1
0
mirror of https://github.com/MariaDB/server.git synced 2025-08-07 00:04:31 +03:00

MDEV-28461 semisync-slave server recovery fails to rollback prepared transaction

that is not in binlog.

Post-crash recovery of --rpl-semi-sync-slave-enabled server
failed to recognize a transaction in-doubt that needed rolled back.
A prepared-but-not-in-binlog transaction gets committed instead
to possibly create inconsistency with a master (e.g the way it was observed
in the bug report).

The semisync recovery is corrected now with initializing binlog coordinates
of any transaction in-doubt to the maximum offset which is
unreachable.
In effect when a prepared transaction that is not found in binlog
it will be decided to rollback because it's guaranteed to reside
in a truncated tail area of binlog.

Mtr tests are reinforced to cover the described scenario.
This commit is contained in:
Andrei
2022-05-11 13:12:48 +03:00
committed by Sergei Golubchik
parent e03e72234a
commit 98ca71ab28
6 changed files with 119 additions and 25 deletions

View File

@@ -4,11 +4,13 @@ RESET MASTER;
SET @@global.sync_binlog=1;
CREATE TABLE t (f INT) ENGINE=INNODB;
CREATE TABLE t2 (f INT) ENGINE=INNODB;
CREATE TABLE t4 (f INT) ENGINE=INNODB;
CREATE TABLE tm (f INT) ENGINE=Aria;
# Case A.
connect master1,localhost,root,,;
connect master2,localhost,root,,;
connect master3,localhost,root,,;
connect master4,localhost,root,,;
connection default;
INSERT INTO t VALUES (10);
INSERT INTO tm VALUES (10);
@@ -23,14 +25,23 @@ connection master3;
SET DEBUG_SYNC= "now WAIT_FOR master2_ready";
SELECT @@global.gtid_binlog_pos as 'Before the crash';
Before the crash
0-1-7
0-1-8
connection master4;
SET DEBUG_SYNC= "ha_commit_trans_before_log_and_order SIGNAL master4_ready WAIT_FOR signal_never_arrives";
INSERT INTO t4 VALUES (13);
connection master3;
SET DEBUG_SYNC= "now WAIT_FOR master4_ready";
SELECT @@global.gtid_binlog_pos as 'Before the crash and never logged trx';
Before the crash and never logged trx
0-1-8
connection default;
# Kill the server
disconnect master1;
disconnect master2;
disconnect master3;
# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
FOUND 1 /Successfully truncated.*to remove transactions starting from GTID 0-1-6/ in mysqld.1.err
disconnect master4;
# restart: --rpl-semi-sync-slave-enabled=1 --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
Log_name Pos Event_type Server_id End_log_pos Info
@@ -39,6 +50,8 @@ master-bin.000001 # Query # # use `test`; CREATE TABLE t (f INT) ENGINE=INNODB
master-bin.000001 # Gtid # # GTID #-#-#
master-bin.000001 # Query # # use `test`; CREATE TABLE t2 (f INT) ENGINE=INNODB
master-bin.000001 # Gtid # # GTID #-#-#
master-bin.000001 # Query # # use `test`; CREATE TABLE t4 (f INT) ENGINE=INNODB
master-bin.000001 # Gtid # # GTID #-#-#
master-bin.000001 # Query # # use `test`; CREATE TABLE tm (f INT) ENGINE=Aria
master-bin.000001 # Gtid # # BEGIN GTID #-#-#
master-bin.000001 # Query # # use `test`; INSERT INTO t VALUES (10)
@@ -48,16 +61,20 @@ master-bin.000001 # Query # # use `test`; INSERT INTO tm VALUES (10)
master-bin.000001 # Query # # COMMIT
SELECT @@global.gtid_binlog_pos as 'After the crash';
After the crash
0-1-5
0-1-6
"One row should be present in table 't'"
SELECT * FROM t;
f
10
"No row should be present in table 't4'"
SELECT * FROM t4;
f
DELETE FROM t;
# Case B.
connect master1,localhost,root,,;
connect master2,localhost,root,,;
connect master3,localhost,root,,;
connect master4,localhost,root,,;
connection default;
INSERT INTO t VALUES (10);
INSERT INTO tm VALUES (10);
@@ -72,14 +89,23 @@ connection master3;
SET DEBUG_SYNC= "now WAIT_FOR master2_ready";
SELECT @@global.gtid_binlog_pos as 'Before the crash';
Before the crash
0-1-10
0-1-11
connection master4;
SET DEBUG_SYNC= "ha_commit_trans_before_log_and_order SIGNAL master4_ready WAIT_FOR signal_never_arrives";
INSERT INTO t4 VALUES (13);
connection master3;
SET DEBUG_SYNC= "now WAIT_FOR master4_ready";
SELECT @@global.gtid_binlog_pos as 'Before the crash and never logged trx';
