WL#7817- RPL Monitoring: Move status variables to

         replication P_S tables                                               
                                                                                
Added more fields to replication P_S tables.
Following variables that were a part of
SHOW STATUS have now been added:                                                 
                                                                                                                    
Show status like 'Slave_last_heartbeat';                                  
Show status like 'Slave_received_heartbeats';                                   
show status like 'Slave_heartbeat_period';
Show status like 'Slave_retried_transactions';

mysql-test/extra/rpl_tests/rpl_deadlock.test

@@ -61,6 +61,15 @@ let $status_var_comparsion= >;
 --source include/wait_for_status_var.inc
 SELECT COUNT(*) FROM t2;
 COMMIT;
+
+--echo
+--echo # Test that the performance schema coulumn shows > 0 values.
+--echo
+
+--let $assert_text= current number of retries should be more than the value saved before deadlock.
+--let $assert_cond= [SELECT COUNT_TRANSACTIONS_RETRIES FROM performance_schema.replication_execute_status, COUNT_TRANSACTIONS_RETRIES, 1] > "$slave_retried_transactions"
+--source include/assert.inc
+
 source include/check_slave_is_running.inc;
 sync_with_master;
 

mysql-test/suite/perfschema/r/table_schema.result

@@ -861,16 +861,20 @@ def	performance_schema	replication_connection_configuration	SSL_CRL_FILE	13	NULL
 def	performance_schema	replication_connection_configuration	SSL_CRL_PATH	14	NULL	NO	varchar	255	765	NULL	NULL	NULL	utf8	utf8_general_ci	varchar(255)			select,insert,update,references	
 def	performance_schema	replication_connection_configuration	CONNECTION_RETRY_INTERVAL	15	NULL	NO	int	NULL	NULL	10	0	NULL	NULL	NULL	int(11)			select,insert,update,references	
 def	performance_schema	replication_connection_configuration	CONNECTION_RETRY_COUNT	16	NULL	NO	bigint	NULL	NULL	20	0	NULL	NULL	NULL	bigint(20) unsigned			select,insert,update,references	
+def	performance_schema	replication_connection_configuration	HEARTBEAT_INTERVAL	17	NULL	NO	double	NULL	NULL	10	3	NULL	NULL	NULL	double(10,3) unsigned			select,insert,update,references	Number of seconds after which a heartbeat will be sent .
 def	performance_schema	replication_connection_status	SOURCE_UUID	1	NULL	NO	char	36	108	NULL	NULL	NULL	utf8	utf8_bin	char(36)			select,insert,update,references	
 def	performance_schema	replication_connection_status	THREAD_ID	2	NULL	YES	bigint	NULL	NULL	20	0	NULL	NULL	NULL	bigint(20) unsigned			select,insert,update,references	
 def	performance_schema	replication_connection_status	SERVICE_STATE	3	NULL	NO	enum	10	30	NULL	NULL	NULL	utf8	utf8_general_ci	enum('ON','OFF','CONNECTING')			select,insert,update,references	
-def	performance_schema	replication_connection_status	RECEIVED_TRANSACTION_SET	4	NULL	NO	text	65535	65535	NULL	NULL	NULL	utf8	utf8_general_ci	text			select,insert,update,references	
