mariadb/mysql-test/main/backup_lock.test

355 lines
8.6 KiB
Text
Raw Normal View History

########################################################################
# Tests BACKUP STAGE locking
########################################################################
--source include/have_innodb.inc
--source include/have_metadata_lock_info.inc
--source include/not_embedded.inc
--source include/no_view_protocol.inc
--echo #
--echo # Testing which locks we get from all stages
--echo #
let $mdl= LOCK_MODE, LOCK_TYPE, TABLE_SCHEMA, TABLE_NAME FROM information_schema.metadata_lock_info
WHERE TABLE_NAME NOT LIKE 'innodb_%_stats';
--source ../suite/innodb/include/wait_all_purged.inc
2023-12-15 20:00:58 +01:00
BACKUP STAGE START;
eval SELECT $mdl;
BACKUP STAGE FLUSH;
eval SELECT $mdl;
BACKUP STAGE BLOCK_DDL;
eval SELECT $mdl;
BACKUP STAGE BLOCK_COMMIT;
eval SELECT $mdl;
BACKUP STAGE END;
eval SELECT $mdl;
--echo #
--echo # testing BACKUP STAGE LOCK's
--echo #
# Following connections are used in a few of the following tests
connect (con1,localhost,root,,);
connect (con2,localhost,root,,);
connection default;
--echo #
--echo # testing if BACKUP STAGE FLUSH causes deadlocks with ALTER TABLE
--echo #
create table t1 (a int) stats_persistent= 0, engine=innodb;
--source ../suite/innodb/include/wait_all_purged.inc
connection con2;
backup stage start;
connection default;
start transaction;
# Acquires MDL lock
insert into t1 values (1);
connection con1;
# Waits on MDL
--send alter table t1 add column (j int), algorithm copy
connection con2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock";
--source include/wait_condition.inc
backup stage flush;
eval SELECT $mdl;
#
# Do first test with max_statement_time, other tests later are done with
# lock_wait_timeout. This is mostly to ensure that both methods works
#
--error ER_STATEMENT_TIMEOUT
SET STATEMENT max_statement_time=1 FOR backup stage block_ddl;
--send backup stage block_ddl
connection default;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for backup lock";
--source include/wait_condition.inc
commit;
# The following select works because alter table is waiting for DDL lock
SELECT * FROM t1;
--error ER_LOCK_WAIT_TIMEOUT
SET STATEMENT lock_wait_timeout=0 FOR INSERT INTO t1 values (2);
--send INSERT INTO t1 values (2,0);
connection con2;
--reap # BLOCK_DDL
backup stage end;
connection con1;
--reap # ALTER TABLE
connection default;
--reap # INSERT
select * from t1;
drop table t1;
--echo # Test with inline alter table, which doesn't block block_commit
create table t1 (a int) engine=innodb;
start transaction;
# Acquires MDL lock
insert into t1 values (1);
connection con1;
# Waits on MDL
--send alter table t1 add column (j int)
connection con2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock";
--source include/wait_condition.inc
backup stage start;
backup stage flush;
eval SELECT $mdl;
backup stage block_ddl;
backup stage block_commit;
connection default;
SELECT * FROM t1;
--send commit
connection con2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for backup lock";
backup stage end;
connection con1;
--reap # ALTER TABLE
connection default;
--reap # commit
drop table t1;
--echo #
--echo # testing if BACKUP STAGE FLUSH causes deadlocks with DROP TABLE
--echo #
create table t1 (a int)stats_persistent=0, engine=innodb;
--source ../suite/innodb/include/wait_all_purged.inc
start transaction;
# Acquires MDL lock
insert into t1 values (1);
connection con1;
# Waits on MDL
--error ER_LOCK_WAIT_TIMEOUT
SET STATEMENT lock_wait_timeout=0 FOR DROP TABLE t1;
--send DROP TABLE t1
connection con2;
backup stage start;
backup stage flush;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for table metadata lock";
--source include/wait_condition.inc
--error ER_LOCK_WAIT_TIMEOUT
SET STATEMENT lock_wait_timeout=0 FOR SELECT * FROM t1;
backup stage block_ddl;
eval SELECT $mdl;
backup stage end;
connection default;
commit;
connection con1;
--reap # DROP TABLE
connection default;
--echo #
--echo # Check if backup stage block_dll + concurrent drop table blocks select
--echo #
create table t1 (a int) engine=innodb;
--source ../suite/innodb/include/wait_all_purged.inc
backup stage start;
backup stage block_ddl;
connection con1;
