int apply_event_and_update_pos(Log_event *ev, ...) + { ... + ev->apply_event(...); + ... + } + + int Log_event::apply_event(...) + { + return do_apply_event(...); + } + +What does it mean to "apply" an Annotate_rows event? It means to set current +thd query to that of the described by the event, i.e. to the query which +caused the subsequent Rows events (see "How Master writes Annotate_rows +events to the binary log" to follow what happens further when the subsequent +Rows events is applied): + + int Annotate_rows_log_event::do_apply_event(...) + { + thd->set_query(m_query_txt, m_query_len); + } + +NOTE. I am not sure, but possibly current values of thd->query and +thd->query_length should be saved before calling set_query() and to be +restored on the Annotate_rows_log_event object deletion. +Is it really needed ? + +After calling this do_apply_event() function we may not delete the +Annotate_rows_log_event object immediatedly (see exec_relay_log_event() +above) because thd->query now points to the string inside this object. +We may keep the pointer to this object in the Relay_log_info: + + class Relay_log_info + { + public: + ... + void set_annotate_event(Annotate_rows_log_event*); + Annotate_rows_log_event* get_annotate_event(); + void free_annotate_event(); + ... + private: + Annotate_rows_log_event* m_annotate_event; + }; + +When the saved Annotate_rows object may be deleted? When all corresponding +Rows events will be processed, i.e. before processing the first non-Rows +event (note that Annotate_rows object resides in the binary log *after* +the (possible) 'BEGIN' Query event which accompanies the rows events; note +also that this deletion is adjusted with the case when some or all +corresponding Rows events are filtered out by replicate filter rules): + + int exec_relay_log_event(THD* thd, Relay_log_info* rli) + { ... + Log_event *ev= next_event(rli); + ... + if (rli->get_annotate_event() && !IS_RBR_EVENT_TYPE(ev->get_type_code())) + rli->free_annotate_event(); + + apply_event_and_update_pos(ev, ...); + + if (ev->get_type_code() == ANNOTATE_ROWS_EVENT) + rli->set_annotate_event((Annotate_rows_log_event*) ev); + else if (ev->get_type_code() != FORMAT_DESCRIPTION_EVENT) + delete ev; + ... + } + +where + + #define IS_RBR_EVENT_TYPE(type) ( (type) == TABLE_MAP_EVENT || \ + (type) == WRITE_ROWS_EVENT || \ + (type) == UPDATE_ROWS_EVENT || \ + (type) == DELETE_ROWS_EVENT ) + +8. General remarks +~~~~~~~~~~~~~~~~~~ +Kristian noticed that introducing new log event type should be coordinated +somehow with MySQL/Sun: + + Kristian: The numeric code for this event must be assigned carefully. + It should be coordinated with MySQL/Sun, otherwise we can get into a + situation where MySQL uses the same numeric code for one event that + MariaDB uses for ANNOTATE_ROWS_EVENT, which would make merging the two + impossible. + Alex: I reserved about 20 numbers not to have possible conflicts + with MySQL. + Kristian: Still, I think it would be appropriate to send a polite email + to internals(a)lists.mysql.com about this and suggesting to reserve the + event number. + +Also we should notice the introduction of the BINLOG_SEND_ANNOTATE_ROWS_EVENT +flag taking into account that MySQL/Sun may also introduce a flag with the +same value to be used in the request_dump-mysql_binlog_send interface. +But this is mainly the question of merging: if a conflict concerning this +flag occur, we may simply change the BINLOG_SEND_ANNOTATE_ROWS_EVENT value +(this does not require additional changes in the code). -=-=(Alexi - Sat, 19 Dec 2009, 15:41)=-=- High-Level Specification modified. --- /tmp/wklog.47.old.14545 2009-12-19 15:41:21.000000000 +0200 +++ /tmp/wklog.47.new.14545 2009-12-19 15:41:21.000000000 +0200 @@ -1,122 +1,107 @@ -First suggestion: - -> I think for this we would actually need a new binlog event type -> (Comment_log_event?). Unless we want to log an empty statement Query_log_event -> containing only a comment (a bit of a hack). - -New server option -~~~~~~~~~~~~~~~~~ - --binlog-annotate-rows-events - -Setting this option makes RBR (rows-) events in the binary log to be -preceded by Annotate rows events (see below). The corresponding -'binlog_annotate_rows_events' system variable is dynamic and has both -global and session values. Default global value is OFF. - -Note. Session values are usefull to make it possible to annotate only - some selected statements: +Content +~~~~~~~ + 1. Annotate_rows_log_event + 2. Server option: --binlog-annotate-rows-events + 3. Server option: --replicate-annotate-rows-events + 4. mysqlbinlog option: --print-annotate-rows-events + 5. mysqlbinlog output + +1. Annotate_rows_log_event [ ANNOTATE_ROWS_EVENT ] +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +Describes the query which caused the corresponding rows events. Has empty +post-header and contains the query text in its data part. Example: + + ************************ + ANNOTATE_RBR_EVENT + ************************ + 00000220 | B6 A0 2C 4B | time_when = 1261215926 + 00000224 | 33 | event_type = 51 + 00000225 | 64 00 00 00 | server_id = 100 + 00000229 | 36 00 00 00 | event_len = 54 + 0000022D | 56 02 00 00 | log_pos = 00000256 + 00000231 | 00 00 | flags = <none> + ------------------------ + 00000233 | 49 4E 53 45 | query = "INSERT INTO t1 VALUES (1), (2), (3)" + 00000237 | 52 54 20 49 | + 0000023B | 4E 54 4F 20 | + 0000023F | 74 31 20 56 | + 00000243 | 41 4C 55 45 | + 00000247 | 53 20 28 31 | + 0000024B | 29 2C 20 28 | + 0000024F | 32 29 2C 20 | + 00000253 | 28 33 29 | + ************************ + +In binary log, Annotate_rows event follows the (possible) 'BEGIN' Query event +and precedes the first of Table map events which accompany the corresponding +rows events. (See example in the "mysqlbinlog output" section below.) + +2. Server option: --binlog-annotate-rows-events +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +Tells the master to write Annotate_rows events to the binary log. + + * Variable Name: binlog_annotate_rows_events + * Scope: Global & Session + * Access Type: Dynamic + * Data Type: bool + * Default Value: OFF +NOTE. Session values allows to annotate only some selected statements: ... SET SESSION binlog_annotate_rows_events=ON; ... statements to be annotated ... SET SESSION binlog_annotate_rows_events=OFF; ... statements not to be annotated ... -New binlog event type -~~~~~~~~~~~~~~~~~~~~~ - Annotate_rows_log_event [ ANNOTATE_ROWS_EVENT ] - -Describes the query which caused the corresponding rows event. In binary log, -precedes each Table_map_log_event. Contains empty post-header and the query -text in its data part. - -The numeric code for this event must be assigned carefully. It should be -coordinated with MySQL/Sun, otherwise we can get into a situation where MySQL -uses the same numeric code for one event that MariaDB uses for -ANNOTATE_ROWS_EVENT, which would make merging the two impossible. - -Example: - - ... - ************************ - ANNOTATE_ROWS_EVENT [51] - ************************ - 000000C7 | 54 1B 12 4B | time_when = 1259477844 - 000000CB | 33 | event_type = 51 - 000000CC | 64 00 00 00 | server_id = 100 - 000000D0 | 2C 00 00 00 | event_len = 44 - 000000D4 | F3 00 00 00 | log_pos = 000000F3 - 000000D8 | 00 00 | flags = <none> - ------------------------ - 000000DA | 69 6E 73 65 | query = "insert into t1 values (1)" - 000000DE | 72 74 20 69 | - 000000E2 | 6E 74 6F 20 | - 000000E6 | 74 31 20 76 | - 000000EA | 61 6C 75 65 | - 000000EE | 73 20 28 31 | - 000000F2 | 29 | - ************************ - TABLE_MAP_EVENT [19] - ************************ - 000000F3 | 54 1B 12 4B | time_when = 1259477844 - 000000F7 | 13 | event_type = 19 - 000000F8 | 64 00 00 00 | server_id = 100 - 000000FC | 29 00 00 00 | event_len = 41 - 00000100 | 1C 01 00 00 | log_pos = 0000011C - 00000104 | 00 00 | flags = <none> - ------------------------ - ... - ************************ - WRITE_ROWS_EVENT [23] - ************************ - 0000011C | 54 1B 12 4B | time_when = 1259477844 - 00000120 | 17 | event_type = 23 - 00000121 | 64 00 00 00 | server_id = 100 - 00000125 | 22 00 00 00 | event_len = 34 - 00000129 | 3E 01 00 00 | log_pos = 0000013E - 0000012D | 10 00 | flags = LOG_EVENT_UPDATE_TABLE_MAP_VERSION_F - ------------------------ - 0000012F | 0F 00 00 00 | table_id = 15 - ... - -New mysqlbinlog option -~~~~~~~~~~~~~~~~~~~~~~ - --print-annotate-rows-events - -With this option, mysqlbinlog prints the content of Annotate-rows -events (if the binary log does contain them). Without this option -(i.e. by default), mysqlbinlog skips Annotate rows events. - - -mysqlbinlog output -~~~~~~~~~~~~~~~~~~ -Something like this: +3. Server option: --replicate-annotate-rows-events +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +Tells the slave to reproduce Annotate_rows events recieved from the master +in its own binary log (sensible only in pair with log-slave-updates option). + + * Variable Name: replicate_annotate_rows_events + * Scope: Global + * Access Type: Read only + * Data Type: bool + * Default Value: OFF + +NOTE. Why do we additionally need this 'replicate' option? Why not to make +the slave to reproduce this events when its binlog-annotate-rows-events +global value is ON? Well, because, for example, we may want to configure +the slave which should reproduce Annotate_rows events but has global +binlog-annotate-rows-events = OFF meaning this to be the default value for +the client threads (see also "How slave treats replicate-annotate-rows-events +option" in LLD part). + +4. mysqlbinlog option: --print-annotate-rows-events +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +With this option, mysqlbinlog prints the content of Annotate_rows events (if +the binary log does contain them). Without this option (i.e. by default), +mysqlbinlog skips Annotate_rows events. +5. mysqlbinlog output +~~~~~~~~~~~~~~~~~~~~~ +With --print-annotate-rows-events, mysqlbinlog outputs Annotate_rows events +in a form like this: ... - # at 199 - # at 243 - # at 284 - #091129 9:57:24 server id 100 end_log_pos 243 Query: `insert into t1 values -(1)` - #091129 9:57:24 server id 100 end_log_pos 284 Table_map: `test`.`t1` mapped -to number 15 - #091129 9:57:24 server id 100 end_log_pos 318 Write_rows: table id 15 + # at 1646 + #091219 12:45:26 server id 100 end_log_pos 1714 Query thread_id=1 +exec_time=0 error_code=0 + SET TIMESTAMP=1261215926/*!*/; + BEGIN + /*!*/; + # at 1714 + # at 1812 + # at 1853 + # at 1894 + # at 1938 + #091219 12:45:26 server id 100 end_log_pos 1812 Query: `DELETE t1, t2 FROM +t1 INNER JOIN t2 INNER JOIN t3 WHERE t1.a=t2.a AND t2.a=t3.a` + #091219 12:45:26 server id 100 end_log_pos 1853 Table_map: `test`.`t1` +mapped to number 16 + #091219 12:45:26 server id 100 end_log_pos 1894 Table_map: `test`.`t2` +mapped to number 17 + #091219 12:45:26 server id 100 end_log_pos 1938 Delete_rows: table id 16 + #091219 12:45:26 server id 100 end_log_pos 1982 Delete_rows: table id 17 flags: STMT_END_F - - BINLOG ' - VBsSSzNkAAAALAAAAPMAAAAAAGluc2VydCBpbnRvIHQxIHZhbHVlcyAoMSk= - VBsSSxNkAAAAKQAAABwBAAAAAA8AAAAAAAAABHRlc3QAAnQxAAEDAAE= - VBsSSxdkAAAAIgAAAD4BAAAQAA8AAAAAAAEAAf/+AQAAAA== - '/*!*/; - ### INSERT INTO test.t1 - ### SET - ### @1=1 /* INT meta=0 nullable=1 is_null=0 */ ... -When master sends Annotate rows events -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -1. Master always sends Annotate_rows events to mysqlbinlog (in - remote case). -2. Master sends Annotate_rows events to a slave only if the slave has - both log-slave-updates and binlog-annotate-rows-events options set. - -=-=(Bothorsen - Fri, 18 Dec 2009, 16:22)=-=- Add estimation time. Worked 5 hours and estimate 35 hours remain (original estimate increased by 5 hours). -=-=(Bothorsen - Fri, 18 Dec 2009, 16:16)=-=- This is the work done on this patch so far. Most of it done by Alex. Worked 15 hours and estimate 035 hours remain (original estimate increased by 50 hours). -=-=(Alexi - Fri, 04 Dec 2009, 13:00)=-=- High-Level Specification modified. --- /tmp/wklog.47.old.14001 