Skip to content

M
mariadb
Commit f8b64e17 authored by rafal@quant.(none)'s avatar rafal@quant.(none)
Browse files
Options

BUG#21842 (Cluster fails to replicate to innodb or myisam with err 134 

using TPC-B):
 
Problem: A RBR event can contain incomplete row data (only key value and
fields which have been changed). In that case, when the row is unpacked
into record and written to a table, the missing fields get incorrect NULL
values leading to master-slave inconsistency.
 
Solution: Use values found in slave's table for columns which are not given
in the rows event. The code for writing a single row uses the following 
algorithm: 

1. unpack row_data into table->record[0],
2. try to insert record,
3. if duplicate record found, fetch it into table->record[0],
4. unpack row_data into table->record[0],
5. write table->record[0] into the table.

Where row_data is the row as stored in the data area of a rows event. 
Thus:

a) unpacking of row_data happens at the time when row is written into 
 a table,

b) when unpacking (in step 4), only columns present in row_data are 
 overwritten - all other columns remain as they were found in the table.
 
Since all data needed for the above algorithm is stored inside 
Rows_log_event class, functions which locate and write rows are turned 
into methods of that class.

replace_record()     -> Rows_log_event::write_row()
find_and_fetch_row() -> Rows_log_event::find_row()

Both methods take row data from event's data buffer - the row being 
processed is pointed by m_curr_row. They unpack the data as needed into 
table's record buffers record[0] or record[1]. When row is unpacked, 
m_curr_row_end is set to point at next row in the data buffer.

Other changes introduced in this changeset:

- Change signature of unpack_row(): don't report errors and don't
setup table's rw_set here. Errors can happen only when setting default 
values in prepare_record() function and are detected there.
 
- In Rows_log_event and derived classes, don't pass arguments to
the execution primitives (do_...() member functions) but use class
members instead.

- Move old row handling code into log_event_old.cc to be used by 
*_rows_log_event_old classes.

Also, a new test rpl_ndb_2other is added which tests basic replication 
from master using ndb tables to slave storing the same tables using 
(possibly) different engine (myisam,innodb).
  
Test is based on existing tests rpl_ndb_2myisam and rpl_ndb_2innodb. 
However, these tests doesn't work for various reasons and currently are 
disabled (see BUG#19227).
  
The new test differs from the ones it is based on as follows:
  
1. Single test tests replication with different storage engines on slave 
(myisam, innodb, ndb).
  
2. Include file extra/rpl_tests/rpl_ndb_2multi_eng.test containing 
original tests is replaced by extra/rpl_tests/rpl_ndb_2multi_basic.test 
which doesn't contain tests using partitioned tables as these don't work 
currently. Instead, it tests replication to a slave which has more or 
less columns than master.
  
3. Include file include/rpl_multi_engine3.inc is replaced with 
include/rpl_multi_engine2.inc. The later differs by performing slightly 
different operations (updating more than one row in the table) and 
clearing table with "TRUNCATE TABLE" statement instead of "DELETE FROM" 
as replication of "DELETE" doesn't work well in this setting.
  