--send DROP TABLE t1
connection con2;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for backup lock";
--source include/wait_condition.inc
connection con2;
eval SELECT $mdl;
# Check that select's are not blocked
SELECT * FROM t1;
connection default;
backup stage end;
connection con1;
--reap
connection default;
--echo #
--echo # Check if backup stage block_dll overrides ddl lock for drop table
--echo #
create table t1 (a int) engine=innodb;
start transaction;
# Acquires MDL lock
insert into t1 values (1);
connection con1;
# Waits on MDL
--error ER_LOCK_WAIT_TIMEOUT
SET STATEMENT lock_wait_timeout=0 FOR DROP TABLE t1;
--send DROP TABLE t1
connection con2;
backup stage start;
backup stage flush;
backup stage block_ddl;
connection default;
commit;
connection con2;
backup stage end;
connection con1;
--reap # DROP TABLE
connection default;
--echo #
--echo # Check if BACKUP STAGE BLOCK_COMMIT blocks commit
--echo #
create table t1 (a int) engine=innodb;
start transaction;
# Acquires MDL lock
insert into t1 values (1);
connection con1;
backup stage start;
backup stage block_commit;
connection default;
--send commit
connection con1;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for backup lock";
backup stage end;
connection default;
--reap # commit
select * from t1;
drop table t1;
--echo #
--echo # Check if BACKUP STAGE BLOCK_DDL blocks create view
--echo #
create table t1 (a int) engine=innodb;
connection con1;
backup stage start;
backup stage block_ddl;
connection default;
--send create view v1 as select * from t1;
connection con1;
--sleep 2
select count(*) = 1 from information_schema.processlist;
let $wait_condition=
select count(*) = 1 from information_schema.processlist
where state = "Waiting for backup lock";
backup stage end;
connection default;
--reap # create
drop table t1;
drop view v1;
#
# End of tests using con1 and con2
#
disconnect con1;
disconnect con2;
--echo #
--echo # Test backup stage and flush tables
--echo #
BACKUP STAGE START ;
BACKUP STAGE BLOCK_DDL ;
FLUSH TABLES;
CREATE TEMPORARY TABLE t12345678_tmp (col1 INT);
drop table t12345678_tmp;
BACKUP STAGE END;
--echo #
--echo # Test BACKUP STAGES with lock timeouts
--echo #
SET GLOBAL lock_wait_timeout=0;
CREATE TABLE t_permanent_innodb (col1 INT) ENGINE = InnoDB;
CREATE TABLE t_permanent_myisam (col1 INT) ENGINE = MyISAM;
CREATE TABLE t_permanent_aria (col1 INT) ENGINE = Aria transactional=1;
Aria will now register it's transactions MDEV-22531 Remove maria::implicit_commit() MDEV-22607 Assertion `ha_info->ht() != binlog_hton' failed in MYSQL_BIN_LOG::unlog_xa_prepare From the handler point of view, Aria now looks like a transactional engine. One effect of this is that we don't need to call maria::implicit_commit() anymore. This change also forces the server to call trans_commit_stmt() after doing any read or writes to system tables. This work will also make it easier to later allow users to have system tables in other engines than Aria. To handle the case that Aria doesn't support rollback, a new handlerton flag, HTON_NO_ROLLBACK, was added to engines that has transactions without rollback (for the moment only binlog and Aria). Other things - Moved freeing of MARIA_SHARE to a separate function as the MARIA_SHARE can be still part of a transaction even if the table has closed. - Changed Aria checkpoint to use the new MARIA_SHARE free function. This fixes a possible memory leak when using S3 tables - Changed testing of binlog_hton to instead test for HTON_NO_ROLLBACK - Removed checking of has_transaction_manager() in handler.cc as we can assume that as the transaction was started by the engine, it does support transactions. - Added new class 'start_new_trans' that can be used to start indepdendent sub transactions, for example while reading mysql.proc, using help or status tables etc. - open_system_tables...() and open_proc_table_for_Read() doesn't anymore take a Open_tables_backup list. This is now handled by 'start_new_trans'. - Split thd::has_transactions() to thd::has_transactions() and thd::has_transactions_and_rollback() - Added handlerton code to free cached transactions objects. Needed by InnoDB. squash! 2ed35999f2a2d84f1c786a21ade5db716b6f1bbc
2020-05-02 12:19:53 +02:00
