mirror of
https://github.com/MariaDB/server.git
synced 2025-10-31 02:46:29 +01:00
0a68328673
This is actually an existing problem in the old binlog implementation, and this patch is applicable to old binlog also. The problem is that RESET MASTER can run concurrently with binlog dump threads / connected slaves. This will remove the binlog from under the feet of the reader, which can cause all sorts of strange behaviour. This patch fixes the problem by disallowing to run RESET MASTER when dump threads (or other RESET MASTER or SHOW BINARY LOGS) are running. An error is thrown in this case, user must stop slaves and/or kill dump threads to make the RESET MASTER go through. A slave that connects in the middle of RESET MASTER will wait for it to complete. Fix a lot of test cases to kill any lingering dump threads before doing RESET MASTER, mostly just by sourcing include/kill_binlog_dump_threads.inc. Signed-off-by: Kristian Nielsen <knielsen@knielsen-hq.org>
292 lines
7.8 KiB
Text
292 lines
7.8 KiB
Text
include/master-slave.inc
|
|
[connection master]
|
|
connection slave;
|
|
include/stop_slave.inc
|
|
RESET MASTER;
|
|
RESET SLAVE;
|
|
connection master;
|
|
include/kill_binlog_dump_threads.inc
|
|
RESET MASTER;
|
|
CREATE TABLE t1 (a int primary key, b text) ENGINE=InnoDB;
|
|
INSERT INTO t1 SET a=25, b='trx0';
|
|
connection slave;
|
|
include/start_slave.inc
|
|
connection master;
|
|
connection slave;
|
|
include/stop_slave.inc
|
|
ALTER TABLE mysql.gtid_slave_pos ENGINE=InnoDB;
|
|
SET @old_parallel_threads=@@GLOBAL.slave_parallel_threads;
|
|
SET GLOBAL slave_parallel_threads=2;
|
|
SET @old_parallel_mode=@@GLOBAL.slave_parallel_mode;
|
|
SET GLOBAL slave_parallel_mode='optimistic';
|
|
connection slave;
|
|
SET @old_max_relay_log_size = @@global.max_relay_log_size;
|
|
SET @@global.max_relay_log_size=4096;
|
|
connection master;
|
|
BEGIN;
|
|
INSERT INTO t1 SET a=1, b='trx1';
|
|
INSERT INTO t1 SET a=2, b='trx1';
|
|
INSERT INTO t1 SET a=3, b='trx1';
|
|
INSERT INTO t1 SET a=4, b='trx1';
|
|
INSERT INTO t1 SET a=5, b='trx1';
|
|
INSERT INTO t1 SET a=6, b='trx1';
|
|
INSERT INTO t1 SET a=7, b='trx1';
|
|
INSERT INTO t1 SET a=8, b='trx1';
|
|
INSERT INTO t1 SET a=9, b='trx1';
|
|
INSERT INTO t1 SET a=10, b='trx1';
|
|
INSERT INTO t1 SET a=11, b='trx1';
|
|
INSERT INTO t1 SET a=12, b='trx1';
|
|
INSERT INTO t1 SET a=13, b='trx1';
|
|
INSERT INTO t1 SET a=14, b='trx1';
|
|
INSERT INTO t1 SET a=15, b='trx1';
|
|
INSERT INTO t1 SET a=16, b='trx1';
|
|
INSERT INTO t1 SET a=17, b='trx1';
|
|
INSERT INTO t1 SET a=18, b='trx1';
|
|
INSERT INTO t1 SET a=19, b='trx1';
|
|
INSERT INTO t1 SET a=20, b='trx1';
|
|
INSERT INTO t1 SET a=21, b='trx1';
|
|
INSERT INTO t1 SET a=22, b='trx1';
|
|
INSERT INTO t1 SET a=23, b='trx1';
|
|
INSERT INTO t1 SET a=24, b='trx1';
|
|
COMMIT;
|
|
FLUSH LOGS;
|
|
BEGIN;
|
|
UPDATE t1 SET b='trx2_0' WHERE a = 25;
|
|