4. Slave must use option --log-slave-updates=0 as otherwise execution of 
replication events generated by ndb fails if table uses a different 
storage engine on slave (see BUG#29569).
parent c7cf3e05
Show whitespace changes
Inline Side-by-side
Showing with 3188 additions and 566 deletions
mysql-test/extra/rpl_tests/rpl_ndb_2multi_basic.test 0 → 100644
View file @ f8b64e17
#######################################
# Author: Rafal Somla #
# Date: 2006-08-20 #
# Purpose: Test replication of basic #
# table operations in various setups #
# #
# Based on rpl_ndb_2multi_eng.test by #
# JBM #
#######################################
--echo --- Doing pre test cleanup ---
connection master;
--disable_warnings
DROP TABLE IF EXISTS t1;
--enable_query_log
#################################################
--echo --- Create Table Section ---
CREATE TABLE t1 (id MEDIUMINT NOT NULL,
b1 INT,
vc VARCHAR(255),
bc CHAR(255),
d DECIMAL(10,4) DEFAULT 0,
f FLOAT DEFAULT 0,
total BIGINT UNSIGNED,
y YEAR,
t DATE,
PRIMARY KEY(id));
--echo --- Show table on master ---
SHOW CREATE TABLE t1;
--echo --- Show table on slave ---
sync_slave_with_master;
SHOW CREATE TABLE t1;
--source include/rpl_multi_engine2.inc
#################################################
# Okay lets see how it holds up to table changes
--echo --- Check that simple Alter statements are replicated correctly --
ALTER TABLE t1 DROP PRIMARY KEY;
# note: table with no PK can't contain blobs if it is to be replicated.
ALTER TABLE t1 MODIFY vc char(32);
--echo --- Show the new improved table on the master ---
SHOW CREATE TABLE t1;
--echo --- Make sure that our tables on slave are still same engine ---
--echo --- and that the alter statements replicated correctly ---
sync_slave_with_master;
SHOW CREATE TABLE t1;
--source include/rpl_multi_engine2.inc
#################################################
--echo --- Check that replication works when slave has more columns than master
connection master;
ALTER TABLE t1 ADD PRIMARY KEY(id,total);
ALTER TABLE t1 MODIFY vc TEXT;
INSERT INTO t1 VALUES(3,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1905-11-14");
INSERT INTO t1 VALUES(20,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1965-11-14");
INSERT INTO t1 VALUES(50,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1985-11-14");
--echo --- Add columns on slave ---
--sync_slave_with_master
ALTER TABLE t1 ADD (u int, v char(16) default 'default');
UPDATE t1 SET u=7 WHERE id < 50;
UPDATE t1 SET v='explicit' WHERE id >10;
--echo --- Show changed table on slave ---
SHOW CREATE TABLE t1;
SELECT *
FROM t1
ORDER BY id;
--source include/rpl_multi_engine2.inc
TRUNCATE TABLE t1;
#################################################
--echo --- Check that replication works when master has more columns than slave
connection master;
--echo --- Remove columns on slave ---
--sync_slave_with_master
ALTER TABLE t1 DROP COLUMN v;
ALTER TABLE t1 DROP COLUMN u;
ALTER TABLE t1 DROP COLUMN t;
ALTER TABLE t1 DROP COLUMN y;
--echo --- Show changed table on slave ---
SHOW CREATE TABLE t1;
--source include/rpl_multi_engine2.inc
TRUNCATE TABLE t1;
#################################################
--echo --- Do Cleanup --
connection master;
DROP TABLE IF EXISTS t1;
sync_slave_with_master;
connection master;
mysql-test/include/rpl_multi_engine2.inc 0 → 100644
View file @ f8b64e17
#############################################################
# Author: Rafal
# Date: 2007-08-20
# based on rpl_multi_engine3.inc
#############################################################
connection slave;
STOP SLAVE;
RESET SLAVE;
connection master;
RESET MASTER;
connection slave;
START SLAVE;
--echo --- Populate t1 with data ---
connection master;
--disable_query_log
INSERT INTO t1 VALUES(42,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1905-11-14");
INSERT INTO t1 VALUES(2,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1965-11-14");
INSERT INTO t1 VALUES(4,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1985-11-14");
INSERT INTO t1 VALUES(142,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1995-11-14");
INSERT INTO t1 VALUES(412,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"2005-11-14");
--enable_query_log
--echo --- Select from t1 on master ---
select *
from t1
order by id;
sync_slave_with_master;
--echo --- Select from t1 on slave ---
select *
from t1
order by id;
--echo --- Perform basic operation on master ---
--echo --- and ensure replicated correctly ---
connection master;
--echo --- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--echo --- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
# Must give injector thread a little time to get update
# into the binlog other wise we will miss the update.
sync_slave_with_master;
--echo --- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
connection master;
--echo --- Remove a record from t1 on master ---
# Note: there is an error in replication of Delete_row
# from NDB to MyISAM (BUG#28538). However, if there is
# only one row in Delete_row event then it works fine,
# as this test demonstrates.
DELETE FROM t1 WHERE id = 412;
--echo --- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
sync_slave_with_master;
--echo --- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
connection master;
TRUNCATE TABLE t1;
sync_slave_with_master;
connection master;
mysql-test/suite/rpl_ndb/r/rpl_ndb_2other.result 0 → 100644
View file @ f8b64e17
stop slave;
drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t9;
reset master;
reset slave;
drop table if exists t1,t2,t3,t4,t5,t6,t7,t8,t9;
start slave;
SET storage_engine=ndb;
=== NDB -> MYISAM ===
SET storage_engine=myisam;
--- Doing pre test cleanup ---
DROP TABLE IF EXISTS t1;
--- Create Table Section ---
CREATE TABLE t1 (id MEDIUMINT NOT NULL,
b1 INT,
vc VARCHAR(255),
bc CHAR(255),
d DECIMAL(10,4) DEFAULT 0,
f FLOAT DEFAULT 0,
total BIGINT UNSIGNED,
y YEAR,
t DATE,
PRIMARY KEY(id));
--- Show table on master ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` varchar(255) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=ndbcluster DEFAULT CHARSET=latin1
--- Show table on slave ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` varchar(255) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
TRUNCATE TABLE t1;
--- Check that simple Alter statements are replicated correctly --
ALTER TABLE t1 DROP PRIMARY KEY;
ALTER TABLE t1 MODIFY vc char(32);
--- Show the new improved table on the master ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` char(32) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL
) ENGINE=ndbcluster DEFAULT CHARSET=latin1
--- Make sure that our tables on slave are still same engine ---
--- and that the alter statements replicated correctly ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` char(32) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL
) ENGINE=MyISAM DEFAULT CHARSET=latin1
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
TRUNCATE TABLE t1;
--- Check that replication works when slave has more columns than master
ALTER TABLE t1 ADD PRIMARY KEY(id,total);
ALTER TABLE t1 MODIFY vc TEXT;
INSERT INTO t1 VALUES(3,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1905-11-14");
INSERT INTO t1 VALUES(20,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1965-11-14");
INSERT INTO t1 VALUES(50,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1985-11-14");
--- Add columns on slave ---
ALTER TABLE t1 ADD (u int, v char(16) default 'default');
UPDATE t1 SET u=7 WHERE id < 50;
UPDATE t1 SET v='explicit' WHERE id >10;
--- Show changed table on slave ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` text,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned NOT NULL DEFAULT '0',
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL,
`u` int(11) DEFAULT NULL,
`v` char(16) DEFAULT 'default',
PRIMARY KEY (`id`,`total`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
SELECT *
FROM t1
ORDER BY id;
id b1 vc bc d f total y t u v
3 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14 7 default
20 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14 7 explicit
50 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14 NULL explicit
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
3 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
20 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
50 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total y t u v
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14 NULL default
3 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14 7 default
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14 NULL default
20 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14 7 explicit
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14 NULL default
50 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14 NULL explicit
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14 NULL default
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14 NULL default
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
3 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
20 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
50 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t u v
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL default
3 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 7 default
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL default
20 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 7 explicit
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL default
50 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL explicit
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
7
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
7
TRUNCATE TABLE t1;
TRUNCATE TABLE t1;
--- Check that replication works when master has more columns than slave
--- Remove columns on slave ---
ALTER TABLE t1 DROP COLUMN v;
ALTER TABLE t1 DROP COLUMN u;
ALTER TABLE t1 DROP COLUMN t;
ALTER TABLE t1 DROP COLUMN y;
--- Show changed table on slave ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` text,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned NOT NULL DEFAULT '0',
PRIMARY KEY (`id`,`total`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
TRUNCATE TABLE t1;
TRUNCATE TABLE t1;
--- Do Cleanup --
DROP TABLE IF EXISTS t1;
=== NDB -> INNODB ===
SET storage_engine=innodb;
--- Doing pre test cleanup ---
DROP TABLE IF EXISTS t1;
--- Create Table Section ---
CREATE TABLE t1 (id MEDIUMINT NOT NULL,
b1 INT,
vc VARCHAR(255),
bc CHAR(255),
d DECIMAL(10,4) DEFAULT 0,