-def	performance_schema	replication_connection_status	LAST_ERROR_NUMBER	5	NULL	NO	int	NULL	NULL	10	0	NULL	NULL	NULL	int(11)			select,insert,update,references	
-def	performance_schema	replication_connection_status	LAST_ERROR_MESSAGE	6	NULL	NO	varchar	1024	3072	NULL	NULL	NULL	utf8	utf8_general_ci	varchar(1024)			select,insert,update,references	
-def	performance_schema	replication_connection_status	LAST_ERROR_TIMESTAMP	7	CURRENT_TIMESTAMP	NO	timestamp	NULL	NULL	NULL	NULL	0	NULL	NULL	timestamp		on update CURRENT_TIMESTAMP	select,insert,update,references	
+def	performance_schema	replication_connection_status	COUNT_RECEIVED_HEARTBEATS	4	0	NO	bigint	NULL	NULL	20	0	NULL	NULL	NULL	bigint(20) unsigned			select,insert,update,references	
+def	performance_schema	replication_connection_status	LAST_HEARTBEAT_TIMESTAMP	5	CURRENT_TIMESTAMP	NO	timestamp	NULL	NULL	NULL	NULL	0	NULL	NULL	timestamp		on update CURRENT_TIMESTAMP	select,insert,update,references	Shows when the most recent heartbeat signal was received.
+def	performance_schema	replication_connection_status	RECEIVED_TRANSACTION_SET	6	NULL	NO	text	65535	65535	NULL	NULL	NULL	utf8	utf8_general_ci	text			select,insert,update,references	
+def	performance_schema	replication_connection_status	LAST_ERROR_NUMBER	7	NULL	NO	int	NULL	NULL	10	0	NULL	NULL	NULL	int(11)			select,insert,update,references	
+def	performance_schema	replication_connection_status	LAST_ERROR_MESSAGE	8	NULL	NO	varchar	1024	3072	NULL	NULL	NULL	utf8	utf8_general_ci	varchar(1024)			select,insert,update,references	
+def	performance_schema	replication_connection_status	LAST_ERROR_TIMESTAMP	9	0000-00-00 00:00:00	NO	timestamp	NULL	NULL	NULL	NULL	0	NULL	NULL	timestamp			select,insert,update,references	
 def	performance_schema	replication_execute_configuration	DESIRED_DELAY	1	NULL	NO	int	NULL	NULL	10	0	NULL	NULL	NULL	int(11)			select,insert,update,references	
 def	performance_schema	replication_execute_status	SERVICE_STATE	1	NULL	NO	enum	3	9	NULL	NULL	NULL	utf8	utf8_general_ci	enum('ON','OFF')			select,insert,update,references	
 def	performance_schema	replication_execute_status	REMAINING_DELAY	2	NULL	YES	int	NULL	NULL	10	0	NULL	NULL	NULL	int(10) unsigned			select,insert,update,references	
+def	performance_schema	replication_execute_status	COUNT_TRANSACTIONS_RETRIES	3	NULL	NO	bigint	NULL	NULL	20	0	NULL	NULL	NULL	bigint(20) unsigned			select,insert,update,references	
 def	performance_schema	replication_execute_status_by_coordinator	THREAD_ID	1	NULL	YES	bigint	NULL	NULL	20	0	NULL	NULL	NULL	bigint(20) unsigned			select,insert,update,references	
 def	performance_schema	replication_execute_status_by_coordinator	SERVICE_STATE	2	NULL	NO	enum	3	9	NULL	NULL	NULL	utf8	utf8_general_ci	enum('ON','OFF')			select,insert,update,references	
 def	performance_schema	replication_execute_status_by_coordinator	LAST_ERROR_NUMBER	3	NULL	NO	int	NULL	NULL	10	0	NULL	NULL	NULL	int(11)			select,insert,update,references	