UPDATE t1 SET b='trx2' WHERE a = 25;
|
|
COMMIT;
|
|
INSERT INTO t1 SET a=26,b='trx3';
|
|
*** case 1 UNTIL inside trx2
|
|
connection slave1;
|
|
BEGIN;
|
|
INSERT INTO t1 SET a= 1;
|
|
connection slave;
|
|
SELECT <pos_0> <= <pos_until> AND <pos_until> < <pos_trx2> as "pos_until < trx0 and is within trx2";
|
|
pos_until < trx0 and is within trx2
|
|
1
|
|
CHANGE MASTER TO MASTER_USE_GTID=no;
|
|
START SLAVE UNTIL MASTER_LOG_FILE = 'file_2', MASTER_LOG_POS = <pos_until>;
|
|
connection slave1;
|
|
ROLLBACK;
|
|
Proof 1: Correct stop
|
|
connection slave;
|
|
include/wait_for_slave_sql_to_stop.inc
|
|
SELECT count(*) = 1 as 'trx2 is committed' FROM t1 WHERE b = 'trx2';
|
|
trx2 is committed
|
|
1
|
|
SELECT count(*) = 0 as 'trx3 is not committed' FROM t1 WHERE b = 'trx3';
|
|
trx3 is not committed
|
|
1
|
|
Proof 2: Resume works out
|
|
include/start_slave.inc
|
|
connection master;
|
|
connection slave;
|
|
*** case 2 UNTIL inside trx2
|
|
connection slave;
|
|
DELETE FROM t1 WHERE a <> 25;
|
|
UPDATE t1 SET b='trx0' WHERE a = 25;
|
|
connection slave1;
|
|
BEGIN;
|
|
INSERT INTO t1 SET a= 1;
|
|
connection slave;
|
|
include/stop_slave.inc
|
|
SELECT <pos_0> <= <pos_until> AND <pos_until> < <pos_trx2> as "pos_until >= trx0 and is within trx2";
|
|
pos_until >= trx0 and is within trx2
|
|
1
|
|
CHANGE MASTER TO MASTER_LOG_FILE = 'file_1', MASTER_LOG_POS = <pos_trx0>, MASTER_USE_GTID=no;
|
|
START SLAVE UNTIL MASTER_LOG_FILE = 'file_2', MASTER_LOG_POS = <pos_until>;
|
|
connection slave1;
|
|
ROLLBACK;
|
|
Proof 1: Correct stop
|
|
connection slave;
|
|
include/wait_for_slave_sql_to_stop.inc
|
|
SELECT count(*) = 1 as 'trx2 is committed' FROM t1 WHERE b = 'trx2';
|
|
trx2 is committed
|
|
1
|
|
SELECT count(*) = 0 as 'trx3 is not committed' FROM t1 WHERE b = 'trx3';
|
|
trx3 is not committed
|
|
1
|
|
Proof 2: Resume works out
|
|
include/start_slave.inc
|
|
connection master;
|
|
connection slave;
|
|
*** case 3 UNTIL inside trx1
|
|
connection slave;
|
|
DELETE FROM t1 WHERE a <> 25;
|
|
UPDATE t1 SET b='trx0' WHERE a = 25;
|
|
connection slave1;
|
|
BEGIN;
|
|
INSERT INTO t1 SET a= 1; # block trx1;
|
|
connection slave;
|
|
include/stop_slave.inc
|
|
SELECT <pos_until> < <pos_0> as "pos_until before trx2 start position";
|
|
pos_until before trx2 start position
|
|
1
|
|
CHANGE MASTER TO MASTER_LOG_FILE = 'file_1', MASTER_LOG_POS = <pos_trx0>, MASTER_USE_GTID=no;
|
|
START SLAVE UNTIL MASTER_LOG_FILE = 'file_2', MASTER_LOG_POS = <pos_until>;
|
|
connection slave1;
|
|
ROLLBACK;
|
|
Proof 1: Correct stop
|
|
connection slave;
|
|
include/wait_for_slave_sql_to_stop.inc
|
|
SELECT count(*) = 25-1 as 'trx1 is committed' FROM t1 WHERE b = 'trx1';
|
|
trx1 is committed
|
|
1
|
|
SELECT count(*) = 0 as 'trx2 is not committed' FROM t1 WHERE b = 'trx2';
|
|
trx2 is not committed
|
|
1
|
|
Proof 2: Resume works out
|
|
include/start_slave.inc
|
|
connection master;
|
