~tsarev/percona-server/5.5-processlist_rows_stats-sporadic_fails-fix : contents of innodb_overwrite_relay_log

~tsarev/percona-server/5.5-processlist_rows_stats-sporadic_fails-fix : (revision 125)
# name       : innodb_overwrite_relay_log_info.patch
# introduced : 11 or before
# maintainer : Yasufumi
#
#!!! notice !!!
# Any small change to this file in the main branch
# should be done or reviewed by the maintainer!
diff -ruN a/storage/innobase/handler/ha_innodb.cc b/storage/innobase/handler/ha_innodb.cc
--- a/storage/innobase/handler/ha_innodb.cc	2010-12-03 15:37:45.516105468 +0900
+++ b/storage/innobase/handler/ha_innodb.cc	2010-12-03 15:38:20.318952987 +0900
@@ -42,6 +42,8 @@
 #pragma implementation				// gcc: Class implementation
 #endif
 
+#define MYSQL_SERVER
+
 #include <sql_table.h>	// explain_filename, nz2, EXPLAIN_PARTITIONS_AS_COMMENT,
 			// EXPLAIN_FILENAME_MAX_EXTRA_LENGTH
 
@@ -53,6 +55,15 @@
 #include <mysql/psi/psi.h>
 #include <my_sys.h>
 
+#ifdef MYSQL_SERVER
+#include <rpl_mi.h>
+#include <slave.h>
+// Defined in slave.cc
+int init_intvar_from_file(int* var, IO_CACHE* f, int default_val);
+int init_strvar_from_file(char *var, int max_size, IO_CACHE *f,
+			  const char *default_val);
+#endif /* MYSQL_SERVER */
+
 /** @file ha_innodb.cc */
 
 /* Include necessary InnoDB headers */
@@ -91,6 +102,14 @@
 #include "ha_innodb.h"
 #include "i_s.h"
 
+#ifdef MYSQL_SERVER
+// Defined in trx0sys.c
+extern char		trx_sys_mysql_master_log_name[];
+extern ib_int64_t	trx_sys_mysql_master_log_pos;
+extern char		trx_sys_mysql_relay_log_name[];
+extern ib_int64_t	trx_sys_mysql_relay_log_pos;
+#endif /* MYSQL_SERVER */
+
 # ifndef MYSQL_PLUGIN_IMPORT
 #  define MYSQL_PLUGIN_IMPORT /* nothing */
 # endif /* MYSQL_PLUGIN_IMPORT */
@@ -163,6 +182,7 @@
 static my_bool	innobase_use_doublewrite		= TRUE;
 static my_bool	innobase_use_checksums			= TRUE;
 static my_bool	innobase_locks_unsafe_for_binlog	= FALSE;
+static my_bool	innobase_overwrite_relay_log_info	= FALSE;
 static my_bool	innobase_rollback_on_timeout		= FALSE;
 static my_bool	innobase_create_status_file		= FALSE;
 static my_bool	innobase_stats_on_metadata		= TRUE;
@@ -2248,6 +2268,89 @@
 	}
 #endif /* DBUG_OFF */
 
