MDEV-5262: Missing retry after temp error in parallel replication

Handle retry of event groups that span multiple relay log files.

 - If retry reaches the end of one relay log file, move on to the next.

 - Handle refcounting of relay log files, and avoid purging relay log
   files until all event groups have completed that might have needed
   them for transaction retry.

2 files changed, 107 insertions, 2 deletions

diff --git a/mysql-test/suite/rpl/r/rpl_parallel_retry.result b/mysql-test/suite/rpl/r/rpl_parallel_retry.result
index 1badf084440..cd12d92430b 100644
--- a/mysql-test/suite/rpl/r/rpl_parallel_retry.result
+++ b/mysql-test/suite/rpl/r/rpl_parallel_retry.result
@@ -141,9 +141,56 @@ a	b
 7	1
 8	1
 9	1
+*** Test retry of event group that spans multiple relay log files. ***
+CREATE TABLE t2 (a int PRIMARY KEY, b BLOB) ENGINE=InnoDB;
+INSERT INTO t2 VALUES (1,"Hulubullu");
+include/stop_slave.inc
+SET @old_max= @@GLOBAL.max_relay_log_size;
+SET GLOBAL max_relay_log_size=4096;
+SET gtid_seq_no = 100;
+SET @old_server_id = @@server_id;
+SET server_id = 12;
+BEGIN;
+INSERT INTO t1 VALUES (10, 4);
+COMMIT;
+SET server_id = @old_server_id;
+SELECT * FROM t1 WHERE a >= 10 ORDER BY a;
+a	b
+10	4
+SELECT a, LENGTH(b) FROM t2 ORDER BY a;
+a	LENGTH(b)
+1	9
+2	5006
+3	5012
+SET @old_dbug= @@GLOBAL.debug_dbug;
+SET GLOBAL debug_dbug="+d,rpl_parallel_simulate_temp_err_gtid_0_x_100";
+include/start_slave.inc
+SET GLOBAL debug_dbug=@old_dbug;
+retries
+1
+SELECT * FROM t1 WHERE a >= 10 ORDER BY a;
+a	b
+10	4
+SELECT a, LENGTH(b) FROM t2 ORDER BY a;
+a	LENGTH(b)
+1	9
+2	5006
+3	5012
+INSERT INTO t1 VALUES (11,11);
+SELECT * FROM t1 WHERE a >= 10 ORDER BY a;
+a	b
+10	4
+11	11
+SELECT a, LENGTH(b) FROM t2 ORDER BY a;
+a	LENGTH(b)
+1	9
+2	5006
+3	5012
+4	5000
+SET GLOBAL max_relay_log_size=@old_max;
 include/stop_slave.inc
 SET GLOBAL slave_parallel_threads=@old_parallel_threads;
 include/start_slave.inc
-DROP TABLE t1;
+DROP TABLE t1, t2;
 DROP function foo;
 include/rpl_end.inc
diff --git a/mysql-test/suite/rpl/t/rpl_parallel_retry.test b/mysql-test/suite/rpl/t/rpl_parallel_retry.test
index a2ccce45f0c..d3be6262cb0 100644
--- a/mysql-test/suite/rpl/t/rpl_parallel_retry.test
+++ b/mysql-test/suite/rpl/t/rpl_parallel_retry.test
@@ -149,6 +149,64 @@ STOP SLAVE IO_THREAD;
 --sync_with_master
 SELECT * FROM t1 ORDER BY a;
 
+--echo *** Test retry of event group that spans multiple relay log files. ***
+
+--connection server_1
+CREATE TABLE t2 (a int PRIMARY KEY, b BLOB) ENGINE=InnoDB;
+INSERT INTO t2 VALUES (1,"Hulubullu");
+--save_master_pos
+
+--connection server_2
+--sync_with_master
+--source include/stop_slave.inc
+SET @old_max= @@GLOBAL.max_relay_log_size;
+SET GLOBAL max_relay_log_size=4096;
+
+--connection server_1
+--let $big= `SELECT REPEAT("*", 5000)`
+SET gtid_seq_no = 100;
+SET @old_server_id = @@server_id;
+SET server_id = 12;
+BEGIN;
+--disable_query_log
+eval INSERT INTO t2 VALUES (2, CONCAT("Hello ", "$big"));
+eval INSERT INTO t2 VALUES (3, CONCAT("Long data:  ", "$big"));
+--enable_query_log
+INSERT INTO t1 VALUES (10, 4);
+COMMIT;
+SET server_id = @old_server_id;
+SELECT * FROM t1 WHERE a >= 10 ORDER BY a;
+SELECT a, LENGTH(b) FROM t2 ORDER BY a;
+--save_master_pos
+
+--connection server_2
+SET @old_dbug= @@GLOBAL.debug_dbug;
+SET GLOBAL debug_dbug="+d,rpl_parallel_simulate_temp_err_gtid_0_x_100";
+let $old_retry= query_get_value(SHOW STATUS LIKE 'Slave_retried_transactions', Value, 1);
+--source include/start_slave.inc
+--sync_with_master
+SET GLOBAL debug_dbug=@old_dbug;
+let $new_retry= query_get_value(SHOW STATUS LIKE 'Slave_retried_transactions', Value, 1);
+--disable_query_log
+eval SELECT $new_retry - $old_retry AS retries;
+--enable_query_log
+
+SELECT * FROM t1 WHERE a >= 10 ORDER BY a;
+SELECT a, LENGTH(b) FROM t2 ORDER BY a;
+
+--connection server_1
+INSERT INTO t1 VALUES (11,11);
+--disable_query_log
+eval INSERT INTO t2 VALUES (4, "$big");
+--enable_query_log
+--save_master_pos
+
+--connection server_2
+--sync_with_master
+SELECT * FROM t1 WHERE a >= 10 ORDER BY a;
+SELECT a, LENGTH(b) FROM t2 ORDER BY a;
+SET GLOBAL max_relay_log_size=@old_max;
+
 
 --connection server_2
 --source include/stop_slave.inc
@@ -156,7 +214,7 @@ SET GLOBAL slave_parallel_threads=@old_parallel_threads;
 --source include/start_slave.inc
 
 --connection server_1
-DROP TABLE t1;
+DROP TABLE t1, t2;
 DROP function foo;
 
 --source include/rpl_end.inc