CREATE TABLE t_permanent_aria2 (col1 INT) ENGINE = Aria transactional=0;
INSERT INTO t_permanent_innodb SET col1 = 1;
INSERT INTO t_permanent_myisam SET col1 = 1;
INSERT INTO t_permanent_aria SET col1 = 1;
Aria will now register it's transactions MDEV-22531 Remove maria::implicit_commit() MDEV-22607 Assertion `ha_info->ht() != binlog_hton' failed in MYSQL_BIN_LOG::unlog_xa_prepare From the handler point of view, Aria now looks like a transactional engine. One effect of this is that we don't need to call maria::implicit_commit() anymore. This change also forces the server to call trans_commit_stmt() after doing any read or writes to system tables. This work will also make it easier to later allow users to have system tables in other engines than Aria. To handle the case that Aria doesn't support rollback, a new handlerton flag, HTON_NO_ROLLBACK, was added to engines that has transactions without rollback (for the moment only binlog and Aria). Other things - Moved freeing of MARIA_SHARE to a separate function as the MARIA_SHARE can be still part of a transaction even if the table has closed. - Changed Aria checkpoint to use the new MARIA_SHARE free function. This fixes a possible memory leak when using S3 tables - Changed testing of binlog_hton to instead test for HTON_NO_ROLLBACK - Removed checking of has_transaction_manager() in handler.cc as we can assume that as the transaction was started by the engine, it does support transactions. - Added new class 'start_new_trans' that can be used to start indepdendent sub transactions, for example while reading mysql.proc, using help or status tables etc. - open_system_tables...() and open_proc_table_for_Read() doesn't anymore take a Open_tables_backup list. This is now handled by 'start_new_trans'. - Split thd::has_transactions() to thd::has_transactions() and thd::has_transactions_and_rollback() - Added handlerton code to free cached transactions objects. Needed by InnoDB. squash! 2ed35999f2a2d84f1c786a21ade5db716b6f1bbc
2020-05-02 12:19:53 +02:00
INSERT INTO t_permanent_aria2 SET col1 = 1;
CREATE TABLE t_con1_innodb (col1 INT) ENGINE = InnoDB;
CREATE TABLE t_con1_myisam (col1 INT) ENGINE = MyISAM;
--connect(con1,localhost,root,,)
--connection default
BACKUP STAGE START;
BACKUP STAGE FLUSH;
BACKUP STAGE BLOCK_DDL;
BACKUP STAGE BLOCK_COMMIT;
--connection con1
Aria will now register it's transactions MDEV-22531 Remove maria::implicit_commit() MDEV-22607 Assertion `ha_info->ht() != binlog_hton' failed in MYSQL_BIN_LOG::unlog_xa_prepare From the handler point of view, Aria now looks like a transactional engine. One effect of this is that we don't need to call maria::implicit_commit() anymore. This change also forces the server to call trans_commit_stmt() after doing any read or writes to system tables. This work will also make it easier to later allow users to have system tables in other engines than Aria. To handle the case that Aria doesn't support rollback, a new handlerton flag, HTON_NO_ROLLBACK, was added to engines that has transactions without rollback (for the moment only binlog and Aria). Other things - Moved freeing of MARIA_SHARE to a separate function as the MARIA_SHARE can be still part of a transaction even if the table has closed. - Changed Aria checkpoint to use the new MARIA_SHARE free function. This fixes a possible memory leak when using S3 tables - Changed testing of binlog_hton to instead test for HTON_NO_ROLLBACK - Removed checking of has_transaction_manager() in handler.cc as we can assume that as the transaction was started by the engine, it does support transactions. - Added new class 'start_new_trans' that can be used to start indepdendent sub transactions, for example while reading mysql.proc, using help or status tables etc. - open_system_tables...() and open_proc_table_for_Read() doesn't anymore take a Open_tables_backup list. This is now handled by 'start_new_trans'. - Split thd::has_transactions() to thd::has_transactions() and thd::has_transactions_and_rollback() - Added handlerton code to free cached transactions objects. Needed by InnoDB. squash! 2ed35999f2a2d84f1c786a21ade5db716b6f1bbc
2020-05-02 12:19:53 +02:00
SET AUTOCOMMIT = 1;
# These should work as values are not changed
UPDATE t_permanent_aria SET col1 = 1;