+#ifndef MYSQL_SERVER
+	innodb_overwrite_relay_log_info = FALSE;
+#endif
+
+#ifdef HAVE_REPLICATION
+#ifdef MYSQL_SERVER
+	/* read master log position from relay-log.info if exists */
+	char fname[FN_REFLEN+128];
+	int pos;
+	int info_fd;
+	IO_CACHE info_file;
+
+	fname[0] = '\0';
+
+	if(innobase_overwrite_relay_log_info) {
+
+	fprintf(stderr,
+		"InnoDB: Warning: innodb_overwrite_relay_log_info is enabled."
+		" Updates by other storage engines may not be synchronized.\n");
+
+	bzero((char*) &info_file, sizeof(info_file));
+	fn_format(fname, relay_log_info_file, mysql_data_home, "", 4+32);
+
+	int error=0;
+
+	if (!access(fname,F_OK)) {
+		/* exist */
+		if ((info_fd = my_open(fname, O_RDWR|O_BINARY, MYF(MY_WME))) < 0) {
+			error=1;
+		} else if (init_io_cache(&info_file, info_fd, IO_SIZE*2,
+					READ_CACHE, 0L, 0, MYF(MY_WME))) {
+			error=1;
+		}
+
+		if (error) {
+relay_info_error:
+			if (info_fd >= 0)
+				my_close(info_fd, MYF(0));
+			fname[0] = '\0';
+			goto skip_relay;
+		}
+	} else {
+		fname[0] = '\0';
+		goto skip_relay;
+	}
+
+	if (init_strvar_from_file(fname, sizeof(fname), &info_file, "") || /* dummy (it is relay-log) */
+	    init_intvar_from_file(&pos, &info_file, BIN_LOG_HEADER_SIZE)) { 
+		end_io_cache(&info_file);
+		error=1;
+		goto relay_info_error;
+	}
+
+	fprintf(stderr,
+		"InnoDB: relay-log.info is detected.\n"
+		"InnoDB: relay log: position %u, file name %s\n",
+		pos, fname);
+
+	strncpy(trx_sys_mysql_relay_log_name, fname, TRX_SYS_MYSQL_MASTER_LOG_NAME_LEN);
+	trx_sys_mysql_relay_log_pos = (ib_int64_t) pos;
+
+	if (init_strvar_from_file(fname, sizeof(fname), &info_file, "") ||
+	    init_intvar_from_file(&pos, &info_file, 0)) {
+		end_io_cache(&info_file);
+		error=1;
+		goto relay_info_error;
+	}
+
+	fprintf(stderr,
+		"InnoDB: master log: position %u, file name %s\n",
+		pos, fname);
+
+	strncpy(trx_sys_mysql_master_log_name, fname, TRX_SYS_MYSQL_MASTER_LOG_NAME_LEN);
+	trx_sys_mysql_master_log_pos = (ib_int64_t) pos;
+
+	end_io_cache(&info_file);
+	if (info_fd >= 0)
+		my_close(info_fd, MYF(0));
+	}
+skip_relay:
+#endif /* MYSQL_SERVER */
+#endif /* HAVE_REPLICATION */
+
 	/* Check that values don't overflow on 32-bit systems. */
 	if (sizeof(ulint) == 4) {
 		if (innobase_buffer_pool_size > UINT_MAX32) {
@@ -2546,6 +2649,76 @@
 		goto mem_free_and_error;
 	}
 