f FLOAT DEFAULT 0,
total BIGINT UNSIGNED,
y YEAR,
t DATE,
PRIMARY KEY(id));
--- Show table on master ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` varchar(255) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=ndbcluster DEFAULT CHARSET=latin1
--- Show table on slave ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` varchar(255) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
TRUNCATE TABLE t1;
--- Check that simple Alter statements are replicated correctly --
ALTER TABLE t1 DROP PRIMARY KEY;
ALTER TABLE t1 MODIFY vc char(32);
--- Show the new improved table on the master ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` char(32) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL
) ENGINE=ndbcluster DEFAULT CHARSET=latin1
--- Make sure that our tables on slave are still same engine ---
--- and that the alter statements replicated correctly ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` char(32) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL
) ENGINE=MyISAM DEFAULT CHARSET=latin1
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
TRUNCATE TABLE t1;
--- Check that replication works when slave has more columns than master
ALTER TABLE t1 ADD PRIMARY KEY(id,total);
ALTER TABLE t1 MODIFY vc TEXT;
INSERT INTO t1 VALUES(3,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1905-11-14");
INSERT INTO t1 VALUES(20,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1965-11-14");
INSERT INTO t1 VALUES(50,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1985-11-14");
--- Add columns on slave ---
ALTER TABLE t1 ADD (u int, v char(16) default 'default');
UPDATE t1 SET u=7 WHERE id < 50;
UPDATE t1 SET v='explicit' WHERE id >10;
--- Show changed table on slave ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` text,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned NOT NULL DEFAULT '0',
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL,
`u` int(11) DEFAULT NULL,
`v` char(16) DEFAULT 'default',
PRIMARY KEY (`id`,`total`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
SELECT *
FROM t1
ORDER BY id;
id b1 vc bc d f total y t u v
3 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14 7 default
20 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14 7 explicit
50 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14 NULL explicit
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
3 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
20 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
50 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total y t u v
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14 NULL default
3 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14 7 default
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14 NULL default
20 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14 7 explicit
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14 NULL default
50 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14 NULL explicit
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14 NULL default
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14 NULL default
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
3 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
20 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
50 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t u v
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL default
3 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 7 default
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL default
20 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 7 explicit
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL default
50 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL explicit
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
7
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
7
TRUNCATE TABLE t1;
TRUNCATE TABLE t1;
--- Check that replication works when master has more columns than slave
--- Remove columns on slave ---
ALTER TABLE t1 DROP COLUMN v;
ALTER TABLE t1 DROP COLUMN u;
ALTER TABLE t1 DROP COLUMN t;
ALTER TABLE t1 DROP COLUMN y;
--- Show changed table on slave ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` text,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned NOT NULL DEFAULT '0',
PRIMARY KEY (`id`,`total`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
TRUNCATE TABLE t1;
TRUNCATE TABLE t1;
--- Do Cleanup --
DROP TABLE IF EXISTS t1;
=== NDB -> NDB ===
SET storage_engine=ndb;
--- Doing pre test cleanup ---
DROP TABLE IF EXISTS t1;
--- Create Table Section ---
CREATE TABLE t1 (id MEDIUMINT NOT NULL,
b1 INT,
vc VARCHAR(255),
bc CHAR(255),
d DECIMAL(10,4) DEFAULT 0,
f FLOAT DEFAULT 0,
total BIGINT UNSIGNED,
y YEAR,
t DATE,
PRIMARY KEY(id));
--- Show table on master ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` varchar(255) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=ndbcluster DEFAULT CHARSET=latin1
--- Show table on slave ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` varchar(255) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL,
PRIMARY KEY (`id`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
TRUNCATE TABLE t1;
--- Check that simple Alter statements are replicated correctly --
ALTER TABLE t1 DROP PRIMARY KEY;
ALTER TABLE t1 MODIFY vc char(32);
--- Show the new improved table on the master ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` char(32) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL
) ENGINE=ndbcluster DEFAULT CHARSET=latin1
--- Make sure that our tables on slave are still same engine ---
--- and that the alter statements replicated correctly ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` char(32) DEFAULT NULL,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned DEFAULT NULL,
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL
) ENGINE=MyISAM DEFAULT CHARSET=latin1
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a coo Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a coo updated 654321.4321 15.21 0 1965 2006-02-22
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
TRUNCATE TABLE t1;
--- Check that replication works when slave has more columns than master
ALTER TABLE t1 ADD PRIMARY KEY(id,total);
ALTER TABLE t1 MODIFY vc TEXT;
INSERT INTO t1 VALUES(3,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1905-11-14");
INSERT INTO t1 VALUES(20,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1965-11-14");
INSERT INTO t1 VALUES(50,1,'Testing MySQL databases is a cool ',
'Must make it bug free for the customer',
654321.4321,15.21,0,1965,"1985-11-14");
--- Add columns on slave ---
ALTER TABLE t1 ADD (u int, v char(16) default 'default');
UPDATE t1 SET u=7 WHERE id < 50;
UPDATE t1 SET v='explicit' WHERE id >10;
--- Show changed table on slave ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` text,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned NOT NULL DEFAULT '0',
`y` year(4) DEFAULT NULL,
`t` date DEFAULT NULL,
`u` int(11) DEFAULT NULL,
`v` char(16) DEFAULT 'default',
PRIMARY KEY (`id`,`total`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
SELECT *
FROM t1
ORDER BY id;
id b1 vc bc d f total y t u v
3 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14 7 default
20 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14 7 explicit
50 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14 NULL explicit
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
3 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
20 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
50 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total y t u v
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14 NULL default
3 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14 7 default
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14 NULL default
20 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14 7 explicit
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14 NULL default
50 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14 NULL explicit
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14 NULL default
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14 NULL default
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
3 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
20 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
50 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t u v
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL default
3 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 7 default
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL default
20 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 7 explicit
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL default
50 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22 NULL explicit
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
7
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
7
TRUNCATE TABLE t1;
TRUNCATE TABLE t1;
--- Check that replication works when master has more columns than slave
--- Remove columns on slave ---
ALTER TABLE t1 DROP COLUMN v;
ALTER TABLE t1 DROP COLUMN u;
ALTER TABLE t1 DROP COLUMN t;
ALTER TABLE t1 DROP COLUMN y;
--- Show changed table on slave ---
SHOW CREATE TABLE t1;
Table Create Table
t1 CREATE TABLE `t1` (
`id` mediumint(9) NOT NULL,
`b1` int(11) DEFAULT NULL,
`vc` text,
`bc` char(255) DEFAULT NULL,
`d` decimal(10,4) DEFAULT '0.0000',
`f` float DEFAULT '0',
`total` bigint(20) unsigned NOT NULL DEFAULT '0',
PRIMARY KEY (`id`,`total`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
STOP SLAVE;
RESET SLAVE;
RESET MASTER;
START SLAVE;
--- Populate t1 with data ---
--- Select from t1 on master ---
select *
from t1
order by id;
id b1 vc bc d f total y t
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1965-11-14
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1985-11-14
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1905-11-14
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 1995-11-14
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0 1965 2005-11-14
--- Select from t1 on slave ---
select *
from t1
order by id;
id b1 vc bc d f total
2 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
4 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
42 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
142 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
412 1 Testing MySQL databases is a cool Must make it bug free for the customer 654321.4321 15.21 0
--- Perform basic operation on master ---
--- and ensure replicated correctly ---
