MDEV-33465: an option to enable semisync recovery

The current semi-sync binlog fail-over recovery process uses
rpl_semi_sync_slave_enabled==TRUE as its condition to truncate a
primary server’s binlog, as it is anticipating the server to re-join
a replication topology as a replica. However, for servers configured
with both rpl_semi_sync_master_enabled=1 and
rpl_semi_sync_slave_enabled=1, if a primary is just re-started (i.e.
retaining its role as master), it can truncate its binlog to drop
transactions which its replica(s) has already received and executed.
If this happens, when the replica reconnects, its gtid_slave_pos can
be ahead of the recovered primary’s gtid_binlog_pos, resulting in an
error state where the replica’s state is ahead of the primary’s.

This patch changes the condition for semi-sync recovery to truncate
the binlog to instead use the configuration variable
--init-rpl-role, when set to SLAVE. This allows for both
rpl_semi_sync_master_enabled and rpl_semi_sync_slave_enabled to be
set for a primary that is restarted, and no transactions will be
lost, so long as --init-rpl-role is not set to SLAVE.

Reviewed By:
============
Sergei Golubchik <serg@mariadb.com>

mysql-test/suite/binlog/r/binlog_truncate_active_log.result

@@ -40,7 +40,7 @@ disconnect master1;
 disconnect master2;
 disconnect master3;
 disconnect master4;
-# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1 --log-warnings=3
+# restart: --init-rpl-role=SLAVE --sync-binlog=1 --log-warnings=3
 FOUND 1 /Successfully truncated.*to remove transactions starting from GTID 0-1-7/ in mysqld.1.err
 Pre-crash binlog file content:
 include/show_binlog_events.inc
@@ -104,7 +104,7 @@ disconnect master1;
 disconnect master2;
 disconnect master3;
 disconnect master4;
-# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1 --log-warnings=3
+# restart: --init-rpl-role=SLAVE --sync-binlog=1 --log-warnings=3
 FOUND 1 /Successfully truncated.*to remove transactions starting from GTID 0-1-11/ in mysqld.1.err
 Pre-crash binlog file content:
 include/show_binlog_events.inc
@@ -173,7 +173,7 @@ disconnect master1;
 disconnect master2;
 disconnect master3;
 disconnect master4;
-# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1 --log-warnings=3
+# restart: --init-rpl-role=SLAVE --sync-binlog=1 --log-warnings=3
 FOUND 1 /Successfully truncated.*to remove transactions starting from GTID 0-1-15/ in mysqld.1.err
 Pre-crash binlog file content:
 include/show_binlog_events.inc
@@ -248,7 +248,7 @@ disconnect master1;
 disconnect master2;
 disconnect master3;
 disconnect master4;
-# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1 --log-warnings=3
+# restart: --init-rpl-role=SLAVE --sync-binlog=1 --log-warnings=3
 FOUND 1 /Successfully truncated.*to remove transactions starting from GTID 0-1-21/ in mysqld.1.err
 Pre-crash binlog file content:
 include/show_binlog_events.inc

mysql-test/suite/binlog/r/binlog_truncate_multi_engine.result

@@ -31,9 +31,9 @@ Log_name	File_size
 master-bin.000001	#
 master-bin.000002	#
 master-bin.000003	#
-# restart the server with --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
+# restart the server with --init-rpl-role=SLAVE --sync-binlog=1
 # the server is restarted
-# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
+# restart: --init-rpl-role=SLAVE --sync-binlog=1
 connection default;
 #
 # *** Summary: 1 row should be present in both tables; binlog is truncated; number of binlogs at reconnect - 3:
@@ -98,7 +98,7 @@ INSERT INTO t2 VALUES (2, REPEAT("x", 4100));
 INSERT INTO t1 VALUES (2, REPEAT("x", 4100));
 COMMIT;
 connection default;
-# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
+# restart: --init-rpl-role=SLAVE --sync-binlog=1
 connection default;
 #
 # *** Summary: 2 rows should be present in both tables; no binlog truncation; one extra binlog file compare with A; number of binlogs at reconnect - 4:
@@ -155,9 +155,9 @@ Log_name	File_size
 master-bin.000001	#
 master-bin.000002	#
 master-bin.000003	#
-# restart the server with --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
+# restart the server with --init-rpl-role=SLAVE --sync-binlog=1
 # the server is restarted
-# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
+# restart: --init-rpl-role=SLAVE --sync-binlog=1
 connection default;
 #
 # *** Summary: 2 rows should be present in both tables; no binlog truncation; the same # of binlog files as in B; number of binlogs at reconnect - 4:

mysql-test/suite/binlog/r/binlog_truncate_multi_log.result

@@ -42,7 +42,7 @@ connection default;
 disconnect master1;
 disconnect master2;
 disconnect master3;
-# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1 --log-warnings=3
+# restart: --init-rpl-role=SLAVE --sync-binlog=1 --log-warnings=3
 FOUND 1 /truncated binlog file:.*master.*000002/ in mysqld.1.err
 "One record should be present in table"
 SELECT * FROM ti;

mysql-test/suite/binlog/t/binlog_truncate_active_log.inc

@@ -42,7 +42,7 @@ SELECT @@global.gtid_binlog_pos as 'Before the crash and never logged trx';
 #
 # Server restart
 #
---let $restart_parameters= --rpl-semi-sync-slave-enabled=1 --sync-binlog=1 --log-warnings=3
+--let $restart_parameters= --init-rpl-role=SLAVE --sync-binlog=1 --log-warnings=3
 --source include/start_mysqld.inc
 
 # Check error log for a successful truncate message.

mysql-test/suite/binlog/t/binlog_truncate_active_log.test

@@ -36,7 +36,7 @@ CREATE TABLE tm (f INT) ENGINE=Aria;
 # Using 'debug_sync' hold 'query1' execution after 'query1' is flushed and
 # synced to binary log but not yet committed. In an another connection hold
 # 'query2' execution after 'query2' is flushed and synced to binlog.
-# Crash and restart server with --rpl-semi-sync-slave-enabled=1
+# Crash and restart server with --init-rpl-role=SLAVE
 #
 # During recovery of binary log 'query1' status is checked with InnoDB engine,
 # it will be in prepared but not yet commited. All transactions starting from

mysql-test/suite/binlog/t/binlog_truncate_multi_engine.test

@@ -28,7 +28,7 @@ CREATE TABLE t2 (a INT PRIMARY KEY, b MEDIUMTEXT) ENGINE=rocksdb;
 # The transaction is killed along with the server after that.
 --let $shutdown_timeout=0
 --let $debug_sync_action = "commit_after_release_LOCK_log SIGNAL con1_ready WAIT_FOR signal_no_signal"
---let $restart_parameters = --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
+--let $restart_parameters = --init-rpl-role=SLAVE --sync-binlog=1
 --let $test_outcome= 1 row should be present in both tables; binlog is truncated; number of binlogs at reconnect - 3
 --source binlog_truncate_multi_engine.inc
 --echo Proof of the truncated binlog file is readable (two transactions must be seen):
@@ -41,7 +41,7 @@ CREATE TABLE t2 (a INT PRIMARY KEY, b MEDIUMTEXT) ENGINE=rocksdb;
 --let $debug_sync_action = ""
 # Both debug_sync and debug-dbug are required to make sure Engines remember the commit state
 # debug_sync alone will not help.
---let $restart_parameters = --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
+--let $restart_parameters = --init-rpl-role=SLAVE --sync-binlog=1
 --let $test_outcome= 2 rows should be present in both tables; no binlog truncation; one extra binlog file compare with A; number of binlogs at reconnect - 4
 --source binlog_truncate_multi_engine.inc
 
@@ -50,7 +50,7 @@ CREATE TABLE t2 (a INT PRIMARY KEY, b MEDIUMTEXT) ENGINE=rocksdb;
 --let $debug_sync_action = "commit_after_run_commit_ordered SIGNAL con1_ready"
 # Hold off after both engines have committed. The server is shut down.
 --let $shutdown_timeout=
---let $restart_parameters = --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
+--let $restart_parameters = --init-rpl-role=SLAVE --sync-binlog=1
 --let $test_outcome= 2 rows should be present in both tables; no binlog truncation; the same # of binlog files as in B; number of binlogs at reconnect - 4
 --source binlog_truncate_multi_engine.inc
 

mysql-test/suite/binlog/t/binlog_truncate_multi_log.test

@@ -63,7 +63,7 @@ SELECT @@global.gtid_binlog_state;
 #
 # Server restart
 #
---let $restart_parameters= --rpl-semi-sync-slave-enabled=1 --sync-binlog=1 --log-warnings=3
+--let $restart_parameters= --init-rpl-role=SLAVE --sync-binlog=1 --log-warnings=3
 --source include/start_mysqld.inc
 