+#ifdef HAVE_REPLICATION
+#ifdef MYSQL_SERVER
+	if(innobase_overwrite_relay_log_info) {
+	/* If InnoDB progressed from relay-log.info, overwrite it */
+	if (fname[0] == '\0') {
+		fprintf(stderr,
+			"InnoDB: Something is wrong with the file relay-info.log. InnoDB will not overwrite it.\n");
+	} else if (0 != strcmp(fname, trx_sys_mysql_master_log_name)
+		   || pos != trx_sys_mysql_master_log_pos) {
+		/* Overwrite relay-log.info */
+		bzero((char*) &info_file, sizeof(info_file));
+		fn_format(fname, relay_log_info_file, mysql_data_home, "", 4+32);
+
+		int error = 0;
+
+		if (!access(fname,F_OK)) {
+			/* exist */
+			if ((info_fd = my_open(fname, O_RDWR|O_BINARY, MYF(MY_WME))) < 0) {
+				error = 1;
+			} else if (init_io_cache(&info_file, info_fd, IO_SIZE*2,
+						WRITE_CACHE, 0L, 0, MYF(MY_WME))) {
+				error = 1;
+			}
+
+			if (error) {
+				if (info_fd >= 0)
+					my_close(info_fd, MYF(0));
+				goto skip_overwrite;
+			}
+		} else {
+			error = 1;
+			goto skip_overwrite;
+		}
+
+		char buff[FN_REFLEN*2+22*2+4], *pos;
+
+		my_b_seek(&info_file, 0L);
+		pos=strmov(buff, trx_sys_mysql_relay_log_name);
+		*pos++='\n';
+		pos=longlong2str(trx_sys_mysql_relay_log_pos, pos, 10);
+		*pos++='\n';
+		pos=strmov(pos, trx_sys_mysql_master_log_name);
+		*pos++='\n';
+		pos=longlong2str(trx_sys_mysql_master_log_pos, pos, 10);
+		*pos='\n';
+
+		if (my_b_write(&info_file, (uchar*) buff, (size_t) (pos-buff)+1))
+			error = 1;
+		if (flush_io_cache(&info_file))
+			error = 1;
+
+		end_io_cache(&info_file);
+		if (info_fd >= 0)
+			my_close(info_fd, MYF(0));
+skip_overwrite:
+		if (error) {
+			fprintf(stderr,
+				"InnoDB: ERROR: An error occurred while overwriting relay-log.info.\n");
+		} else {
+			fprintf(stderr,
+				"InnoDB: The file relay-log.info was successfully overwritten.\n");
+		}
+	} else {
+		fprintf(stderr,
+			"InnoDB: InnoDB and relay-log.info are synchronized. InnoDB will not overwrite it.\n");
+	}
+	}
+#endif /* MYSQL_SERVER */
+#endif /* HAVE_REPLICATION */
+
 	innobase_old_blocks_pct = buf_LRU_old_ratio_update(
 		innobase_old_blocks_pct, TRUE);
 
@@ -2658,6 +2831,25 @@
 	trx_t*	trx)	/*!< in: transaction handle */
 {
 	if (trx_is_started(trx)) {
+#ifdef HAVE_REPLICATION
+#ifdef MYSQL_SERVER
+		THD *thd=current_thd;
+
+		if (thd && thd->slave_thread) {
+			/* Update the replication position info inside InnoDB */
+			trx->mysql_master_log_file_name
+				= active_mi->rli.group_master_log_name;
+			trx->mysql_master_log_pos
+				= ((ib_int64_t)active_mi->rli.group_master_log_pos +
+				   ((ib_int64_t)active_mi->rli.future_event_relay_log_pos -
+				    (ib_int64_t)active_mi->rli.group_relay_log_pos));
+			trx->mysql_relay_log_file_name
+				= active_mi->rli.group_relay_log_name;
+			trx->mysql_relay_log_pos
+				= (ib_int64_t)active_mi->rli.future_event_relay_log_pos;
+		}
+#endif /* MYSQL_SERVER */
+#endif /* HAVE_REPLICATION */
 
 		trx_commit_for_mysql(trx);
 	}
@@ -11014,6 +11206,12 @@
   "The common part for InnoDB table spaces.",
   NULL, NULL, NULL);
 
+static MYSQL_SYSVAR_BOOL(recovery_update_relay_log, innobase_overwrite_relay_log_info,
+  PLUGIN_VAR_NOCMDARG | PLUGIN_VAR_READONLY,
+  "During InnoDB crash recovery on slave overwrite relay-log.info "
+  "to align master log file position if information in InnoDB and relay-log.info is different.",
+  NULL, NULL, FALSE);
+
 static MYSQL_SYSVAR_BOOL(doublewrite, innobase_use_doublewrite,
   PLUGIN_VAR_NOCMDARG | PLUGIN_VAR_READONLY,
   "Enable InnoDB doublewrite buffer (enabled by default). "
@@ -11474,6 +11672,7 @@
   MYSQL_SYSVAR(old_blocks_pct),
   MYSQL_SYSVAR(old_blocks_time),
   MYSQL_SYSVAR(open_files),
+  MYSQL_SYSVAR(recovery_update_relay_log),
   MYSQL_SYSVAR(rollback_on_timeout),
   MYSQL_SYSVAR(stats_on_metadata),
   MYSQL_SYSVAR(stats_sample_pages),
diff -ruN a/storage/innobase/include/trx0sys.h b/storage/innobase/include/trx0sys.h
--- a/storage/innobase/include/trx0sys.h	2010-11-03 07:01:13.000000000 +0900
+++ b/storage/innobase/include/trx0sys.h	2010-12-03 15:38:20.321953297 +0900
@@ -53,6 +53,9 @@
 extern ib_int64_t	trx_sys_mysql_master_log_pos;
 /* @} */
 