--- Update t1 on master --
UPDATE t1 SET b1 = 0, bc='updated', t="2006-02-22"
WHERE id < 100
ORDER BY id;
--- Check the update on master ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total y t
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0 1965 2006-02-22
--- Check Update on slave ---
SELECT *
FROM t1
WHERE id < 100
ORDER BY id;
id b1 vc bc d f total
2 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0
4 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0
42 0 Testing MySQL databases is a cool updated 654321.4321 15.21 0
--- Remove a record from t1 on master ---
DELETE FROM t1 WHERE id = 412;
--- Show current count on master for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
--- Show current count on slave for t1 ---
SELECT COUNT(*) FROM t1;
COUNT(*)
4
TRUNCATE TABLE t1;
TRUNCATE TABLE t1;
--- Do Cleanup --
DROP TABLE IF EXISTS t1;
mysql-test/suite/rpl_ndb/t/rpl_ndb_2other.test 0 → 100644
View file @ f8b64e17
#############################################################
# Author: Rafal Somla
# Date: 2006-08-20
# Purpose: Trying to test ability to replicate from cluster
# to other engines (innodb, myisam).
##############################################################
--source include/have_ndb.inc
--source include/have_innodb.inc
--source include/have_binlog_format_mixed_or_row.inc
--source include/master-slave.inc
# On master use NDB as storage engine.
connection master;
SET storage_engine=ndb;
--echo
--echo === NDB -> MYISAM ===
--echo
connection slave;
SET storage_engine=myisam;
--source extra/rpl_tests/rpl_ndb_2multi_basic.test
--echo
--echo === NDB -> INNODB ===
--echo
connection slave;
SET storage_engine=innodb;
--source extra/rpl_tests/rpl_ndb_2multi_basic.test
--echo
--echo === NDB -> NDB ===
--echo
connection slave;
SET storage_engine=ndb;
--source extra/rpl_tests/rpl_ndb_2multi_basic.test
sql/log_event.cc
View file @ f8b64e17
......@@ -5654,13 +5654,14 @@ Rows_log_event::Rows_log_event(THD *thd_arg, TABLE *tbl_arg, ulong tid,
m_table_id(tid),
m_width(tbl_arg ? tbl_arg->s->fields : 1),
m_rows_buf(0), m_rows_cur(0), m_rows_end(0),
m_flags(0)
m_curr_row(NULL), m_curr_row_end(NULL),
m_flags(0), m_key(NULL)
{
/*
We allow a special form of dummy event when the table, and cols
are null and the table id is ~0UL. This is a temporary
solution, to be able to terminate a started statement in the
binary log: the extreneous events will be removed in the future.
binary log: the extraneous events will be removed in the future.
*/
DBUG_ASSERT(tbl_arg && tbl_arg->s && tid != ~0UL ||
!tbl_arg && !cols && tid == ~0UL);
......@@ -5669,7 +5670,7 @@ Rows_log_event::Rows_log_event(THD *thd_arg, TABLE *tbl_arg, ulong tid,
set_flags(NO_FOREIGN_KEY_CHECKS_F);
if (thd_arg->options & OPTION_RELAXED_UNIQUE_CHECKS)
set_flags(RELAXED_UNIQUE_CHECKS_F);
/* if bitmap_init fails, catched in is_valid() */
/* if bitmap_init fails, caught in is_valid() */
if (likely(!bitmap_init(&m_cols,
m_width <= sizeof(m_bitbuf)*8 ? m_bitbuf : NULL,
m_width,
......@@ -5696,7 +5697,10 @@ Rows_log_event::Rows_log_event(const char *buf, uint event_len,
*description_event)
: Log_event(buf, description_event),
m_row_count(0),
m_rows_buf(0), m_rows_cur(0), m_rows_end(0)
m_table(NULL),
m_rows_buf(0), m_rows_cur(0), m_rows_end(0),
m_curr_row(NULL), m_curr_row_end(NULL),
m_key(NULL)
{
DBUG_ENTER("Rows_log_event::Rows_log_event(const char*,...)");
uint8 const common_header_len= description_event->common_header_len;
......@@ -5755,7 +5759,7 @@ Rows_log_event::Rows_log_event(const char *buf, uint event_len,
{
DBUG_PRINT("debug", ("Reading from %p", ptr_after_width));
/* if bitmap_init fails, catched in is_valid() */
/* if bitmap_init fails, caught in is_valid() */
if (likely(!bitmap_init(&m_cols_ai,
m_width <= sizeof(m_bitbuf_ai)*8 ? m_bitbuf_ai : NULL,
m_width,
......@@ -5785,6 +5789,7 @@ Rows_log_event::Rows_log_event(const char *buf, uint event_len,
m_rows_buf= (uchar*) my_malloc(data_size, MYF(MY_WME));
if (likely((bool)m_rows_buf))
{
m_curr_row= m_rows_buf;
m_rows_end= m_rows_buf + data_size;
m_rows_cur= m_rows_end;
memcpy(m_rows_buf, ptr_rows_data, data_size);
......@@ -5889,7 +5894,6 @@ int Rows_log_event::do_apply_event(RELAY_LOG_INFO const *rli)
{
DBUG_ENTER("Rows_log_event::do_apply_event(st_relay_log_info*)");
int error= 0;
uchar const *row_start= m_rows_buf;
/*
If m_table_id == ~0UL, then we have a dummy event that does not
......@@ -6049,7 +6053,9 @@ int Rows_log_event::do_apply_event(RELAY_LOG_INFO const *rli)
#endif
}
TABLE* table= const_cast<RELAY_LOG_INFO*>(rli)->m_table_map.get_table(m_table_id);
TABLE*
table=
m_table= const_cast<RELAY_LOG_INFO*>(rli)->m_table_map.get_table(m_table_id);
if (table)
{
......@@ -6096,22 +6102,41 @@ int Rows_log_event::do_apply_event(RELAY_LOG_INFO const *rli)
*/
const_cast<RELAY_LOG_INFO*>(rli)->set_flag(RELAY_LOG_INFO::IN_STMT);
error= do_before_row_operations(table);
while (error == 0 && row_start < m_rows_end)
{
uchar const *row_end= NULL;
if ((error= do_prepare_row(thd, rli, table, row_start, &row_end)))
break; // We should perform the after-row operation even in
// the case of error
if ( m_width == table->s->fields && bitmap_is_set_all(&m_cols))
set_flags(COMPLETE_ROWS_F);
/*
Set tables write and read sets.
Read_set contains all slave columns (in case we are going to fetch
a complete record from slave)
Write_set equals the m_cols bitmap sent from master but it can be
longer if slave has extra columns.
*/
DBUG_ASSERT(row_end != NULL); // cannot happen
DBUG_ASSERT(row_end <= m_rows_end);
DBUG_PRINT_BITSET("debug", "Setting table's write_set from: %s", &m_cols);
bitmap_set_all(table->read_set);
bitmap_set_all(table->write_set);
if (!get_flags(COMPLETE_ROWS_F))
bitmap_intersect(table->write_set,&m_cols);
// Do event specific preparations
error= do_before_row_operations(rli);
// row processing loop
while (error == 0 && m_curr_row < m_rows_end)
{
/* in_use can have been set to NULL in close_tables_for_reopen */
THD* old_thd= table->in_use;
if (!table->in_use)
table->in_use= thd;
error= do_exec_row(table);
error= do_exec_row(rli);
table->in_use = old_thd;
switch (error)
{
......@@ -6132,21 +6157,38 @@ int Rows_log_event::do_apply_event(RELAY_LOG_INFO const *rli)
break;
}
row_start= row_end;
}
/*
If m_curr_row_end was not set during event execution (e.g., because
of errors) we can't proceed to the next row. If the error is transient
(i.e., error==0 at this point) we must call unpack_current_row() to set
m_curr_row_end.
*/
if (!m_curr_row_end && !error)
unpack_current_row(rli);
// at this moment m_curr_row_end should be set
DBUG_ASSERT(error || m_curr_row_end != NULL);
DBUG_ASSERT(error || m_curr_row < m_curr_row_end);
DBUG_ASSERT(error || m_curr_row_end <= m_rows_end);
m_curr_row= m_curr_row_end;
} // row processing loop
DBUG_EXECUTE_IF("STOP_SLAVE_after_first_Rows_event",
const_cast<RELAY_LOG_INFO*>(rli)->abort_slave= 1;);
error= do_after_row_operations(table, error);
error= do_after_row_operations(rli, error);
if (!cache_stmt)
{
DBUG_PRINT("info", ("Marked that we need to keep log"));
thd->options|= OPTION_KEEP_LOG;
}
}
} // if (table)
/*
We need to delay this clear until the table def is no longer needed.
The table def is needed in unpack_row().
We need to delay this clear until the table def stored in m_table_def is no
longer needed. It is used in unpack_current_row().
*/
if (rli->tables_to_lock && get_flags(STMT_END_F))
const_cast<RELAY_LOG_INFO*>(rli)->clear_tables_to_lock();
......@@ -6199,7 +6241,7 @@ int Rows_log_event::do_apply_event(RELAY_LOG_INFO const *rli)
wait (reached end of last relay log and nothing gets appended
there), we timeout after one minute, and notify DBA about the
problem. When WL#2975 is implemented, just remove the member
st_relay_log_info::last_event_start_time and all its occurences.
st_relay_log_info::last_event_start_time and all its occurrences.
*/
const_cast<RELAY_LOG_INFO*>(rli)->last_event_start_time= my_time(0);
}
......@@ -6603,7 +6645,7 @@ Table_map_log_event::Table_map_log_event(THD *thd, TABLE *tbl, ulong tid,
else
m_data_size+= m_field_metadata_size + 1;
/* If malloc fails, catched in is_valid() */
/* If malloc fails, caught in is_valid() */
if ((m_memory= (uchar*) my_malloc(m_colcnt, MYF(MY_WME))))
{
m_coltype= reinterpret_cast<uchar*>(m_memory);
......@@ -6708,7 +6750,7 @@ Table_map_log_event::Table_map_log_event(const char *buf, uint event_len,
(ulong) m_tbllen, (long) (ptr_tbllen-(const uchar*)vpart),
m_colcnt, (long) (ptr_colcnt-(const uchar*)vpart)));
/* Allocate mem for all fields in one go. If fails, catched in is_valid() */
/* Allocate mem for all fields in one go. If fails, caught in is_valid() */
m_memory= (uchar*) my_multi_malloc(MYF(MY_WME),
&m_dbnam, (uint) m_dblen + 1,
&m_tblnam, (uint) m_tbllen + 1,
......@@ -7034,7 +7076,8 @@ Write_rows_log_event::Write_rows_log_event(const char *buf, uint event_len,
#endif
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
int Write_rows_log_event::do_before_row_operations(TABLE *table)
int
Write_rows_log_event::do_before_row_operations(const Slave_reporting_capability *const)
{
int error= 0;
......@@ -7056,26 +7099,26 @@ int Write_rows_log_event::do_before_row_operations(TABLE *table)
/*
Do not raise the error flag in case of hitting to an unique attribute
*/
table->file->extra(HA_EXTRA_IGNORE_DUP_KEY);
m_table->file->extra(HA_EXTRA_IGNORE_DUP_KEY);
/*
NDB specific: update from ndb master wrapped as Write_rows
*/
/*
so that the event should be applied to replace slave's row
*/
table->file->extra(HA_EXTRA_WRITE_CAN_REPLACE);
m_table->file->extra(HA_EXTRA_WRITE_CAN_REPLACE);
/*
NDB specific: if update from ndb master wrapped as Write_rows
does not find the row it's assumed idempotent binlog applying
is taking place; don't raise the error.
*/
table->file->extra(HA_EXTRA_IGNORE_NO_KEY);
m_table->file->extra(HA_EXTRA_IGNORE_NO_KEY);
/*
TODO: the cluster team (Tomas?) says that it's better if the engine knows
how many rows are going to be inserted, then it can allocate needed memory
from the start.
*/
table->file->ha_start_bulk_insert(0);
m_table->file->ha_start_bulk_insert(0);
/*
We need TIMESTAMP_NO_AUTO_SET otherwise ha_write_row() will not use fill
any TIMESTAMP column with data from the row but instead will use
......@@ -7091,47 +7134,31 @@ int Write_rows_log_event::do_before_row_operations(TABLE *table)
some cases we won't want TIMESTAMP_NO_AUTO_SET (will require some code to
analyze if explicit data is provided for slave's TIMESTAMP columns).
*/
table->timestamp_field_type= TIMESTAMP_NO_AUTO_SET;