mysql-test/suite/rpl/r/rpl_deadlock_innodb.result

@@ -46,6 +46,10 @@ SELECT COUNT(*) FROM t2;
 COUNT(*)
 0
 COMMIT;
+
+# Test that the performance schema coulumn shows > 0 values.
+
+include/assert.inc [current number of retries should be more than the value saved before deadlock.]
 include/check_slave_is_running.inc
 SELECT * FROM t1;
 a

mysql-test/suite/rpl/r/rpl_perfschema_connect_config.result

@@ -27,6 +27,11 @@ include/assert.inc [Value returned by SSS and PS table for SSL_Crl_Path should b
 include/assert.inc [Value returned by SSS and PS table for Connection_Retry_Interval should be same.]
 include/assert.inc [Value returned by SSS and PS table for Connection_Retry_Count should be same.]
 
+# Heartbeat_Interval is part of I_S and P_S. We will compare the
+# two to make sure both match.
+
+include/assert.inc [Value returned by IS and PS table for Heartbeat_Interval should be same.]
+
 # Change configuration parameters and verify that these changes
 # are shown correctly by SELECTs from PS table.
 

mysql-test/suite/rpl/r/rpl_perfschema_connect_status.result

@@ -12,10 +12,27 @@ include/assert.inc [On master, the table should return an empty set.]
 include/assert.inc [Value returned by SSS and PS table for Source_UUID should be same.]
 include/assert.inc [thread_name should should indicate io thread.]
 include/assert.inc [SSS shows Slave_IO_Running as "Yes". So, Service_State from this PS table should be "ON".]
+include/assert.inc [IS and PS tables should both show that no heartbeats have been received.]
+include/assert.inc [IS table should give an empty output and PS table should output 0000-00-00 00:00:00 both indicating no heartbeats have been received]
 include/assert.inc [Value returned by SSS and PS table for Received_Transaction_Set should be same.]
 include/assert.inc [Value returned by SSS and PS table for Last_Error_Number should be same.]
 include/assert.inc [Value returned by SSS and PS table for Last_Error_Message should be same.]
 include/assert.inc [Value returned by PS table for Last_Error_Timestamp should be 0000-00-00 00:00:00.]
+include/stop_slave.inc
+
+# heartbeat count and last heartbeat timestamp are indeterministic so we
+# can not test for their exact values. We will thus check for >0 number
+# of heartbeats and last heartbeat timestamp!= zeros.
+
+change master to master_heartbeat_period= 0.5;
+set @restore_slave_net_timeout= @@global.slave_net_timeout;
+set @@global.slave_net_timeout= 10;
+include/start_slave.inc
+
+A heartbeat has been received by the slave
+
+include/assert.inc [last_heartbeat_timestamp should NOT be zero]
+set @@global.slave_net_timeout= @restore_slave_net_timeout;
 
 # We now introduce an error in the IO thread and check for the correctness
 # of error number, message and timestamp fields.

mysql-test/suite/rpl/r/rpl_perfschema_execute_status.result

@@ -12,10 +12,12 @@ include/assert.inc [On master, the table should return an empty set.]
 # Verify that output is same as SSS on a fresh slave.
 
 include/assert.inc [SSS shows Slave_SQL_Running as "Yes". So, Service_State from this PS table should be "ON".]
+include/assert.inc [COUNT_TRANSACTION_RETRIES should be 0.]
 
-# Verify that the value of this field is correct after STOP SLAVE.
+# Verify that the fields show the correct values after STOP SLAVE.
 
 include/stop_slave.inc
 include/assert.inc [SSS shows Slave_SQL_Running as "No". So, Service_State from this PS table should be "OFF".]
+include/assert.inc [COUNT_TRANSACTION_RETRIES should be 0.]
 include/start_slave.inc
 include/rpl_end.inc

mysql-test/suite/rpl/t/rpl_perfschema_connect_config.test

@@ -30,6 +30,8 @@
 #  WL#3656: PERFORMANCE SCHEMA table for SHOW SLAVE STATUS
 #  Bug#18174719: P_S.REPLICATION_CONNECTION_CONFIGURATION'S COL
 #                SSL_CRL_PATH THROWS WARNING 1366
+#  WL#7817: RPL Monitoring: Move status variables to replication P_S tables
+#           for multi-source monitoring
 
 source include/master-slave.inc;
 source include/have_ssl_communication.inc;
@@ -148,6 +150,17 @@ let $assert_text= Value returned by SSS and PS table for Connection_Retry_Count
 let $assert_cond= "$sss_value" = "$ps_value";
 source include/assert.inc;
 
+--echo
+--echo # Heartbeat_Interval is part of I_S and P_S. We will compare the
+--echo # two to make sure both match.
+--echo
+
+let $is_value= query_get_value(select Variable_Value from information_schema.GLOBAL_STATUS where variable_name= 'Slave_heartbeat_period', Variable_Value, 1);
+let $ps_value= query_get_value(select Heartbeat_Interval from performance_schema.replication_connection_configuration, Heartbeat_Interval, 1);
+let $assert_text= Value returned by IS and PS table for Heartbeat_Interval should be same.;
+let $assert_cond= $is_value = $ps_value;
+source include/assert.inc;
+
 --echo
 --echo # Change configuration parameters and verify that these changes
 --echo # are shown correctly by SELECTs from PS table.