2009-12-04 13:00:24.000000000 +0200 +++ /tmp/wklog.47.new.14001 2009-12-04 13:00:24.000000000 +0200 @@ -6,27 +6,27 @@ New server option ~~~~~~~~~~~~~~~~~ - --binlog-annotate-row-events + --binlog-annotate-rows-events -Setting this option makes RBR (row-) events in the binary log to be +Setting this option makes RBR (rows-) events in the binary log to be preceded by Annotate rows events (see below). The corresponding -'binlog_annotate_row_events' system variable is dynamic and has both +'binlog_annotate_rows_events' system variable is dynamic and has both global and session values. Default global value is OFF. Note. Session values are usefull to make it possible to annotate only some selected statements: ... - SET SESSION binlog_annotate_row_events=ON; + SET SESSION binlog_annotate_rows_events=ON; ... statements to be annotated ... - SET SESSION binlog_annotate_row_events=OFF; + SET SESSION binlog_annotate_rows_events=OFF; ... statements not to be annotated ... New binlog event type ~~~~~~~~~~~~~~~~~~~~~ Annotate_rows_log_event [ ANNOTATE_ROWS_EVENT ] -Describes the query which caused the corresponding row event. In binary log, +Describes the query which caused the corresponding rows event. In binary log, precedes each Table_map_log_event. Contains empty post-header and the query text in its data part. @@ -79,6 +79,15 @@ 0000012F | 0F 00 00 00 | table_id = 15 ... +New mysqlbinlog option +~~~~~~~~~~~~~~~~~~~~~~ + --print-annotate-rows-events + +With this option, mysqlbinlog prints the content of Annotate-rows +events (if the binary log does contain them). Without this option +(i.e. by default), mysqlbinlog skips Annotate rows events. + + mysqlbinlog output ~~~~~~~~~~~~~~~~~~ Something like this: @@ -109,5 +118,5 @@ 1. Master always sends Annotate_rows events to mysqlbinlog (in remote case). 2. Master sends Annotate_rows events to a slave only if the slave has - both log-slave-updates and binlog-annotate-row-events options set. + both log-slave-updates and binlog-annotate-rows-events options set. -=-=(Alexi - Wed, 02 Dec 2009, 13:32)=-=- Low Level Design modified. --- /tmp/wklog.47.old.17456 2009-12-02 13:32:18.000000000 +0200 +++ /tmp/wklog.47.new.17456 2009-12-02 13:32:18.000000000 +0200 @@ -1,8 +1 @@ -mysql_binlog_send() [sql/sql_repl.cc] -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ -1. When sending events to a slave, master should simply skip - Annotate_rows events (they are not needed for replication). - [ ??? Multi-master - currently not clear ] -2. When sending events to mysqlbinlog (remote case), master - must send Annotate_rows events as well. -=-=(Alexi - Wed, 02 Dec 2009, 13:31)=-=- High-Level Specification modified. --- /tmp/wklog.47.old.17414 2009-12-02 11:31:59.000000000 +0000 +++ /tmp/wklog.47.new.17414 2009-12-02 11:31:59.000000000 +0000 @@ -104,3 +104,10 @@ ### @1=1 /* INT meta=0 nullable=1 is_null=0 */ ... +When master sends Annotate rows events +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +1. Master always sends Annotate_rows events to mysqlbinlog (in + remote case). +2. Master sends Annotate_rows events to a slave only if the slave has + both log-slave-updates and binlog-annotate-row-events options set. + -=-=(Knielsen - Mon, 30 Nov 2009, 11:21)=-=- High-Level Specification modified. --- /tmp/wklog.47.old.18210 2009-11-30 11:21:33.000000000 +0200 +++ /tmp/wklog.47.new.18210 2009-11-30 11:21:33.000000000 +0200 @@ -28,7 +28,14 @@ Describes the query which caused the corresponding row event. In binary log, precedes each Table_map_log_event. Contains empty post-header and the query -text in its data part. Example: +text in its data part. + +The numeric code for this event must be assigned carefully. It should be +coordinated with MySQL/Sun, otherwise we can get into a situation where MySQL +uses the same numeric code for one event that MariaDB uses for +ANNOTATE_ROWS_EVENT, which would make merging the two impossible. + +Example: ... ************************ -=-=(Alexi - Mon, 30 Nov 2009, 