m_table->timestamp_field_type= TIMESTAMP_NO_AUTO_SET;
return error;
}
int Write_rows_log_event::do_after_row_operations(TABLE *table, int error)
int
Write_rows_log_event::do_after_row_operations(const Slave_reporting_capability *const,
int error)
{
int local_error= 0;
table->file->extra(HA_EXTRA_NO_IGNORE_DUP_KEY);
table->file->extra(HA_EXTRA_WRITE_CANNOT_REPLACE);
m_table->file->extra(HA_EXTRA_NO_IGNORE_DUP_KEY);
m_table->file->extra(HA_EXTRA_WRITE_CANNOT_REPLACE);
/*
reseting the extra with
table->file->extra(HA_EXTRA_NO_IGNORE_NO_KEY);
fires bug#27077
todo: explain or fix
*/
if ((local_error= table->file->ha_end_bulk_insert()))
if ((local_error= m_table->file->ha_end_bulk_insert()))
{
table->file->print_error(local_error, MYF(0));
m_table->file->print_error(local_error, MYF(0));
}
return error? error : local_error;
}
int Write_rows_log_event::do_prepare_row(THD *thd_arg,
RELAY_LOG_INFO const *rli,
TABLE *table,
uchar const *const row_start,
uchar const **const row_end)
{
DBUG_ASSERT(table != NULL);
DBUG_ASSERT(row_start && row_end);
if (int error= unpack_row(rli, table, m_width, row_start, &m_cols, row_end,
&m_master_reclength, table->write_set, WRITE_ROWS_EVENT))
{
thd_arg->net.last_errno= error;
return error;
}
bitmap_copy(table->read_set, table->write_set);
return 0;
}
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
/*
Check if there are more UNIQUE keys after the given key.
......@@ -7145,135 +7172,6 @@ last_uniq_key(TABLE *table, uint keyno)
return 1;
}
/* Anonymous namespace for template functions/classes */
namespace {
/*
Smart pointer that will automatically call my_afree (a macro) when
the pointer goes out of scope. This is used so that I do not have
to remember to call my_afree() before each return. There is no
overhead associated with this, since all functions are inline.
I (Matz) would prefer to use the free function as a template
parameter, but that is not possible when the "function" is a
macro.
*/
template <class Obj>
class auto_afree_ptr
{
Obj* m_ptr;
public:
auto_afree_ptr(Obj* ptr) : m_ptr(ptr) { }
~auto_afree_ptr() { if (m_ptr) my_afree(m_ptr); }
void assign(Obj* ptr) {
/* Only to be called if it hasn't been given a value before. */
DBUG_ASSERT(m_ptr == NULL);
m_ptr= ptr;
}
Obj* get() { return m_ptr; }
};
}
/*
Copy "extra" columns from record[1] to record[0].
Copy the extra fields that are not present on the master but are
present on the slave from record[1] to record[0]. This is used
after fetching a record that are to be updated, either inside
replace_record() or as part of executing an update_row().
*/
static int
copy_extra_record_fields(TABLE *table,
size_t master_reclength,
my_ptrdiff_t master_fields)
{
DBUG_ENTER("copy_extra_record_fields(table, master_reclen, master_fields)");
DBUG_PRINT("info", ("Copying to 0x%lx "
"from field %lu at offset %lu "
"to field %d at offset %lu",
(long) table->record[0],
(ulong) master_fields, (ulong) master_reclength,
table->s->fields, table->s->reclength));
/*
Copying the extra fields of the slave that does not exist on
master into record[0] (which are basically the default values).
*/
if (table->s->fields < (uint) master_fields)
DBUG_RETURN(0);
DBUG_ASSERT(master_reclength <= table->s->reclength);
if (master_reclength < table->s->reclength)
bmove_align(table->record[0] + master_reclength,
table->record[1] + master_reclength,
table->s->reclength - master_reclength);
/*
Bit columns are special. We iterate over all the remaining
columns and copy the "extra" bits to the new record. This is
not a very good solution: it should be refactored on
opportunity.
REFACTORING SUGGESTION (Matz). Introduce a member function
similar to move_field_offset() called copy_field_offset() to
copy field values and implement it for all Field subclasses. Use
this function to copy data from the found record to the record
that are going to be inserted.
The copy_field_offset() function need to be a virtual function,
which in this case will prevent copying an entire range of
fields efficiently.
*/
{
Field **field_ptr= table->field + master_fields;
for ( ; *field_ptr ; ++field_ptr)
{
/*
Set the null bit according to the values in record[1]
*/
if ((*field_ptr)->maybe_null() &&
(*field_ptr)->is_null_in_record(reinterpret_cast<uchar*>(table->record[1])))
(*field_ptr)->set_null();
else
(*field_ptr)->set_notnull();
/*
Do the extra work for special columns.
*/
switch ((*field_ptr)->real_type())
{
default:
/* Nothing to do */
break;
case MYSQL_TYPE_BIT:
Field_bit *f= static_cast<Field_bit*>(*field_ptr);
if (f->bit_len > 0)
{
my_ptrdiff_t const offset= table->record[1] - table->record[0];
uchar const bits=
get_rec_bits(f->bit_ptr + offset, f->bit_ofs, f->bit_len);
set_rec_bits(bits, f->bit_ptr, f->bit_ofs, f->bit_len);
}
break;
}
}
}
DBUG_RETURN(0); // All OK
}
#define DBUG_PRINT_BITSET(N,FRM,BS) \
do { \
char buf[256]; \
for (uint i = 0 ; i < (BS)->n_bits ; ++i) \
buf[i] = bitmap_is_set((BS), i) ? '1' : '0'; \
buf[(BS)->n_bits] = '\0'; \
DBUG_PRINT((N), ((FRM), buf)); \
} while (0)
/**
Check if an error is a duplicate key error.
......@@ -7299,45 +7197,76 @@ is_duplicate_key_error(int errcode)
return false;
}
/**
Write the current row into event's table.
The row is located in the row buffer, pointed by @c m_curr_row member.
Number of columns of the row is stored in @c m_width member (it can be
different from the number of columns in the table to which we insert).
Bitmap @c m_cols indicates which columns are present in the row. It is assumed
that event's table is already open and pointed by @c m_table.
If the same record already exists in the table it can be either overwritten
or an error is reported depending on the value of @c overwrite flag
(error reporting not yet implemented). Note that the matching record can be
different from the row we insert if we use primary keys to identify records in
the table.
The row to be inserted can contain values only for selected columns. The
missing columns are filled with default values using @c prepare_record()
function. If a matching record is found in the table and @c overwritte is
true, the missing columns are taken from it.
@param rli Relay log info (needed for row unpacking).
@param overwrite
Shall we overwrite if the row already exists or signal
error (currently ignored).
@returns Error code on failure, 0 on success.
This method, if successful, sets @c m_curr_row_end pointer to point at the
next row in the rows buffer. This is done when unpacking the row to be
inserted.
@note If a matching record is found, it is either updated using
@c ha_update_row() or first deleted and then new record written.
*/
/*
Replace the provided record in the database.
SYNOPSIS
replace_record()
thd Thread context for writing the record.
table Table to which record should be written.
master_reclength
Offset to first column that is not present on the master,
alternatively the length of the record on the master
side.
RETURN VALUE
Error code on failure, 0 on success.
DESCRIPTION
Similar to how it is done in mysql_insert(), we first try to do
a ha_write_row() and of that fails due to duplicated keys (or
indices), we do an ha_update_row() or a ha_delete_row() instead.
*/
static int
replace_record(THD *thd, TABLE *table,
ulong const master_reclength,
uint const master_fields)
int
Rows_log_event::write_row(const RELAY_LOG_INFO *const rli,
const bool overwrite)
{
DBUG_ENTER("replace_record");
DBUG_ASSERT(table != NULL && thd != NULL);
DBUG_ENTER("write_row");
DBUG_ASSERT(m_table != NULL && thd != NULL);
TABLE *table= m_table; // pointer to event's table
int error;
int keynum;
auto_afree_ptr<char> key(NULL);
/* fill table->record[0] with default values */
if ((error= prepare_record(rli, table, m_width,
TRUE /* check if columns have def. values */)))
DBUG_RETURN(error);
/* unpack row into table->record[0] */
error= unpack_current_row(rli); // TODO: how to handle errors?
#ifndef DBUG_OFF
DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
DBUG_PRINT_BITSET("debug", "write_set = %s", table->write_set);
DBUG_PRINT_BITSET("debug", "read_set = %s", table->read_set);
#endif
/*
Try to write record. If a corresponding record already exists in the table,
we try to change it using ha_update_row() if possible. Otherwise we delete
it and repeat the whole process again.
TODO: Add safety measures against infinite looping.
*/
while ((error= table->file->ha_write_row(table->record[0])))
{
if (error == HA_ERR_LOCK_DEADLOCK || error == HA_ERR_LOCK_WAIT_TIMEOUT)
......@@ -7347,6 +7276,7 @@ replace_record(THD *thd, TABLE *table,
}
if ((keynum= table->file->get_dup_key(error)) < 0)
{
DBUG_PRINT("info",("Can't locate duplicate key (get_dup_key returns %d)",keynum));
table->file->print_error(error, MYF(0));
/*
We failed to retrieve the duplicate key
......@@ -7368,17 +7298,22 @@ replace_record(THD *thd, TABLE *table,
*/
if (table->file->ha_table_flags() & HA_DUPLICATE_POS)
{
DBUG_PRINT("info",("Locating offending record using rnd_pos()"));
error= table->file->rnd_pos(table->record[1], table->file->dup_ref);
if (error)
{
DBUG_PRINT("info",("rnd_pos() returns error %d",error));
table->file->print_error(error, MYF(0));
DBUG_RETURN(error);
}
}
else
{
DBUG_PRINT("info",("Locating offending record using index_read_idx()"));
if (table->file->extra(HA_EXTRA_FLUSH_CACHE))
{
DBUG_PRINT("info",("Error when setting HA_EXTRA_FLUSH_CACHE"));
DBUG_RETURN(my_errno);
}
......@@ -7386,8 +7321,11 @@ replace_record(THD *thd, TABLE *table,
{
key.assign(static_cast<char*>(my_alloca(table->s->max_unique_length)));
if (key.get() == NULL)
{
DBUG_PRINT("info",("Can't allocate key buffer"));
DBUG_RETURN(ENOMEM);
}
}
key_copy((uchar*)key.get(), table->record[0], table->key_info + keynum,
0);
......@@ -7397,20 +7335,33 @@ replace_record(THD *thd, TABLE *table,
HA_READ_KEY_EXACT);
if (error)
{
DBUG_PRINT("info",("index_read_idx() returns error %d",error));
table->file->print_error(error, MYF(0));
DBUG_RETURN(error);
}
}
/*
Now, table->record[1] should contain the offending row. That
Now, record[1] should contain the offending row. That
will enable us to update it or, alternatively, delete it (so
that we can insert the new row afterwards).
*/
First we copy the columns into table->record[0] that are not
present on the master from table->record[1], if there are any.
/*
If row is incomplete we will use the record found to fill
missing columns.
*/
copy_extra_record_fields(table, master_reclength, master_fields);
if (!get_flags(COMPLETE_ROWS_F))
{
restore_record(table,record[1]);
error= unpack_current_row(rli);
}
#ifndef DBUG_OFF