--error ER_LOCK_WAIT_TIMEOUT
UPDATE t_permanent_innodb SET col1 = 1;
--error ER_LOCK_WAIT_TIMEOUT
UPDATE t_permanent_innodb SET col1 = 8;
--error ER_LOCK_WAIT_TIMEOUT
UPDATE t_permanent_myisam SET col1 = 8;
Aria will now register it's transactions MDEV-22531 Remove maria::implicit_commit() MDEV-22607 Assertion `ha_info->ht() != binlog_hton' failed in MYSQL_BIN_LOG::unlog_xa_prepare From the handler point of view, Aria now looks like a transactional engine. One effect of this is that we don't need to call maria::implicit_commit() anymore. This change also forces the server to call trans_commit_stmt() after doing any read or writes to system tables. This work will also make it easier to later allow users to have system tables in other engines than Aria. To handle the case that Aria doesn't support rollback, a new handlerton flag, HTON_NO_ROLLBACK, was added to engines that has transactions without rollback (for the moment only binlog and Aria). Other things - Moved freeing of MARIA_SHARE to a separate function as the MARIA_SHARE can be still part of a transaction even if the table has closed. - Changed Aria checkpoint to use the new MARIA_SHARE free function. This fixes a possible memory leak when using S3 tables - Changed testing of binlog_hton to instead test for HTON_NO_ROLLBACK - Removed checking of has_transaction_manager() in handler.cc as we can assume that as the transaction was started by the engine, it does support transactions. - Added new class 'start_new_trans' that can be used to start indepdendent sub transactions, for example while reading mysql.proc, using help or status tables etc. - open_system_tables...() and open_proc_table_for_Read() doesn't anymore take a Open_tables_backup list. This is now handled by 'start_new_trans'. - Split thd::has_transactions() to thd::has_transactions() and thd::has_transactions_and_rollback() - Added handlerton code to free cached transactions objects. Needed by InnoDB. squash! 2ed35999f2a2d84f1c786a21ade5db716b6f1bbc
2020-05-02 12:19:53 +02:00
--error ER_LOCK_WAIT_TIMEOUT
UPDATE t_permanent_aria SET col1 = 8;
Aria will now register it's transactions MDEV-22531 Remove maria::implicit_commit() MDEV-22607 Assertion `ha_info->ht() != binlog_hton' failed in MYSQL_BIN_LOG::unlog_xa_prepare From the handler point of view, Aria now looks like a transactional engine. One effect of this is that we don't need to call maria::implicit_commit() anymore. This change also forces the server to call trans_commit_stmt() after doing any read or writes to system tables. This work will also make it easier to later allow users to have system tables in other engines than Aria. To handle the case that Aria doesn't support rollback, a new handlerton flag, HTON_NO_ROLLBACK, was added to engines that has transactions without rollback (for the moment only binlog and Aria). Other things - Moved freeing of MARIA_SHARE to a separate function as the MARIA_SHARE can be still part of a transaction even if the table has closed. - Changed Aria checkpoint to use the new MARIA_SHARE free function. This fixes a possible memory leak when using S3 tables - Changed testing of binlog_hton to instead test for HTON_NO_ROLLBACK - Removed checking of has_transaction_manager() in handler.cc as we can assume that as the transaction was started by the engine, it does support transactions. - Added new class 'start_new_trans' that can be used to start indepdendent sub transactions, for example while reading mysql.proc, using help or status tables etc. - open_system_tables...() and open_proc_table_for_Read() doesn't anymore take a Open_tables_backup list. This is now handled by 'start_new_trans'. - Split thd::has_transactions() to thd::has_transactions() and thd::has_transactions_and_rollback() - Added handlerton code to free cached transactions objects. Needed by InnoDB. squash! 2ed35999f2a2d84f1c786a21ade5db716b6f1bbc
2020-05-02 12:19:53 +02:00
--error ER_LOCK_WAIT_TIMEOUT
UPDATE t_permanent_aria2 SET col1 = 8;
select * from t_permanent_innodb;
select * from t_permanent_myisam;
select * from t_permanent_aria;
select * from t_permanent_aria2;
SET AUTOCOMMIT = 0;
UPDATE t_permanent_innodb SET col1 = 9;
UPDATE t_permanent_aria SET col1 = 9;