10:33)=-=- Low Level Design modified. --- /tmp/wklog.47.old.16188 2009-11-30 10:33:44.000000000 +0200 +++ /tmp/wklog.47.new.16188 2009-11-30 10:33:44.000000000 +0200 @@ -2,6 +2,7 @@ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ 1. When sending events to a slave, master should simply skip Annotate_rows events (they are not needed for replication). + [ ??? Multi-master - currently not clear ] 2. When sending events to mysqlbinlog (remote case), master must send Annotate_rows events as well. -=-=(Alexi - Sun, 29 Nov 2009, 13:00)=-=- Low Level Design modified. --- /tmp/wklog.47.old.32047 2009-11-29 13:00:21.000000000 +0200 +++ /tmp/wklog.47.new.32047 2009-11-29 13:00:21.000000000 +0200 @@ -1 +1,7 @@ +mysql_binlog_send() [sql/sql_repl.cc] +~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ +1. When sending events to a slave, master should simply skip + Annotate_rows events (they are not needed for replication). +2. When sending events to mysqlbinlog (remote case), master + must send Annotate_rows events as well. ------------------------------------------------------------ -=-=(View All Progress Notes, 14 total)=-=- http://askmonty.org/worklog/index.pl?tid=47&nolimit=1 DESCRIPTION: Store in binlog (and show in mysqlbinlog output) texts of statements that caused RBR events This is needed for (list from Monty): - Easier to understand why updates happened - Would make it easier to find out where in application things went wrong (as you can search for exact strings) - Allow one to filter things based on comments in the statement. The cost of this can be that the binlog will be approximately 2x in size (especially insert of big blob's would be a bit painful), so this should be an optional feature. HIGH-LEVEL SPECIFICATION: Content ~~~~~~~ 1. Annotate_rows_log_event 2. Server option: --binlog-annotate-rows-events 3. Server option: --replicate-annotate-rows-events 4. mysqlbinlog option: --print-annotate-rows-events 5. mysqlbinlog output 1. Annotate_rows_log_event [ ANNOTATE_ROWS_EVENT ] ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Describes the query which caused the corresponding rows events. Has empty post-header and contains the query text in its data part. Example: ************************ ANNOTATE_RBR_EVENT ************************ 00000220 | B6 A0 2C 4B | time_when = 1261215926 00000224 | 33 | event_type = 51 00000225 | 64 00 00 00 | server_id = 100 00000229 | 36 00 00 00 | event_len = 54 0000022D | 56 02 00 00 | log_pos = 00000256 00000231 | 00 00 | flags = <none> ------------------------ 00000233 | 49 4E 53 45 | query = "INSERT INTO t1 VALUES (1), (2), (3)" 00000237 | 52 54 20 49 | 0000023B | 4E 54 4F 20 | 0000023F | 74 31 20 56 | 00000243 | 41 4C 55 45 | 00000247 | 53 20 28 31 | 0000024B | 29 2C 20 28 | 0000024F | 32 29 2C 20 | 00000253 | 28 33 29 | ************************ In binary log, Annotate_rows event follows the (possible) 'BEGIN' Query event and precedes the first of Table map events which accompany the corresponding rows events. (See example in the "mysqlbinlog output" section below.) 2. Server option: --binlog-annotate-rows-events ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Tells the master to write Annotate_rows events to the binary log. * Variable Name: binlog_annotate_rows_events * Scope: Global & Session * Access Type: Dynamic * Data Type: bool * Default Value: OFF NOTE. Session values allows to annotate only some selected statements: ... SET SESSION binlog_annotate_rows_events=ON; ... statements to be annotated ... SET SESSION binlog_annotate_rows_events=OFF; ... statements not to be annotated ... 