DBUG_PRINT("debug",("preparing for update: before and after image"));
DBUG_DUMP("record[1] (before)", table->record[1], table->s->reclength);
DBUG_DUMP("record[0] (after)", table->record[0], table->s->reclength);
#endif
/*
REPLACE is defined as either INSERT or DELETE + INSERT. If
......@@ -7430,18 +7381,32 @@ replace_record(THD *thd, TABLE *table,
if (last_uniq_key(table, keynum) &&
!table->file->referenced_by_foreign_key())
{
DBUG_PRINT("info",("Updating row using ha_update_row()"));
error=table->file->ha_update_row(table->record[1],
table->record[0]);
if (error && error != HA_ERR_RECORD_IS_THE_SAME)
table->file->print_error(error, MYF(0));
else
switch (error) {
case HA_ERR_RECORD_IS_THE_SAME:
DBUG_PRINT("info",("ignoring HA_ERR_RECORD_IS_THE_SAME error from"
" ha_update_row()"));
error= 0;
case 0:
break;
default:
DBUG_PRINT("info",("ha_update_row() returns error %d",error));
table->file->print_error(error, MYF(0));
}
DBUG_RETURN(error);
}
else
{
DBUG_PRINT("info",("Deleting offending row and trying to write new one again"));
if ((error= table->file->ha_delete_row(table->record[1])))
{
DBUG_PRINT("info",("ha_delete_row() returns error %d",error));
table->file->print_error(error, MYF(0));
DBUG_RETURN(error);
}
......@@ -7452,12 +7417,20 @@ replace_record(THD *thd, TABLE *table,
DBUG_RETURN(error);
}
int Write_rows_log_event::do_exec_row(TABLE *table)
#endif
int
Write_rows_log_event::do_exec_row(const RELAY_LOG_INFO *const rli)
{
DBUG_ASSERT(table != NULL);
int error= replace_record(thd, table, m_master_reclength, m_width);
DBUG_ASSERT(m_table != NULL);
int error= write_row(rli, TRUE /* overwrite */);
if (error && !thd->net.last_errno)
thd->net.last_errno= error;
return error;
}
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
#ifdef MYSQL_CLIENT
......@@ -7550,40 +7523,52 @@ record_compare_exit:
return result;
}
/**
Locate the current row in event's table.
/*
Find the row given by 'key', if the table has keys, or else use a table scan
to find (and fetch) the row.
The current row is pointed by @c m_curr_row. Member @c m_width tells how many
columns are there in the row (this can be differnet from the number of columns
in the table). It is assumed that event's table is already open and pointed
by @c m_table.
If the engine allows random access of the records, a combination of
position() and rnd_pos() will be used.
If a corresponding record is found in the table it is stored in
@c m_table->record[0]. Note that when record is located based on a primary
key, it is possible that the record found differs from the row being located.
@param table Pointer to table to search
@param key Pointer to key to use for search, if table has key
If no key is specified or table does not have keys, a table scan is used to
find the row. In that case the row should be complete and contain values for
all columns. However, it can still be shorter than the table, i.e. the table
can contain extra columns not present in the row. It is also possible that
the table has fewer columns than the row being located.
@pre <code>table->record[0]</code> shall contain the row to locate
and <code>key</code> shall contain a key to use for searching, if
the engine has a key.
@returns Error code on failure, 0 on success.
@post If the return value is zero, <code>table->record[1]</code>
will contain the fetched row and the internal "cursor" will refer to
the row. If the return value is non-zero,
<code>table->record[1]</code> is undefined. In either case,
<code>table->record[0]</code> is undefined.
@post In case of success @c m_table->record[0] contains the record found.
Also, the internal "cursor" of the table is positioned at the record found.
@return Zero if the row was successfully fetched into
<code>table->record[1]</code>, error code otherwise.
@note If the engine allows random access of the records, a combination of
@c position() and @c rnd_pos() will be used.
*/
static int find_and_fetch_row(TABLE *table, uchar *key)
int Rows_log_event::find_row(const RELAY_LOG_INFO *rli)
{
DBUG_ENTER("find_and_fetch_row(TABLE *table, uchar *key, uchar *record)");
DBUG_PRINT("enter", ("table: 0x%lx, key: 0x%lx record: 0x%lx",
(long) table, (long) key, (long) table->record[1]));
DBUG_ENTER("find_row");
DBUG_ASSERT(m_table && m_table->in_use != NULL);
TABLE *table= m_table;
int error;
/* unpack row - missing fields get default values */
DBUG_ASSERT(table->in_use != NULL);
// TODO: shall we check and report errors here?
prepare_record(NULL,table,m_width,FALSE /* don't check errors */);
error= unpack_current_row(rli);
#ifndef DBUG_OFF
DBUG_PRINT("info",("looking for the following record"));
DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
#endif
if ((table->file->ha_table_flags() & HA_PRIMARY_KEY_REQUIRED_FOR_POSITION) &&
table->s->primary_key < MAX_KEY)
......@@ -7606,34 +7591,48 @@ static int find_and_fetch_row(TABLE *table, uchar *key)
table->s->reclength) == 0);
*/
DBUG_PRINT("info",("locating record using primary key (position)"));
table->file->position(table->record[0]);
int error= table->file->rnd_pos(table->record[0], table->file->ref);
/*
rnd_pos() returns the record in table->record[0], so we have to
move it to table->record[1].
*/
bmove_align(table->record[1], table->record[0], table->s->reclength);
error= table->file->rnd_pos(table->record[0], table->file->ref);
if (error)
{
DBUG_PRINT("info",("rnd_pos returns error %d",error));
table->file->print_error(error, MYF(0));
}
DBUG_RETURN(error);
}
/* We need to retrieve all fields */
/* TODO: Move this out from this function to main loop */
// We can't use pisition() - try other methods.
/*
We need to retrieve all fields
TODO: Move this out from this function to main loop
*/
table->use_all_columns();
if (table->s->keys > 0)
{
int error;
DBUG_PRINT("info",("locating record using primary key (index_read)"));
/* We have a key: search the table using the index */
if (!table->file->inited && (error= table->file->ha_index_init(0, FALSE)))
{
DBUG_PRINT("info",("ha_index_init returns error %d",error));
table->file->print_error(error, MYF(0));
DBUG_RETURN(error);
}
/* Fill key data for the row */
DBUG_ASSERT(m_key);
key_copy(m_key, table->record[0], table->key_info, 0);
/*
Don't print debug messages when running valgrind since they can
trigger false warnings.
*/
#ifndef HAVE_purify
DBUG_DUMP("table->record[0]", table->record[0], table->s->reclength);
DBUG_DUMP("table->record[1]", table->record[1], table->s->reclength);
DBUG_DUMP("key data", m_key, table->key_info->key_length);
#endif
/*
......@@ -7645,9 +7644,11 @@ static int find_and_fetch_row(TABLE *table, uchar *key)
my_ptrdiff_t const pos=
table->s->null_bytes > 0 ? table->s->null_bytes - 1 : 0;
table->record[1][pos]= 0xFF;
if ((error= table->file->index_read_map(table->record[1], key, HA_WHOLE_KEY,
if ((error= table->file->index_read_map(table->record[1], m_key,
HA_WHOLE_KEY,
HA_READ_KEY_EXACT)))
{
DBUG_PRINT("info",("no record matching the key found in the table"));
table->file->print_error(error, MYF(0));
table->file->ha_index_end();
DBUG_RETURN(error);
......@@ -7658,8 +7659,8 @@ static int find_and_fetch_row(TABLE *table, uchar *key)
trigger false warnings.
*/
#ifndef HAVE_purify
DBUG_DUMP("table->record[0]", table->record[0], table->s->reclength);
DBUG_DUMP("table->record[1]", table->record[1], table->s->reclength);
DBUG_PRINT("info",("found first matching record"));
DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
#endif
/*
Below is a minor "optimization". If the key (i.e., key number
......@@ -7681,10 +7682,15 @@ static int find_and_fetch_row(TABLE *table, uchar *key)
DBUG_RETURN(0);
}
/*
In case key is not unique, we still have to iterate over records found
and find the one which is identical to the row given. The row is unpacked
in record[1] where missing columns are filled with default values.
*/
DBUG_PRINT("info",("non-unique index, scanning it to find matching record"));
while (record_compare(table))
{
int error;
/*
We need to set the null bytes to ensure that the filler bit
are all set when returning. There are storage engines that
......@@ -7702,6 +7708,7 @@ static int find_and_fetch_row(TABLE *table, uchar *key)
if ((error= table->file->index_next(table->record[1])))
{
DBUG_PRINT("info",("no record matching the given row found"));
table->file->print_error(error, MYF(0));
table->file->ha_index_end();
DBUG_RETURN(error);
......@@ -7715,23 +7722,29 @@ static int find_and_fetch_row(TABLE *table, uchar *key)
}
else
{
DBUG_PRINT("info",("locating record using table scan (rnd_next)"));
int restart_count= 0; // Number of times scanning has restarted from top
int error;
/* We don't have a key: search the table using rnd_next() */
if ((error= table->file->ha_rnd_init(1)))
return error;
{
DBUG_PRINT("info",("error initializing table scan"
" (ha_rnd_init returns %d)",error));
table->file->print_error(error, MYF(0));
DBUG_RETURN(error);
}
/* Continue until we find the right record or have made a full loop */
do
{
error= table->file->rnd_next(table->record[1]);
DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
DBUG_DUMP("record[1]", table->record[1], table->s->reclength);
switch (error) {
case 0:
DBUG_DUMP("record found", table->record[0], table->s->reclength);
case HA_ERR_RECORD_DELETED:
break;
......@@ -7741,18 +7754,25 @@ static int find_and_fetch_row(TABLE *table, uchar *key)
break;
default:
DBUG_PRINT("info", ("Failed to get next record"
" (rnd_next returns %d)",error));
table->file->print_error(error, MYF(0));
DBUG_PRINT("info", ("Record not found"));
table->file->ha_rnd_end();
DBUG_RETURN(error);
}
}
while (restart_count < 2 && record_compare(table));
/*
Note: above record_compare will take into accout all record fields
which might be incorrect in case a partial row was given in the event
*/
/*
Have to restart the scan to be able to fetch the next row.
*/
DBUG_PRINT("info", ("Record %sfound", restart_count == 2 ? "not " : ""));
if (restart_count == 2)
DBUG_PRINT("info", ("Record not found"));
table->file->ha_rnd_end();
DBUG_ASSERT(error == HA_ERR_END_OF_FILE || error == 0);
......@@ -7761,6 +7781,7 @@ static int find_and_fetch_row(TABLE *table, uchar *key)
DBUG_RETURN(0);
}
#endif
/*
......@@ -7772,9 +7793,6 @@ Delete_rows_log_event::Delete_rows_log_event(THD *thd_arg, TABLE *tbl_arg,
ulong tid, MY_BITMAP const *cols,
bool is_transactional)
: Rows_log_event(thd_arg, tbl_arg, tid, cols, is_transactional)
#ifdef HAVE_REPLICATION
,m_memory(NULL), m_key(NULL), m_after_image(NULL)
#endif
{
}
#endif /* #if !defined(MYSQL_CLIENT) */