 # Check error log for a successful truncate message.

mysql-test/suite/binlog/t/binlog_truncate_multi_log_unsafe.test

@@ -92,7 +92,7 @@ SELECT @@global.gtid_binlog_state;
 #
 --echo # Failed restart as the semisync slave
 --error 1
---exec $MYSQLD_LAST_CMD  --rpl-semi-sync-slave-enabled=1 >> $MYSQLTEST_VARDIR/log/mysqld.1.err 2>&1
+--exec $MYSQLD_LAST_CMD  --init-rpl-role=SLAVE >> $MYSQLTEST_VARDIR/log/mysqld.1.err 2>&1
 
 --echo # Normal restart
 --source include/start_mysqld.inc

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

@@ -50,7 +50,7 @@ on slave must be 2
 SELECT @@GLOBAL.gtid_current_pos;
 @@GLOBAL.gtid_current_pos
 0-1-4
-# restart: --skip-slave-start=1 --rpl-semi-sync-slave-enabled=1
+# restart: --skip-slave-start=1 --rpl-semi-sync-slave-enabled=1 --init-rpl-role=SLAVE
 connection server_1;
 # Ensuring variable rpl_semi_sync_slave_enabled is ON..
 # Ensuring status rpl_semi_sync_slave_status is OFF..
@@ -136,7 +136,7 @@ on slave must be 5
 SELECT @@GLOBAL.gtid_current_pos;
 @@GLOBAL.gtid_current_pos
 0-2-7
-# restart: --skip-slave-start=1 --rpl-semi-sync-slave-enabled=1
+# restart: --skip-slave-start=1 --rpl-semi-sync-slave-enabled=1 --init-rpl-role=SLAVE
 connection server_2;
 # Ensuring variable rpl_semi_sync_slave_enabled is ON..
 # Ensuring status rpl_semi_sync_slave_status is OFF..
@@ -221,7 +221,7 @@ on slave must be 7
 SELECT @@GLOBAL.gtid_current_pos;
 @@GLOBAL.gtid_current_pos
 0-1-9
-# restart: --skip-slave-start=1 --rpl-semi-sync-slave-enabled=1
+# restart: --skip-slave-start=1 --rpl-semi-sync-slave-enabled=1 --init-rpl-role=SLAVE
 connection server_1;
 # Ensuring variable rpl_semi_sync_slave_enabled is ON..
 # Ensuring status rpl_semi_sync_slave_status is OFF..

mysql-test/suite/rpl/t/rpl_semi_sync_crash.inc

@@ -74,7 +74,7 @@ source include/wait_for_slave_param.inc;
 
 SELECT @@GLOBAL.gtid_current_pos;
 
---let $restart_parameters=--skip-slave-start=1 --rpl-semi-sync-slave-enabled=1
+--let $restart_parameters=--skip-slave-start=1 --rpl-semi-sync-slave-enabled=1 --init-rpl-role=SLAVE
 --let $allow_rpl_inited=1
 --source include/start_mysqld.inc
 --connection server_$server_to_crash

sql/log.cc

@@ -40,6 +40,7 @@
 #include "sql_audit.h"
 #include "mysqld.h"
 #include "ddl_log.h"
+#include "repl_failsafe.h"
 
 #include <my_dir.h>
 #include <m_ctype.h>				// For test_if_number
@@ -11025,7 +11026,7 @@ Recovery_context::Recovery_context() :
   prev_event_pos(0),
   last_gtid_standalone(false), last_gtid_valid(false), last_gtid_no2pc(false),
   last_gtid_engines(0),
-  do_truncate(global_rpl_semi_sync_slave_enabled),
+  do_truncate(rpl_status == RPL_IDLE_SLAVE),
   truncate_validated(false), truncate_reset_done(false),
   truncate_set_in_1st(false), id_binlog(MAX_binlog_id),
   checksum_alg(BINLOG_CHECKSUM_ALG_UNDEF), gtid_maybe_to_truncate(NULL)

sql/mysqld.cc

@@ -6627,7 +6627,7 @@ struct my_option my_long_options[]=
 #ifdef HAVE_REPLICATION
   {"init-rpl-role", 0, "Set the replication role",
    &rpl_status, &rpl_status, &rpl_role_typelib,
-   GET_ENUM, REQUIRED_ARG, 0, 0, 0, 0, 0, 0},
+   GET_ENUM, REQUIRED_ARG, RPL_AUTH_MASTER, 0, 0, 0, 0, 0},
 #endif /* HAVE_REPLICATION */
   {"memlock", 0, "Lock mysqld in memory.", &locked_in_memory,
    &locked_in_memory, 0, GET_BOOL, NO_ARG, 0, 0, 0, 0, 0, 0},