3. Server option: --replicate-annotate-rows-events ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Tells the slave to reproduce Annotate_rows events recieved from the master in its own binary log (sensible only in pair with log-slave-updates option). * Variable Name: replicate_annotate_rows_events * Scope: Global * Access Type: Read only * Data Type: bool * Default Value: OFF NOTE. Why do we additionally need this 'replicate' option? Why not to make the slave to reproduce this events when its binlog-annotate-rows-events global value is ON? Well, because, for example, we may want to configure the slave which should reproduce Annotate_rows events but has global binlog-annotate-rows-events = OFF meaning this to be the default value for the client threads (see also "How slave treats replicate-annotate-rows-events option" in LLD part). 4. mysqlbinlog option: --print-annotate-rows-events ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ With this option, mysqlbinlog prints the content of Annotate_rows events (if the binary log does contain them). Without this option (i.e. by default), mysqlbinlog skips Annotate_rows events. 5. mysqlbinlog output ~~~~~~~~~~~~~~~~~~~~~ With --print-annotate-rows-events, mysqlbinlog outputs Annotate_rows events in a form like this: ... # at 1646 #091219 12:45:26 server id 100 end_log_pos 1714 Query thread_id=1 exec_time=0 error_code=0 SET TIMESTAMP=1261215926/*!*/; BEGIN /*!*/; # at 1714 # at 1812 # at 1853 # at 1894 # at 1938 #091219 12:45:26 server id 100 end_log_pos 1812 Query: `DELETE t1, t2 FROM t1 INNER JOIN t2 INNER JOIN t3 WHERE t1.a=t2.a AND t2.a=t3.a` #091219 12:45:26 server id 100 end_log_pos 1853 Table_map: `test`.`t1` mapped to number 16 #091219 12:45:26 server id 100 end_log_pos 1894 Table_map: `test`.`t2` mapped to number 17 #091219 12:45:26 server id 100 end_log_pos 1938 Delete_rows: table id 16 #091219 12:45:26 server id 100 end_log_pos 1982 Delete_rows: table id 17 flags: STMT_END_F ... LOW-LEVEL DESIGN: Content ~~~~~~~ 1. Annotate_rows event number 2. Outline of Annotate_rows event behavior 3. How Master writes Annotate_rows events to the binary log 4. How slave treats replicate-annotate-rows-events option 5. How slave IO thread requests Annotate_rows events 6. How master executes the request 7. How slave SQL thread processes Annotate_rows events 8. General remarks 1. Annotate_rows event number ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ To avoid possible event numbers conflict with MySQL/Sun, we leave a gap between the last MySQL event number and the Annotate_rows event number: enum Log_event_type { ... INCIDENT_EVENT= 26, // New MySQL event numbers are to be added here MYSQL_EVENTS_END, MARIA_EVENTS_BEGIN= 51, // New Maria event numbers start from here ANNOTATE_ROWS_EVENT= 51, ENUM_END_EVENT }; together with the corresponding extension of 'post_header_len' array in the Format description event. (This extension does not affect the compatibility of the binary log). Here is how Format description event looks like with this extension: ************************ FORMAT_DESCRIPTION_EVENT ************************ 00000004 | A1 A0 2C 4B | time_when = 1261215905 00000008 | 0F | event_type = 15 00000009 | 64 00 00 00 | server_id = 100 0000000D | 7F 00 00 00 | event_len = 127 00000011 | 83 00 00 00 | log_pos = 00000083 00000015 | 01 00 | flags = LOG_EVENT_BINLOG_IN_USE_F ------------------------ 00000017 | 04 00 | binlog_ver = 4 00000019 | 35 2E 32 2E | server_ver = 5.2.0-MariaDB-alpha-debug-log ..... ... 0000004B | A1 A0 2C 4B | time_created = 1261215905 0000004F | 13 | common_header_len = 19 ------------------------ post_header_len ------------------------ 00000050 | 38 | 56 - START_EVENT_V3 [1] ..... ... 00000069 | 02 | 2 - INCIDENT_EVENT [26] 0000006A | 00 | 0 - RESERVED [27] ..... ... 00000081 | 00 | 0 - RESERVED [50] 00000082 | 00 | 0 - ANNOTATE_RBR_EVENT [51] ************************ 2. Outline of Annotate_rows event behavior ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Each Annotate_rows_log_event object has two private members describing the corresponding query: char *m_query_txt; uint m_query_len; When the object is created for writing to a binary log, this query is taken from 'thd' (for short, below we omit the 'Annotate_rows_log_event::' prefix as well as other implementation details): Annotate_rows_log_event(THD *thd) { m_query_txt = thd->query(); m_query_len = thd->query_length(); } When the object is read from a binary log, the query is taken from the buffer containing the binary log representation of the event (this buffer is allocated in Log_event object from which all Log events are derived): Annotate_rows_log_event(char *buf, uint event_len, Format_description_log_event *desc) { m_query_len = event_len - desc->common_header_len; m_query_txt = buf + desc->common_header_len; } The events are written to the binary log by the Log_event::write() member which calls virtual write_data_header() and write_data_body() members ("data header" and "post header" are synonym in replication terminology). In our case, data header is empty and data body is just the query: bool write_data_body(IO_CACHE *file) { return my_b_safe_write(file, (uchar*) m_query_txt, m_query_len); } Printing the event is just printing the query: void Annotate_rows_log_event::print(FILE *file, PRINT_EVENT_INFO *pinfo) { my_b_printf(&pinfo->head_cache, "\tQuery: `%s`\n", m_query_txt); } 3. How Master writes Annotate_rows events to the binary log ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The event is written to the binary log just before the group of Table_map events which precede corresponding Rows events (one query may generate several Table map events in the binary log, but the corresponding Annotate_rows event must be written only once before the first Table map event; hence the boolean variable 'with_annotate' below): int write_locked_table_maps(THD *thd) { ... bool with_annotate= thd->variables.binlog_annotate_rows_events; ... for (uint i= 0; i < ... <number of tables> ...; ++i) { ... thd->binlog_write_table_map(table, ..., with_annotate); with_annotate= 0; // write Annotate_event not more than once ... } ... } int THD::binlog_write_table_map(TABLE *table, ..., bool with_annotate) { ... Table_map_log_event the_event(...); ... if (with_annotate) { Annotate_rows_log_event anno(this); mysql_bin_log.write(&anno); } mysql_bin_log.write(&the_event); ... } 4. How slave treats replicate-annotate-rows-events option ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The replicate-annotate-rows-events option is treated just as the session value of the binlog_annotate_rows_events variable for the slave IO and SQL threads. This setting is done during initialization of these threads: pthread_handler_t handle_slave_io(void *arg) { THD *thd= new THD; ... init_slave_thread(thd, SLAVE_THD_IO); ... } pthread_handler_t handle_slave_sql(void *arg) { THD *thd= new THD; ... init_slave_thread(thd, SLAVE_THD_SQL); ... } int init_slave_thread(THD* thd, SLAVE_THD_TYPE thd_type) { ... thd->variables.binlog_annotate_rows_events= opt_replicate_annotate_rows_events; ... } 5. How slave IO thread requests Annotate_rows events ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ When requesting an event, the slave should inform the master whether it should send Annotate_rows events or not. To that end we add a new BINLOG_SEND_ANNOTATE_ROWS_EVENT flag used when requesting an event: #define BINLOG_DUMP_NON_BLOCK 1 #define BINLOG_SEND_ANNOTATE_ROWS_EVENT 2 pthread_handler_t handle_slave_io(void *arg) { ... request_dump(mysql, ...); ... } int request_dump(MYSQL* mysql, ...) { ... if (opt_log_slave_updates && mi->io_thd->variables.binlog_annotate_rows_events) binlog_flags|= BINLOG_SEND_ANNOTATE_ROWS_EVENT; ... int2store(buf + 4, binlog_flags); ... simple_command(mysql, COM_BINLOG_DUMP, buf, ...); ... } 6. How master executes the request ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ case COM_BINLOG_DUMP: { ... flags= uint2korr(packet + 4); ... mysql_binlog_send(thd, ..., flags); ... } void mysql_binlog_send(THD* thd, ..., ushort flags) { ... Log_event::read_log_event(&log, packet, ...); ... if ((*packet)[EVENT_TYPE_OFFSET + 1] != ANNOTATE_ROWS_EVENT || flags & BINLOG_SEND_ANNOTATE_ROWS_EVENT || thd->server_id == 0 /* slave == mysqlbinlog */ ) { my_net_write(net, packet->ptr(), packet->length()); } ... } 7. How slave SQL thread processes