--error ER_LOCK_WAIT_TIMEOUT
UPDATE t_permanent_myisam SET col1 = 9;
--error ER_LOCK_WAIT_TIMEOUT
UPDATE t_permanent_aria2 SET col1 = 9;
--error ER_LOCK_WAIT_TIMEOUT
DROP TABLE t_con1_innodb;
--error ER_LOCK_WAIT_TIMEOUT
DROP TABLE t_con1_myisam;
--connection default
BACKUP STAGE END;
select * from t_permanent_innodb;
select * from t_permanent_myisam;
select * from t_permanent_aria;
Aria will now register it's transactions MDEV-22531 Remove maria::implicit_commit() MDEV-22607 Assertion `ha_info->ht() != binlog_hton' failed in MYSQL_BIN_LOG::unlog_xa_prepare From the handler point of view, Aria now looks like a transactional engine. One effect of this is that we don't need to call maria::implicit_commit() anymore. This change also forces the server to call trans_commit_stmt() after doing any read or writes to system tables. This work will also make it easier to later allow users to have system tables in other engines than Aria. To handle the case that Aria doesn't support rollback, a new handlerton flag, HTON_NO_ROLLBACK, was added to engines that has transactions without rollback (for the moment only binlog and Aria). Other things - Moved freeing of MARIA_SHARE to a separate function as the MARIA_SHARE can be still part of a transaction even if the table has closed. - Changed Aria checkpoint to use the new MARIA_SHARE free function. This fixes a possible memory leak when using S3 tables - Changed testing of binlog_hton to instead test for HTON_NO_ROLLBACK - Removed checking of has_transaction_manager() in handler.cc as we can assume that as the transaction was started by the engine, it does support transactions. - Added new class 'start_new_trans' that can be used to start indepdendent sub transactions, for example while reading mysql.proc, using help or status tables etc. - open_system_tables...() and open_proc_table_for_Read() doesn't anymore take a Open_tables_backup list. This is now handled by 'start_new_trans'. - Split thd::has_transactions() to thd::has_transactions() and thd::has_transactions_and_rollback() - Added handlerton code to free cached transactions objects. Needed by InnoDB. squash! 2ed35999f2a2d84f1c786a21ade5db716b6f1bbc
2020-05-02 12:19:53 +02:00
select * from t_permanent_aria2;
Aria will now register it's transactions MDEV-22531 Remove maria::implicit_commit() MDEV-22607 Assertion `ha_info->ht() != binlog_hton' failed in MYSQL_BIN_LOG::unlog_xa_prepare From the handler point of view, Aria now looks like a transactional engine. One effect of this is that we don't need to call maria::implicit_commit() anymore. This change also forces the server to call trans_commit_stmt() after doing any read or writes to system tables. This work will also make it easier to later allow users to have system tables in other engines than Aria. To handle the case that Aria doesn't support rollback, a new handlerton flag, HTON_NO_ROLLBACK, was added to engines that has transactions without rollback (for the moment only binlog and Aria). Other things - Moved freeing of MARIA_SHARE to a separate function as the MARIA_SHARE can be still part of a transaction even if the table has closed. - Changed Aria checkpoint to use the new MARIA_SHARE free function. This fixes a possible memory leak when using S3 tables - Changed testing of binlog_hton to instead test for HTON_NO_ROLLBACK - Removed checking of has_transaction_manager() in handler.cc as we can assume that as the transaction was started by the engine, it does support transactions. - Added new class 'start_new_trans' that can be used to start indepdendent sub transactions, for example while reading mysql.proc, using help or status tables etc. - open_system_tables...() and open_proc_table_for_Read() doesn't anymore take a Open_tables_backup list. This is now handled by 'start_new_trans'. - Split thd::has_transactions() to thd::has_transactions() and thd::has_transactions_and_rollback() - Added handlerton code to free cached transactions objects. Needed by InnoDB. squash! 2ed35999f2a2d84f1c786a21ade5db716b6f1bbc
2020-05-02 12:19:53 +02:00
DROP TABLE t_permanent_myisam, t_permanent_innodb, t_permanent_aria, t_permanent_aria2;
DROP TABLE t_con1_innodb, t_con1_myisam;
--disconnect con1
set global lock_wait_timeout=default;