+extern char		trx_sys_mysql_relay_log_name[];
+extern ib_int64_t	trx_sys_mysql_relay_log_pos;
+
 /** If this MySQL server uses binary logging, after InnoDB has been inited
 and if it has done a crash recovery, we store the binlog file name and position
 here. */
@@ -287,7 +290,8 @@
 void
 trx_sys_update_mysql_binlog_offset(
 /*===============================*/
-	const char*	file_name,/*!< in: MySQL log file name */
+	trx_sysf_t*	sys_header,
+	const char*	file_name_in,/*!< in: MySQL log file name */
 	ib_int64_t	offset,	/*!< in: position in that log file */
 	ulint		field,	/*!< in: offset of the MySQL log info field in
 				the trx sys header */
@@ -482,6 +486,7 @@
 @see trx_sys_mysql_master_log_name
 @see trx_sys_mysql_bin_log_name */
 #define TRX_SYS_MYSQL_LOG_NAME_LEN	512
+#define TRX_SYS_MYSQL_MASTER_LOG_NAME_LEN	480	/* (500 - 12) is dead line. */
 /** Contents of TRX_SYS_MYSQL_LOG_MAGIC_N_FLD */
 #define TRX_SYS_MYSQL_LOG_MAGIC_N	873422344
 
@@ -491,6 +496,7 @@
 /** The offset of the MySQL replication info in the trx system header;
 this contains the same fields as TRX_SYS_MYSQL_LOG_INFO below */
 #define TRX_SYS_MYSQL_MASTER_LOG_INFO	(UNIV_PAGE_SIZE - 2000)
+#define TRX_SYS_MYSQL_RELAY_LOG_INFO	(UNIV_PAGE_SIZE - 1500)
 
 /** The offset of the MySQL binlog offset info in the trx system header */
 #define TRX_SYS_MYSQL_LOG_INFO		(UNIV_PAGE_SIZE - 1000)
diff -ruN a/storage/innobase/include/trx0trx.h b/storage/innobase/include/trx0trx.h
--- a/storage/innobase/include/trx0trx.h	2010-12-03 15:18:48.894955550 +0900
+++ b/storage/innobase/include/trx0trx.h	2010-12-03 15:38:20.323953416 +0900
@@ -580,6 +580,20 @@
 	ib_int64_t	mysql_log_offset;/* if MySQL binlog is used, this field
 					contains the end offset of the binlog
 					entry */
+	const char*	mysql_master_log_file_name;
+					/* if the database server is a MySQL
+					replication slave, we have here the
+					master binlog name up to which
+					replication has processed; otherwise
+					this is a pointer to a null
+					character */
+	ib_int64_t	mysql_master_log_pos;
+					/* if the database server is a MySQL
+					replication slave, this is the
+					position in the log file up to which
+					replication has processed */
+	const char*	mysql_relay_log_file_name;
+	ib_int64_t	mysql_relay_log_pos;
 	/*------------------------------*/
 	ulint		n_mysql_tables_in_use; /* number of Innobase tables
 					used in the processing of the current
diff -ruN a/storage/innobase/trx/trx0sys.c b/storage/innobase/trx/trx0sys.c
--- a/storage/innobase/trx/trx0sys.c	2010-11-03 07:01:13.000000000 +0900
+++ b/storage/innobase/trx/trx0sys.c	2010-12-03 15:38:20.325956917 +0900
@@ -76,13 +76,16 @@
 file name and position here. */
 /* @{ */
 /** Master binlog file name */
-UNIV_INTERN char	trx_sys_mysql_master_log_name[TRX_SYS_MYSQL_LOG_NAME_LEN];
+UNIV_INTERN char	trx_sys_mysql_master_log_name[TRX_SYS_MYSQL_MASTER_LOG_NAME_LEN];
 /** Master binlog file position.  We have successfully got the updates
 up to this position.  -1 means that no crash recovery was needed, or
 there was no master log position info inside InnoDB.*/
 UNIV_INTERN ib_int64_t	trx_sys_mysql_master_log_pos	= -1;
 /* @} */
 