Annotate_rows events ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ The slave processes each recieved event by "applying" it, i.e. by calling the Log_event::apply_event() function which in turn calls the virtual do_apply_event() member specific for each type of the event. int exec_relay_log_event(THD* thd, Relay_log_info* rli) { ... Log_event *ev = next_event(rli); ... apply_event_and_update_pos(ev, ...); if (ev->get_type_code() != FORMAT_DESCRIPTION_EVENT) delete ev; ... } int apply_event_and_update_pos(Log_event *ev, ...) { ... ev->apply_event(...); ... } int Log_event::apply_event(...) { return do_apply_event(...); } What does it mean to "apply" an Annotate_rows event? It means to set current thd query to that of the described by the event, i.e. to the query which caused the subsequent Rows events (see "How Master writes Annotate_rows events to the binary log" to follow what happens further when the subsequent Rows events is applied): int Annotate_rows_log_event::do_apply_event(...) { thd->set_query(m_query_txt, m_query_len); } NOTE. I am not sure, but possibly current values of thd->query and thd->query_length should be saved before calling set_query() and to be restored on the Annotate_rows_log_event object deletion. Is it really needed ? After calling this do_apply_event() function we may not delete the Annotate_rows_log_event object immediatedly (see exec_relay_log_event() above) because thd->query now points to the string inside this object. We may keep the pointer to this object in the Relay_log_info: class Relay_log_info { public: ... void set_annotate_event(Annotate_rows_log_event*); Annotate_rows_log_event* get_annotate_event(); void free_annotate_event(); ... private: Annotate_rows_log_event* m_annotate_event; }; When the saved Annotate_rows object may be deleted? When all corresponding Rows events will be processed, i.e. before processing the first non-Rows event (note that Annotate_rows object resides in the binary log *after* the (possible) 'BEGIN' Query event which accompanies the rows events; note also that this deletion is adjusted with the case when some or all corresponding Rows events are filtered out by replicate filter rules): int exec_relay_log_event(THD* thd, Relay_log_info* rli) { ... Log_event *ev= next_event(rli); ... if (rli->get_annotate_event() && !IS_RBR_EVENT_TYPE(ev->get_type_code())) rli->free_annotate_event(); apply_event_and_update_pos(ev, ...); if (ev->get_type_code() == ANNOTATE_ROWS_EVENT) rli->set_annotate_event((Annotate_rows_log_event*) ev); else if (ev->get_type_code() != FORMAT_DESCRIPTION_EVENT) delete ev; ... } where #define IS_RBR_EVENT_TYPE(type) ( (type) == TABLE_MAP_EVENT || \ (type) == WRITE_ROWS_EVENT || \ (type) == UPDATE_ROWS_EVENT || \ (type) == DELETE_ROWS_EVENT ) 8. General remarks ~~~~~~~~~~~~~~~~~~ Kristian noticed that introducing new log event type should be coordinated somehow with MySQL/Sun: Kristian: The numeric code for this event must be assigned carefully. It should be coordinated with MySQL/Sun, otherwise we can get into a situation where MySQL uses the same numeric code for one event that MariaDB uses for ANNOTATE_ROWS_EVENT, which would make merging the two impossible. Alex: I reserved about 20 numbers not to have possible conflicts with MySQL. Kristian: Still, I think it would be appropriate to send a polite email to internals(a)lists.mysql.com about this and suggesting to reserve the event number. Also we should notice the introduction of the BINLOG_SEND_ANNOTATE_ROWS_EVENT flag taking into account that MySQL/Sun may also introduce a flag with the same value to be used in the request_dump-mysql_binlog_send interface. But this is mainly the question of merging: if a conflict concerning this flag occur, we may simply change the BINLOG_SEND_ANNOTATE_ROWS_EVENT value (this does not require additional changes in the code). ESTIMATED WORK TIME ESTIMATED COMPLETION DATE ----------------------------------------------------------------------- WorkLog (v3.5.9)

3 2