Before the crash and never logged trx
0-1-11
connection default;
# Kill the server
disconnect master1;
disconnect master2;
disconnect master3;
# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
FOUND 1 /Successfully truncated.*to remove transactions starting from GTID 0-1-10/ in mysqld.1.err
disconnect master4;
# restart: --rpl-semi-sync-slave-enabled=1 --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
Log_name Pos Event_type Server_id End_log_pos Info
@@ -97,11 +123,14 @@ master-bin.000002 # Query # # use `test`; DELETE FROM t2 WHERE f = 0
master-bin.000002 # Query # # COMMIT
SELECT @@global.gtid_binlog_pos as 'After the crash';
After the crash
0-1-9
0-1-10
"One row should be present in table 't'"
SELECT * FROM t;
f
10
"No row should be present in table 't4'"
SELECT * FROM t4;
f
DELETE FROM t;
# Case C.
CREATE PROCEDURE sp_blank_xa()
@@ -114,6 +143,7 @@ END|
connect master1,localhost,root,,;
connect master2,localhost,root,,;
connect master3,localhost,root,,;
connect master4,localhost,root,,;
connection default;
INSERT INTO t VALUES (10);
INSERT INTO tm VALUES (10);
@@ -128,14 +158,23 @@ connection master3;
SET DEBUG_SYNC= "now WAIT_FOR master2_ready";
SELECT @@global.gtid_binlog_pos as 'Before the crash';
Before the crash
0-1-15
0-1-16
connection master4;
SET DEBUG_SYNC= "ha_commit_trans_before_log_and_order SIGNAL master4_ready WAIT_FOR signal_never_arrives";
INSERT INTO t4 VALUES (13);
connection master3;
SET DEBUG_SYNC= "now WAIT_FOR master4_ready";
SELECT @@global.gtid_binlog_pos as 'Before the crash and never logged trx';
Before the crash and never logged trx
0-1-16
connection default;
# Kill the server
disconnect master1;
disconnect master2;
disconnect master3;
# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
FOUND 1 /Successfully truncated.*to remove transactions starting from GTID 0-1-14/ in mysqld.1.err
disconnect master4;
# restart: --rpl-semi-sync-slave-enabled=1 --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
Log_name Pos Event_type Server_id End_log_pos Info
@@ -158,11 +197,14 @@ master-bin.000003 # Query # # use `test`; INSERT INTO tm VALUES (10)
master-bin.000003 # Query # # COMMIT
SELECT @@global.gtid_binlog_pos as 'After the crash';
After the crash
0-1-13
0-1-14
"One row should be present in table 't'"
SELECT * FROM t;
f
10
"No row should be present in table 't4'"
SELECT * FROM t4;
f
DELETE FROM t;
DROP PROCEDURE sp_blank_xa;
# Case D.
@@ -176,6 +218,7 @@ END|
connect master1,localhost,root,,;
connect master2,localhost,root,,;
connect master3,localhost,root,,;
connect master4,localhost,root,,;
connection default;
INSERT INTO t VALUES (10);
INSERT INTO tm VALUES (10);
@@ -190,14 +233,23 @@ connection master3;
SET DEBUG_SYNC= "now WAIT_FOR master2_ready";
SELECT @@global.gtid_binlog_pos as 'Before the crash';
Before the crash
0-1-20
0-1-21
connection master4;
SET DEBUG_SYNC= "ha_commit_trans_before_log_and_order SIGNAL master4_ready WAIT_FOR signal_never_arrives";
INSERT INTO t4 VALUES (13);
connection master3;
SET DEBUG_SYNC= "now WAIT_FOR master4_ready";
SELECT @@global.gtid_binlog_pos as 'Before the crash and never logged trx';
Before the crash and never logged trx
0-1-21
connection default;
# Kill the server
disconnect master1;
disconnect master2;
disconnect master3;
# restart: --rpl-semi-sync-slave-enabled=1 --sync-binlog=1
FOUND 1 /Successfully truncated.*to remove transactions starting from GTID 0-1-20/ in mysqld.1.err
disconnect master4;
# restart: --rpl-semi-sync-slave-enabled=1 --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
Log_name Pos Event_type Server_id End_log_pos Info
@@ -226,11 +278,14 @@ master-bin.000004 # Query # # XA END X'786964',X'',1
master-bin.000004 # XA_prepare # # XA PREPARE X'786964',X'',1
SELECT @@global.gtid_binlog_pos as 'After the crash';
After the crash
0-1-19
0-1-20
"One row should be present in table 't'"
SELECT * FROM t;
f
10
"No row should be present in table 't4'"
SELECT * FROM t4;
f
DELETE FROM t;
DROP PROCEDURE sp_xa;
# Cleanup