+UNIV_INTERN char	trx_sys_mysql_relay_log_name[TRX_SYS_MYSQL_MASTER_LOG_NAME_LEN];
+UNIV_INTERN ib_int64_t	trx_sys_mysql_relay_log_pos	= -1;
+
 /** If this MySQL server uses binary logging, after InnoDB has been inited
 and if it has done a crash recovery, we store the binlog file name and position
 here. */
@@ -684,23 +687,25 @@
 void
 trx_sys_update_mysql_binlog_offset(
 /*===============================*/
-	const char*	file_name,/*!< in: MySQL log file name */
+	trx_sysf_t*	sys_header,
+	const char*	file_name_in,/*!< in: MySQL log file name */
 	ib_int64_t	offset,	/*!< in: position in that log file */
 	ulint		field,	/*!< in: offset of the MySQL log info field in
 				the trx sys header */
 	mtr_t*		mtr)	/*!< in: mtr */
 {
-	trx_sysf_t*	sys_header;
+	const char*	file_name;
 
-	if (ut_strlen(file_name) >= TRX_SYS_MYSQL_LOG_NAME_LEN) {
+	if (ut_strlen(file_name_in) >= TRX_SYS_MYSQL_MASTER_LOG_NAME_LEN) {
 
 		/* We cannot fit the name to the 512 bytes we have reserved */
+		/* -> To store relay log file information, file_name must fit to the 480 bytes */
 
-		return;
+		file_name = "";
+	} else {
+		file_name = file_name_in;
 	}
 
-	sys_header = trx_sysf_get(mtr);
-
 	if (mach_read_from_4(sys_header + field
 			     + TRX_SYS_MYSQL_LOG_MAGIC_N_FLD)
 	    != TRX_SYS_MYSQL_LOG_MAGIC_N) {
@@ -822,13 +827,26 @@
 					 + TRX_SYS_MYSQL_LOG_OFFSET_LOW),
 		sys_header + TRX_SYS_MYSQL_MASTER_LOG_INFO
 		+ TRX_SYS_MYSQL_LOG_NAME);
+
+	fprintf(stderr,
+		"InnoDB: and relay log file\n"
+		"InnoDB: position %lu %lu, file name %s\n",
+		(ulong) mach_read_from_4(sys_header
+					 + TRX_SYS_MYSQL_RELAY_LOG_INFO
+					 + TRX_SYS_MYSQL_LOG_OFFSET_HIGH),
+		(ulong) mach_read_from_4(sys_header
+					 + TRX_SYS_MYSQL_RELAY_LOG_INFO
+					 + TRX_SYS_MYSQL_LOG_OFFSET_LOW),
+		sys_header + TRX_SYS_MYSQL_RELAY_LOG_INFO
+		+ TRX_SYS_MYSQL_LOG_NAME);
+
 	/* Copy the master log position info to global variables we can
 	use in ha_innobase.cc to initialize glob_mi to right values */
 
 	ut_memcpy(trx_sys_mysql_master_log_name,
 		  sys_header + TRX_SYS_MYSQL_MASTER_LOG_INFO
 		  + TRX_SYS_MYSQL_LOG_NAME,
-		  TRX_SYS_MYSQL_LOG_NAME_LEN);
+		  TRX_SYS_MYSQL_MASTER_LOG_NAME_LEN);
 