mysql-test/suite/rpl/t/rpl_perfschema_connect_status.test

@@ -64,6 +64,25 @@ let $assert_text= SSS shows Slave_IO_Running as "Yes". So, Service_State from th
 let $assert_cond= "$sss_value" = "Yes" AND "$ps_value"= "ON";
 source include/assert.inc;
 
+#
+# Heartbeat count and last hearbeat timestamp are not in SSS, but these are in
+# IS. So to test count_received_heartbeats and last_heartbeat_timestamp
+# we compare values from IS and PS unlike SSS and PS for other fields in this
+# test.
+#
+
+let $is_value= query_get_value(select Variable_Value from information_schema.global_status where Variable_name = 'slave_received_heartbeats', Variable_Value, 1);
+let $ps_value= query_get_value(select count_received_heartbeats from performance_schema.replication_connection_status, count_received_heartbeats, 1);
+let $assert_text= IS and PS tables should both show that no heartbeats have been received.;
+let $assert_cond= "$is_value" = "$ps_value";
+source include/assert.inc;
+
+let $is_value= query_get_value(select Variable_Value from information_schema.global_status where Variable_name = 'slave_last_heartbeat', Variable_Value, 1);
+let $ps_value= query_get_value(select last_heartbeat_timestamp from performance_schema.replication_connection_status, last_heartbeat_timestamp, 1);
+let $assert_text= IS table should give an empty output and PS table should output 0000-00-00 00:00:00 both indicating no heartbeats have been received;
+let $assert_cond= "$is_value" = "" AND "$ps_value"= "0000-00-00 00:00:00";
+source include/assert.inc;
+
 let $sss_value= query_get_value(SHOW SLAVE STATUS, Retrieved_Gtid_Set, 1);
 let $ps_value= query_get_value(select Received_Transaction_Set from performance_schema.replication_connection_status, Received_Transaction_Set, 1);
 let $assert_text= Value returned by SSS and PS table for Received_Transaction_Set should be same.;
@@ -88,6 +107,67 @@ let $assert_text= Value returned by PS table for Last_Error_Timestamp should be
 let $assert_cond= "$ps_value" = "0000-00-00 00:00:00";
 source include/assert.inc;
 
+source include/stop_slave.inc;
+
+--echo
+--echo # heartbeat count and last heartbeat timestamp are indeterministic so we
+--echo # can not test for their exact values. We will thus check for >0 number
+--echo # of heartbeats and last heartbeat timestamp!= zeros.
+--echo
+
+change master to master_heartbeat_period= 0.5;
+
+set @restore_slave_net_timeout= @@global.slave_net_timeout;
+set @@global.slave_net_timeout= 10;
+
+source include/start_slave.inc;
+
+# there is an explicit sleep lasting longer than slave_net_timeout
+# to ensure that nothing will come to slave from master for that period.
+# That would cause reconnecting and relaylog rotation w/o the fix i.e
+# without a heartbeat received.
+
+real_sleep 15;
+
+#
+# proof that there has been received at least one heartbeat;
+# The exact number of received heartbeat is an indeterministic value
+# and therefore it's not recorded into results.
+#
+
+let $slave_wait_param_counter= 300;
+let $slave_value= query_get_value(select count_received_heartbeats from performance_schema.replication_connection_status, count_received_heartbeats, 1);
+
+# Checking the fact that at least one heartbeat is received
+while (!$slave_value)
+{
+  dec $slave_wait_param_counter;
+  if (!$slave_wait_param_counter)
+  {
+    query_vertical show slave status;
+    select count_received_heartbeats from performance_schema.replication_connection_status;
+    --die ERROR: failed while waiting for slave to generate a heartbeat.
+    exit;
+  }
+  sleep 0.1;
+  let $slave_value= query_get_value(select count_received_heartbeats from performance_schema.replication_connection_status, count_received_heartbeats, 1);
+}
+
+--echo
+--echo A heartbeat has been received by the slave
+--echo
+
+#
+# check for non-zero value of the Slave_last_heartbeat variable when the slave is enabled
+#
+let $ps_value= query_get_value(select last_heartbeat_timestamp from performance_schema.replication_connection_status, last_heartbeat_timestamp, 1);
+let $assert_text= last_heartbeat_timestamp should NOT be zero;
+let $assert_cond= "$ps_value" != "0000-00-00 00:00:00";
+source include/assert.inc;
+
+# Reset slave_net_timeout to default after testing heartbeats.
+set @@global.slave_net_timeout= @restore_slave_net_timeout;
+
 --echo
 --echo # We now introduce an error in the IO thread and check for the correctness
 --echo # of error number, message and timestamp fields.