|
connection slave;
|
|
*** case 4 Relay-log UNTIL inside trx1
|
|
connection slave;
|
|
DELETE FROM t1 WHERE a <> 25;
|
|
UPDATE t1 SET b='trx0' WHERE a = 25;
|
|
connection slave1;
|
|
BEGIN;
|
|
INSERT INTO t1 SET a= 1; # block trx1;
|
|
connection slave;
|
|
include/stop_slave.inc
|
|
CHANGE MASTER TO MASTER_LOG_FILE = 'file_1', MASTER_LOG_POS = <pos_trx0>, MASTER_USE_GTID=no;
|
|
START SLAVE IO_THREAD;
|
|
include/wait_for_slave_io_to_start.inc
|
|
START SLAVE UNTIL RELAY_LOG_FILE = 'file_2', RELAY_LOG_POS = <pos_until>;
|
|
connection slave1;
|
|
ROLLBACK;
|
|
Proof 1: Correct stop
|
|
connection slave;
|
|
include/wait_for_slave_sql_to_stop.inc
|
|
SELECT count(*) = 25-1 as 'trx1 is committed' FROM t1 WHERE b = 'trx1';
|
|
trx1 is committed
|
|
1
|
|
SELECT count(*) = 0 as 'trx2 is not committed' FROM t1 WHERE b = 'trx2';
|
|
trx2 is not committed
|
|
1
|
|
Proof 2: Resume works out
|
|
include/start_slave.inc
|
|
connection master;
|
|
connection slave;
|
|
*** case 5 Relay-log UNTIL inside a "big" trx that spawns few relay logs
|
|
connection master;
|
|
CREATE TABLE t2 (a TEXT) ENGINE=InnoDB;
|
|
FLUSH LOGS;
|
|
connection slave;
|
|
connection slave;
|
|
include/stop_slave.inc
|
|
connection master;
|
|
BEGIN;
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
INSERT INTO t2 SET a=repeat('a',1024);
|
|
COMMIT;
|
|
INSERT INTO t2 SET a='a';
|
|
connection slave;
|
|
START SLAVE IO_THREAD;
|
|
include/wait_for_slave_io_to_start.inc
|
|
START SLAVE UNTIL RELAY_LOG_FILE = 'file_2', RELAY_LOG_POS = <pos_until>;
|
|
Proof 1: Correct stop
|
|
connection slave;
|
|
include/wait_for_slave_sql_to_stop.inc
|
|
Proof 2: Resume works out
|
|
include/start_slave.inc
|
|
connection master;
|
|
connection slave;
|
|
include/diff_tables.inc [master:t2,slave:t2]
|
|
*** case 6 Relay-log UNTIL inside a small trx inside a sequence of relay logs
|
|
connection slave;
|
|
include/stop_slave.inc
|
|
connection master;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
BEGIN;
|
|
DELETE FROM t2 LIMIT 1;
|
|
COMMIT;
|
|
COMMIT;
|
|
connection slave;
|
|
START SLAVE IO_THREAD;
|
|
include/wait_for_slave_io_to_start.inc
|
|
connection master;
|
|
include/sync_slave_io_with_master.inc
|
|
connection slave;
|
|
START SLAVE UNTIL RELAY_LOG_FILE = 'file_2', RELAY_LOG_POS = <pos_until>;
|
|
Proof 1: Correct stop
|
|
connection slave;
|
|
include/wait_for_slave_sql_to_stop.inc
|
|
Proof 2: Resume works out
|
|
include/start_slave.inc
|
|
connection master;
|
|
connection slave;
|
|
include/diff_tables.inc [master:t2,slave:t2]
|
|
connection slave;
|
|
include/stop_slave.inc
|
|
SET GLOBAL max_relay_log_size=@old_max_relay_log_size;
|
|
SET GLOBAL slave_parallel_mode=@old_parallel_mode;
|
|
SET GLOBAL slave_parallel_threads=@old_parallel_threads;
|
|
include/start_slave.inc
|
|
connection master;
|
|
DROP TABLE t1, t2;
|
|
connection slave;
|
|
include/rpl_end.inc
|