 	trx_sys_mysql_master_log_pos
 		= (((ib_int64_t) mach_read_from_4(
@@ -837,6 +855,19 @@
 		+ ((ib_int64_t) mach_read_from_4(
 			   sys_header + TRX_SYS_MYSQL_MASTER_LOG_INFO
 			   + TRX_SYS_MYSQL_LOG_OFFSET_LOW));
+
+	ut_memcpy(trx_sys_mysql_relay_log_name,
+		  sys_header + TRX_SYS_MYSQL_RELAY_LOG_INFO
+		  + TRX_SYS_MYSQL_LOG_NAME,
+		  TRX_SYS_MYSQL_MASTER_LOG_NAME_LEN);
+
+	trx_sys_mysql_relay_log_pos
+		= (((ib_int64_t) mach_read_from_4(
+			    sys_header + TRX_SYS_MYSQL_RELAY_LOG_INFO
+			    + TRX_SYS_MYSQL_LOG_OFFSET_HIGH)) << 32)
+		+ ((ib_int64_t) mach_read_from_4(
+			   sys_header + TRX_SYS_MYSQL_RELAY_LOG_INFO
+			   + TRX_SYS_MYSQL_LOG_OFFSET_LOW));
 	mtr_commit(&mtr);
 }
 
diff -ruN a/storage/innobase/trx/trx0trx.c b/storage/innobase/trx/trx0trx.c
--- a/storage/innobase/trx/trx0trx.c	2010-12-03 15:37:45.549028990 +0900
+++ b/storage/innobase/trx/trx0trx.c	2010-12-03 15:38:20.328957217 +0900
@@ -138,6 +138,10 @@
 
 	trx->mysql_log_file_name = NULL;
 	trx->mysql_log_offset = 0;
+	trx->mysql_master_log_file_name = "";
+	trx->mysql_master_log_pos = 0;
+	trx->mysql_relay_log_file_name = "";
+	trx->mysql_relay_log_pos = 0;
 
 	mutex_create(trx_undo_mutex_key, &trx->undo_mutex, SYNC_TRX_UNDO);
 
@@ -820,6 +824,7 @@
 {
 	mtr_t		mtr;
 	trx_rseg_t*	rseg;
+	trx_sysf_t*	sys_header = NULL;
 
 	ut_ad(!mutex_own(&kernel_mutex));
 
@@ -873,8 +878,12 @@
 
 	if (trx->mysql_log_file_name
 	    && trx->mysql_log_file_name[0] != '\0') {
+		if (!sys_header) {
+			sys_header = trx_sysf_get(&mtr);
+		}
 
 		trx_sys_update_mysql_binlog_offset(
+			sys_header,
 			trx->mysql_log_file_name,
 			trx->mysql_log_offset,
 			TRX_SYS_MYSQL_LOG_INFO, &mtr);
@@ -882,6 +891,27 @@
 		trx->mysql_log_file_name = NULL;
 	}
 
+	if (trx->mysql_master_log_file_name[0] != '\0') {
+		/* This database server is a MySQL replication slave */
+		if (!sys_header) {
+			sys_header = trx_sysf_get(&mtr);
+		}
+
+		trx_sys_update_mysql_binlog_offset(
+			sys_header,
+			trx->mysql_relay_log_file_name,
+			trx->mysql_relay_log_pos,
+			TRX_SYS_MYSQL_RELAY_LOG_INFO, &mtr);
+
+		trx_sys_update_mysql_binlog_offset(
+			sys_header,
+			trx->mysql_master_log_file_name,
+			trx->mysql_master_log_pos,
+			TRX_SYS_MYSQL_MASTER_LOG_INFO, &mtr);
+
+		trx->mysql_master_log_file_name = "";
+	}
+
 	/* The following call commits the mini-transaction, making the
 	whole transaction committed in the file-based world, at this
 	log sequence number. The transaction becomes 'durable' when