......@@ -7786,23 +7804,18 @@ Delete_rows_log_event::Delete_rows_log_event(THD *thd_arg, TABLE *tbl_arg,
Delete_rows_log_event::Delete_rows_log_event(const char *buf, uint event_len,
const Format_description_log_event
*description_event)
#if defined(MYSQL_CLIENT)
: Rows_log_event(buf, event_len, DELETE_ROWS_EVENT, description_event)
#else
: Rows_log_event(buf, event_len, DELETE_ROWS_EVENT, description_event),
m_memory(NULL), m_key(NULL), m_after_image(NULL)
#endif
{
}
#endif
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
int Delete_rows_log_event::do_before_row_operations(TABLE *table)
{
DBUG_ASSERT(m_memory == NULL);
if ((table->file->ha_table_flags() & HA_PRIMARY_KEY_REQUIRED_FOR_POSITION) &&
table->s->primary_key < MAX_KEY)
int
Delete_rows_log_event::do_before_row_operations(const Slave_reporting_capability *const)
{
if ((m_table->file->ha_table_flags() & HA_PRIMARY_KEY_REQUIRED_FOR_POSITION) &&
m_table->s->primary_key < MAX_KEY)
{
/*
We don't need to allocate any memory for m_after_image and
......@@ -7811,82 +7824,39 @@ int Delete_rows_log_event::do_before_row_operations(TABLE *table)
return 0;
}
int error= 0;
if (table->s->keys > 0)
{
m_memory= (uchar*) my_multi_malloc(MYF(MY_WME),
&m_after_image,
(uint) table->s->reclength,
&m_key,
(uint) table->key_info->key_length,
NullS);
}
else
if (m_table->s->keys > 0)
{
m_after_image= (uchar*) my_malloc(table->s->reclength, MYF(MY_WME));
m_memory= (uchar*)m_after_image;
m_key= NULL;
}
if (!m_memory)
// Allocate buffer for key searches
m_key= (uchar*)my_malloc(m_table->key_info->key_length, MYF(MY_WME));
if (!m_key)
return HA_ERR_OUT_OF_MEM;
return error;
}
return 0;
}
int Delete_rows_log_event::do_after_row_operations(TABLE *table, int error)
int
Delete_rows_log_event::do_after_row_operations(const Slave_reporting_capability *const,
int error)
{
/*error= ToDo:find out what this should really be, this triggers close_scan in nbd, returning error?*/
table->file->ha_index_or_rnd_end();
my_free(m_memory, MYF(MY_ALLOW_ZERO_PTR)); // Free for multi_malloc
m_memory= NULL;
m_after_image= NULL;
m_table->file->ha_index_or_rnd_end();
my_free(m_key, MYF(MY_ALLOW_ZERO_PTR));
m_key= NULL;
return error;
}
int Delete_rows_log_event::do_prepare_row(THD *thd_arg,
RELAY_LOG_INFO const *rli,
TABLE *table,
uchar const *const row_start,
uchar const **const row_end)
{
DBUG_ASSERT(row_start && row_end);
if (int error= unpack_row(rli, table, m_width, row_start, &m_cols, row_end,
&m_master_reclength, table->read_set, DELETE_ROWS_EVENT))
{
thd_arg->net.last_errno= error;
return error;
}
/*
If we will access rows using the random access method, m_key will
be set to NULL, so we do not need to make a key copy in that case.
*/
if (m_key)
{
KEY *const key_info= table->key_info;
key_copy(m_key, table->record[0], key_info, 0);
}
return 0;
}
int Delete_rows_log_event::do_exec_row(TABLE *table)
int Delete_rows_log_event::do_exec_row(const RELAY_LOG_INFO *const rli)
{
int error;
DBUG_ASSERT(table != NULL);
DBUG_ASSERT(m_table != NULL);
if (!(error= find_and_fetch_row(table, m_key)))
if (!(error= find_row(rli)))
{
/*
Now we should have the right row to delete. We are using
record[0] since it is guaranteed to point to a record with the
correct value.
Delete the record found, located in record[0]
*/
error= table->file->ha_delete_row(table->record[0]);
error= m_table->file->ha_delete_row(m_table->record[0]);
}
return error;
}
......@@ -7916,10 +7886,6 @@ Update_rows_log_event::Update_rows_log_event(THD *thd_arg, TABLE *tbl_arg,
MY_BITMAP const *cols_ai,
bool is_transactional)
: Rows_log_event(thd_arg, tbl_arg, tid, cols_bi, is_transactional)
#ifdef HAVE_REPLICATION
, m_memory(NULL), m_key(NULL)
#endif
{
init(cols_ai);
}
......@@ -7929,16 +7895,13 @@ Update_rows_log_event::Update_rows_log_event(THD *thd_arg, TABLE *tbl_arg,
MY_BITMAP const *cols,
bool is_transactional)
: Rows_log_event(thd_arg, tbl_arg, tid, cols, is_transactional)
#ifdef HAVE_REPLICATION
, m_memory(NULL), m_key(NULL)
#endif
{
init(cols);
}
void Update_rows_log_event::init(MY_BITMAP const *cols)
{
/* if bitmap_init fails, catched in is_valid() */
/* if bitmap_init fails, caught in is_valid() */
if (likely(!bitmap_init(&m_cols_ai,
m_width <= sizeof(m_bitbuf_ai)*8 ? m_bitbuf_ai : NULL,
m_width,
......@@ -7971,157 +7934,87 @@ Update_rows_log_event::Update_rows_log_event(const char *buf, uint event_len,
const
Format_description_log_event
*description_event)
#if defined(MYSQL_CLIENT)
: Rows_log_event(buf, event_len, UPDATE_ROWS_EVENT, description_event)
#else
: Rows_log_event(buf, event_len, UPDATE_ROWS_EVENT, description_event),
m_memory(NULL), m_key(NULL)
#endif
{
}
#endif
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
int Update_rows_log_event::do_before_row_operations(TABLE *table)
{
DBUG_ASSERT(m_memory == NULL);
int error= 0;
if (table->s->keys > 0)
{
m_memory= (uchar*) my_multi_malloc(MYF(MY_WME),
&m_after_image,
(uint) table->s->reclength,
&m_key,
(uint) table->key_info->key_length,
NullS);
}
else
int
Update_rows_log_event::do_before_row_operations(const Slave_reporting_capability *const)
{
if (m_table->s->keys > 0)
{
m_after_image= (uchar*) my_malloc(table->s->reclength, MYF(MY_WME));
m_memory= m_after_image;
m_key= NULL;
}
if (!m_memory)
// Allocate buffer for key searches
m_key= (uchar*)my_malloc(m_table->key_info->key_length, MYF(MY_WME));
if (!m_key)
return HA_ERR_OUT_OF_MEM;
}
table->timestamp_field_type= TIMESTAMP_NO_AUTO_SET;
m_table->timestamp_field_type= TIMESTAMP_NO_AUTO_SET;
return error;
return 0;
}
int Update_rows_log_event::do_after_row_operations(TABLE *table, int error)
int
Update_rows_log_event::do_after_row_operations(const Slave_reporting_capability *const,
int error)
{
/*
error= ToDo:find out what this should really be, this triggers
close_scan in nbd, returning error?
*/
table->file->ha_index_or_rnd_end();
my_free(m_memory, MYF(MY_ALLOW_ZERO_PTR));
m_memory= NULL;
m_after_image= NULL;
/*error= ToDo:find out what this should really be, this triggers close_scan in nbd, returning error?*/
m_table->file->ha_index_or_rnd_end();
my_free(m_key, MYF(MY_ALLOW_ZERO_PTR)); // Free for multi_malloc
m_key= NULL;
return error;
}
int Update_rows_log_event::do_prepare_row(THD *thd_arg,
RELAY_LOG_INFO const *rli,
TABLE *table,
uchar const *const row_start,
uchar const **const row_end)
int
Update_rows_log_event::do_exec_row(const RELAY_LOG_INFO *const rli)
{
int error;
DBUG_ASSERT(row_start && row_end);
/*
We need to perform some juggling below since unpack_row() always
unpacks into table->record[0]. For more information, see the
comments for unpack_row().
*/
/* record[0] is the before image for the update */
if ((error= unpack_row(rli, table, m_width, row_start, &m_cols, row_end,
&m_master_reclength, table->read_set,
UPDATE_ROWS_EVENT)))
{
thd_arg->net.last_errno= error;
return error;
}
DBUG_ASSERT(m_table != NULL);
store_record(table, record[1]);
uchar const *next_start = *row_end;
/* m_after_image is the after image for the update */
if ((error= unpack_row(rli, table, m_width, next_start, &m_cols_ai, row_end,
&m_master_reclength, table->write_set,
UPDATE_ROWS_EVENT)))
{
thd_arg->net.last_errno= error;
int error= find_row(rli);
if (error)
return error;
}
bmove_align(m_after_image, table->record[0], table->s->reclength);
restore_record(table, record[1]);
/*
Don't print debug messages when running valgrind since they can
trigger false warnings.
*/
#ifndef HAVE_purify
DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
DBUG_DUMP("m_after_image", m_after_image, table->s->reclength);
#endif
/*
If we will access rows using the random access method, m_key will
be set to NULL, so we do not need to make a key copy in that case.
*/
if (m_key)
{
KEY *const key_info= table->key_info;
This is the situation after locating BI:
key_copy(m_key, table->record[0], key_info, 0);
}
===|=== before image ====|=== after image ===|===
^ ^
m_curr_row m_curr_row_end
return error;
}
BI found in the table is stored in record[0]. We copy it to record[1]
and unpack AI to record[0].
*/
int Update_rows_log_event::do_exec_row(TABLE *table)
{
DBUG_ASSERT(table != NULL);
store_record(m_table,record[1]);
int error= find_and_fetch_row(table, m_key);
if (error)
return error;
m_curr_row= m_curr_row_end;
error= unpack_current_row(rli); // this also updates m_curr_row_end
/*
We have to ensure that the new record (i.e., the after image) is
in record[0] and the old record (i.e., the before image) is in
record[1]. This since some storage engines require this (for
example, the partition engine).
Since find_and_fetch_row() puts the fetched record (i.e., the old
record) in record[1], we can keep it there. We put the new record
(i.e., the after image) into record[0], and copy the fields that
are on the slave (i.e., in record[1]) into record[0], effectively
overwriting the default values that where put there by the
unpack_row() function.
Now we have the right row to update. The old row (the one we're
looking for) is in record[1] and the new row is in record[0].
*/
bmove_align(table->record[0], m_after_image, table->s->reclength);
copy_extra_record_fields(table, m_master_reclength, m_width);
#ifndef HAVE_purify
/*
Now we have the right row to update. The old row (the one we're
looking for) is in record[1] and the new row has is in record[0].
We also have copied the original values already in the slave's
database into the after image delivered from the master.
Don't print debug messages when running valgrind since they can
trigger false warnings.
*/
error= table->file->ha_update_row(table->record[1], table->record[0]);
DBUG_PRINT("info",("Updating row in table"));
DBUG_DUMP("old record", m_table->record[1], m_table->s->reclength);
DBUG_DUMP("new values", m_table->record[0], m_table->s->reclength);
#endif
error= m_table->file->ha_update_row(m_table->record[1], m_table->record[0]);
if (error == HA_ERR_RECORD_IS_THE_SAME)
error= 0;
return error;
}
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
#ifdef MYSQL_CLIENT
......
sql/log_event.h
View file @ f8b64e17
......@@ -23,6 +23,10 @@
#include <my_bitmap.h>
#include "rpl_constants.h"
#ifndef MYSQL_CLIENT
#include "rpl_record.h"
#include "rpl_reporting.h"
#endif
#define LOG_READ_EOF -1
#define LOG_READ_BOGUS -2
......@@ -2180,7 +2184,13 @@ public:
NO_FOREIGN_KEY_CHECKS_F = (1U << 1),
/* Value of the OPTION_RELAXED_UNIQUE_CHECKS flag in thd->options */
RELAXED_UNIQUE_CHECKS_F = (1U << 2)
RELAXED_UNIQUE_CHECKS_F = (1U << 2),
/**
Indicates that rows in this event are complete, that is contain
values for all columns of the table.
*/
COMPLETE_ROWS_F = (1U << 3)
};
typedef uint16 flag_set;