mysql-test/suite/rpl/t/rpl_perfschema_execute_status.test

@@ -11,11 +11,14 @@
 #  - Verify that output is same as SSS on a fresh slave.
 #  - Verify that the value of this field is correct after STOP SLAVE.
 #  - Remaining delay is not tested.
+#  - Count_trnsaction is partially tested here making sure it can be queried.
+#    More testing in extra/rpl_deadlock.test
 #
 #  ==== Related Worklog ====
 #
 #  WL#3656: PERFORMANCE SCHEMA table for SHOW SLAVE STATUS
-
+#  WL#7817: RPL Monitoring: Move status variables to replication P_S tables for
+#           multi-source monitoring
 
 source include/master-slave.inc;
 source include/have_binlog_format_mixed.inc;
@@ -41,8 +44,13 @@ let $assert_text= SSS shows Slave_SQL_Running as "Yes". So, Service_State from t
 let $assert_cond= "$sss_value" = "Yes" AND "$ps_value"= "ON";
 source include/assert.inc;
 
+let $ps_value= query_get_value(select count_transactions_retries from performance_schema.replication_execute_status, count_transactions_retries, 1);
+let $assert_text= COUNT_TRANSACTION_RETRIES should be 0.;
+let $assert_cond= "$ps_value"= 0;
+source include/assert.inc;
+
 --echo
---echo # Verify that the value of this field is correct after STOP SLAVE.
+--echo # Verify that the fields show the correct values after STOP SLAVE.
 --echo
 
 source include/stop_slave.inc;
@@ -53,6 +61,11 @@ let $assert_text= SSS shows Slave_SQL_Running as "No". So, Service_State from th
 let $assert_cond= "$sss_value" = "No" AND "$ps_value"= "OFF";
 source include/assert.inc;
 
+let $ps_value= query_get_value(select count_transactions_retries from performance_schema.replication_execute_status, count_transactions_retries, 1);
+let $assert_text= COUNT_TRANSACTION_RETRIES should be 0.;
+let $assert_cond= "$ps_value"= 0;
+source include/assert.inc;
+
 source include/start_slave.inc;
 source include/rpl_end.inc;
 

scripts/mysql_system_tables.sql

@@ -2144,7 +2144,8 @@ SET @cmd="CREATE TABLE performance_schema.replication_connection_configuration("
   "SSL_CRL_FILE VARCHAR(255) not null,"
   "SSL_CRL_PATH VARCHAR(255) not null,"
   "CONNECTION_RETRY_INTERVAL INTEGER not null,"
-  "CONNECTION_RETRY_COUNT BIGINT unsigned NOT NULL"
+  "CONNECTION_RETRY_COUNT BIGINT unsigned not null,"
+  "HEARTBEAT_INTERVAL DOUBLE(10,3) unsigned not null COMMENT 'Number of seconds after which a heartbeat will be sent .'"
   ") ENGINE=PERFORMANCE_SCHEMA;";
 
 SET @str = IF(@have_pfs = 1, @cmd, 'SET @dummy = 0');
@@ -2160,6 +2161,8 @@ SET @cmd="CREATE TABLE performance_schema.replication_connection_status("
   "SOURCE_UUID CHAR(36) collate utf8_bin not null,"
   "THREAD_ID BIGINT unsigned,"
   "SERVICE_STATE ENUM('ON','OFF','CONNECTING') not null,"
+  "COUNT_RECEIVED_HEARTBEATS bigint unsigned NOT NULL DEFAULT 0,"
+  "LAST_HEARTBEAT_TIMESTAMP TIMESTAMP(0) not null COMMENT 'Shows when the most recent heartbeat signal was received.',"
   "RECEIVED_TRANSACTION_SET TEXT not null,"
   "LAST_ERROR_NUMBER INTEGER not null,"
   "LAST_ERROR_MESSAGE VARCHAR(1024) not null,"
@@ -2190,7 +2193,8 @@ DROP PREPARE stmt;
 
 SET @cmd="CREATE TABLE performance_schema.replication_execute_status("
   "SERVICE_STATE ENUM('ON','OFF') not null,"
-  "REMAINING_DELAY INTEGER unsigned"
+  "REMAINING_DELAY INTEGER unsigned,"
+  "COUNT_TRANSACTIONS_RETRIES BIGINT unsigned not null"
   ") ENGINE=PERFORMANCE_SCHEMA;";
 
 SET @str = IF(@have_pfs = 1, @cmd, 'SET @dummy = 0');

storage/perfschema/pfs_engine_table.cc

@@ -602,6 +602,13 @@ void PFS_engine_table::set_field_timestamp(Field *f, ulonglong value)
   f2->store_timestamp(& tm);
 }
 
+void PFS_engine_table::set_field_double(Field *f, double value)
+{
+  DBUG_ASSERT(f->real_type() == MYSQL_TYPE_DOUBLE);
+  Field_double *f2= (Field_double*) f;
+  f2->store(value);
+}
+
 ulonglong PFS_engine_table::get_field_enum(Field *f)
 {
   DBUG_ASSERT(f->real_type() == MYSQL_TYPE_ENUM);

storage/perfschema/pfs_engine_table.h

@@ -144,6 +144,12 @@ class PFS_engine_table
     @param value the value to assign
   */
   static void set_field_timestamp(Field *f, ulonglong value);
+  /**
+    Helper, assign a value to a double field.
+    @param f the field to set
+    @param value the value to assign
+  */
+  static void set_field_double(Field *f, double value);
   /**
     Helper, read a value from an enum field.
     @param f the field to read

storage/perfschema/table_replication_connection_configuration.cc

@@ -116,12 +116,17 @@ static const TABLE_FIELD_TYPE field_types[]=
     {C_STRING_WITH_LEN("CONNECTION_RETRY_COUNT")},
     {C_STRING_WITH_LEN("bigint")},
     {NULL, 0}
-  }
+  },
+  {
+    {C_STRING_WITH_LEN("HEARTBEAT_INTERVAL")},
+    {C_STRING_WITH_LEN("double(10,3)")},
+    {NULL, 0}
+   }
 };
 
 TABLE_FIELD_DEF
 table_replication_connection_configuration::m_field_def=
-{ 16, field_types };
+{ 17, field_types };
 
 PFS_engine_table_share
 table_replication_connection_configuration::m_share=
@@ -279,6 +284,8 @@ void table_replication_connection_configuration::make_row()
 
   m_row.connection_retry_count= (ulong) active_mi->retry_count;
 
+  m_row.heartbeat_interval= (double)active_mi->heartbeat_period;
+
   mysql_mutex_unlock(&active_mi->rli->data_lock);
   mysql_mutex_unlock(&active_mi->data_lock);
   mysql_mutex_unlock(&LOCK_active_mi);
@@ -358,6 +365,9 @@ int table_replication_connection_configuration::read_row_values(TABLE *table,
       case 15: /** connect_retry_count */
         set_field_ulonglong(f, m_row.connection_retry_count);
         break;
+      case 16:/** number of seconds after which heartbeat will be sent */
+        set_field_double(f, m_row.heartbeat_interval);
+        break;
       default:
         DBUG_ASSERT(false);
       }

storage/perfschema/table_replication_connection_configuration.h

@@ -1,5 +1,5 @@
 /*
-   Copyright (c) 2013, Oracle and/or its affiliates. All rights reserved.
+   Copyright (c) 2013, 2014 Oracle and/or its affiliates. All rights reserved.
 
    This program is free software; you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -79,6 +79,7 @@ struct st_row_connect_config {
   uint ssl_crl_path_length;
   uint connection_retry_interval;
   ulong connection_retry_count;
+  double heartbeat_interval;
 };
 
 /** Table PERFORMANCE_SCHEMA.TABLE_REPLICATION_CONNECTION_CONFIGURATION. */