......@@ -2284,7 +2294,26 @@ protected:
uchar *m_rows_cur; /* One-after the end of the data */
uchar *m_rows_end; /* One-after the end of the allocated space */
const uchar *m_curr_row; /* Start of the row being processed */
const uchar *m_curr_row_end; /* One-after the end of the current row */
flag_set m_flags; /* Flags for row-level events */
uchar *m_key; /* Buffer to keep key value during searches */
/* helper functions */
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
int find_row(const RELAY_LOG_INFO *const);
int write_row(const RELAY_LOG_INFO *const, const bool);
// Unpack the current row into m_table->record[0]
int unpack_current_row(const RELAY_LOG_INFO *const rli)
{
DBUG_ASSERT(m_table);
return ::unpack_row(rli, m_table, m_width, m_curr_row, &m_cols,
&m_curr_row_end, &m_master_reclength);
}
#endif
private:
......@@ -2309,7 +2338,8 @@ private:
The member function will return 0 if all went OK, or a non-zero
error code otherwise.
*/
virtual int do_before_row_operations(TABLE *table) = 0;
virtual
int do_before_row_operations(const Slave_reporting_capability *const log) = 0;
/*
Primitive to clean up after a sequence of row executions.
......@@ -2319,45 +2349,33 @@ private:
After doing a sequence of do_prepare_row() and do_exec_row(),
this member function should be called to clean up and release
any allocated buffers.
*/
virtual int do_after_row_operations(TABLE *table, int error) = 0;
/*
Primitive to prepare for handling one row in a row-level event.
DESCRIPTION
The member function prepares for execution of operations needed for one
row in a row-level event by reading up data from the buffer containing
the row. No specific interpretation of the data is normally done here,
since SQL thread specific data is not available: that data is made
available for the do_exec function.
A pointer to the start of the next row, or NULL if the preparation
failed. Currently, preparation cannot fail, but don't rely on this
behavior.
RETURN VALUE
Error code, if something went wrong, 0 otherwise.
The error argument, if non-zero, indicates an error which happened during
row processing before this function was called. In this case, even if
function is successful, it should return the error code given in the argument.
*/
virtual int do_prepare_row(THD*, RELAY_LOG_INFO const*, TABLE*,
uchar const *row_start,
uchar const **row_end) = 0;
virtual
int do_after_row_operations(const Slave_reporting_capability *const log,
int error) = 0;
/*
Primitive to do the actual execution necessary for a row.
DESCRIPTION
The member function will do the actual execution needed to handle a row.
The row is located at m_curr_row. When the function returns,
m_curr_row_end should point at the next row (one byte after the end
of the current row).
RETURN VALUE
0 if execution succeeded, 1 if execution failed.
*/
virtual int do_exec_row(TABLE *table) = 0;
virtual int do_exec_row(const RELAY_LOG_INFO *const rli) = 0;
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
};
friend class Old_rows_log_event;
};
/*****************************************************************************
......@@ -2407,14 +2425,9 @@ private:
#endif
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
uchar *m_memory;
uchar *m_after_image;
virtual int do_before_row_operations(TABLE *table);
virtual int do_after_row_operations(TABLE *table, int error);
virtual int do_prepare_row(THD*, RELAY_LOG_INFO const*, TABLE*,
uchar const *row_start, uchar const **row_end);
virtual int do_exec_row(TABLE *table);
virtual int do_before_row_operations(const Slave_reporting_capability *const);
virtual int do_after_row_operations(const Slave_reporting_capability *const,int);
virtual int do_exec_row(const RELAY_LOG_INFO *const);
#endif
};
......@@ -2486,15 +2499,9 @@ protected:
#endif
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
uchar *m_memory;
uchar *m_key;
uchar *m_after_image;
virtual int do_before_row_operations(TABLE *table);
virtual int do_after_row_operations(TABLE *table, int error);
virtual int do_prepare_row(THD*, RELAY_LOG_INFO const*, TABLE*,
uchar const *row_start, uchar const **row_end);
virtual int do_exec_row(TABLE *table);
virtual int do_before_row_operations(const Slave_reporting_capability *const);
virtual int do_after_row_operations(const Slave_reporting_capability *const,int);
virtual int do_exec_row(const RELAY_LOG_INFO *const);
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
};
......@@ -2557,15 +2564,9 @@ protected:
#endif
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
uchar *m_memory;
uchar *m_key;
uchar *m_after_image;
virtual int do_before_row_operations(TABLE *table);
virtual int do_after_row_operations(TABLE *table, int error);
virtual int do_prepare_row(THD*, RELAY_LOG_INFO const*, TABLE*,
uchar const *row_start, uchar const **row_end);
virtual int do_exec_row(TABLE *table);
virtual int do_before_row_operations(const Slave_reporting_capability *const);
virtual int do_after_row_operations(const Slave_reporting_capability *const,int);
virtual int do_exec_row(const RELAY_LOG_INFO *const);
#endif
};
......
sql/log_event_old.cc
View file @ f8b64e17
#include "mysql_priv.h"
#ifndef MYSQL_CLIENT
#include "rpl_rli.h"
#include "rpl_utility.h"
#endif
#include "log_event_old.h"
#include "rpl_record_old.h"
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
// Old implementation of do_apply_event()
int
Old_rows_log_event::do_apply_event(Rows_log_event *ev, const RELAY_LOG_INFO *rli)
{
DBUG_ENTER("Rows_log_event::do_apply_event(st_relay_log_info*)");
int error= 0;
THD *thd= ev->thd;
uchar const *row_start= ev->m_rows_buf;
/*
If m_table_id == ~0UL, then we have a dummy event that does not
contain any data. In that case, we just remove all tables in the
tables_to_lock list, close the thread tables, and return with
success.
*/
if (ev->m_table_id == ~0UL)
{
/*
This one is supposed to be set: just an extra check so that
nothing strange has happened.
*/
DBUG_ASSERT(ev->get_flags(Rows_log_event::STMT_END_F));
const_cast<RELAY_LOG_INFO*>(rli)->clear_tables_to_lock();
close_thread_tables(thd);
thd->clear_error();
DBUG_RETURN(0);
}
/*
'thd' has been set by exec_relay_log_event(), just before calling
do_apply_event(). We still check here to prevent future coding
errors.
*/
DBUG_ASSERT(rli->sql_thd == thd);
/*
If there is no locks taken, this is the first binrow event seen
after the table map events. We should then lock all the tables
used in the transaction and proceed with execution of the actual
event.
*/
if (!thd->lock)
{
bool need_reopen= 1; /* To execute the first lap of the loop below */
/*
lock_tables() reads the contents of thd->lex, so they must be
initialized. Contrary to in
Table_map_log_event::do_apply_event() we don't call
mysql_init_query() as that may reset the binlog format.
*/
lex_start(thd);
while ((error= lock_tables(thd, rli->tables_to_lock,
rli->tables_to_lock_count, &need_reopen)))
{
if (!need_reopen)
{
if (thd->query_error || thd->is_fatal_error)
{
/*
Error reporting borrowed from Query_log_event with many excessive
simplifications (we don't honour --slave-skip-errors)
*/
uint actual_error= thd->net.last_errno;
rli->report(ERROR_LEVEL, actual_error,
"Error '%s' in %s event: when locking tables",
(actual_error ? thd->net.last_error :
"unexpected success or fatal error"),
ev->get_type_str());
thd->is_fatal_error= 1;
}
else
{
rli->report(ERROR_LEVEL, error,
"Error in %s event: when locking tables",
ev->get_type_str());
}
const_cast<RELAY_LOG_INFO*>(rli)->clear_tables_to_lock();
DBUG_RETURN(error);
}
/*
So we need to reopen the tables.
We need to flush the pending RBR event, since it keeps a
pointer to an open table.
ALTERNATIVE SOLUTION (not implemented): Extract a pointer to
the pending RBR event and reset the table pointer after the
tables has been reopened.
NOTE: For this new scheme there should be no pending event:
need to add code to assert that is the case.
*/
thd->binlog_flush_pending_rows_event(false);
TABLE_LIST *tables= rli->tables_to_lock;
close_tables_for_reopen(thd, &tables);
uint tables_count= rli->tables_to_lock_count;
if ((error= open_tables(thd, &tables, &tables_count, 0)))
{
if (thd->query_error || thd->is_fatal_error)
{
/*
Error reporting borrowed from Query_log_event with many excessive
simplifications (we don't honour --slave-skip-errors)
*/
uint actual_error= thd->net.last_errno;
rli->report(ERROR_LEVEL, actual_error,
"Error '%s' on reopening tables",
(actual_error ? thd->net.last_error :
"unexpected success or fatal error"));
thd->query_error= 1;
}
const_cast<RELAY_LOG_INFO*>(rli)->clear_tables_to_lock();
DBUG_RETURN(error);
}
}
/*
When the open and locking succeeded, we check all tables to
ensure that they still have the correct type.
We can use a down cast here since we know that every table added
to the tables_to_lock is a RPL_TABLE_LIST.
*/
{
RPL_TABLE_LIST *ptr= rli->tables_to_lock;
for ( ; ptr ; ptr= static_cast<RPL_TABLE_LIST*>(ptr->next_global))
{
if (ptr->m_tabledef.compatible_with(rli, ptr->table))
{
mysql_unlock_tables(thd, thd->lock);
thd->lock= 0;
thd->query_error= 1;
const_cast<RELAY_LOG_INFO*>(rli)->clear_tables_to_lock();
DBUG_RETURN(Rows_log_event::ERR_BAD_TABLE_DEF);
}
}
}
/*
... and then we add all the tables to the table map and remove
them from tables to lock.
We also invalidate the query cache for all the tables, since
they will now be changed.
TODO [/Matz]: Maybe the query cache should not be invalidated
here? It might be that a table is not changed, even though it
was locked for the statement. We do know that each
Rows_log_event contain at least one row, so after processing one
Rows_log_event, we can invalidate the query cache for the
associated table.
*/
for (TABLE_LIST *ptr= rli->tables_to_lock ; ptr ; ptr= ptr->next_global)
{
const_cast<RELAY_LOG_INFO*>(rli)->m_table_map.set_table(ptr->table_id, ptr->table);
}
#ifdef HAVE_QUERY_CACHE
query_cache.invalidate_locked_for_write(rli->tables_to_lock);
#endif
}
TABLE* table= const_cast<RELAY_LOG_INFO*>(rli)->m_table_map.get_table(ev->m_table_id);
if (table)
{
/*
table == NULL means that this table should not be replicated
(this was set up by Table_map_log_event::do_apply_event()
which tested replicate-* rules).
*/
/*
It's not needed to set_time() but
1) it continues the property that "Time" in SHOW PROCESSLIST shows how
much slave is behind
2) it will be needed when we allow replication from a table with no
TIMESTAMP column to a table with one.
So we call set_time(), like in SBR. Presently it changes nothing.
*/
thd->set_time((time_t)ev->when);
/*
There are a few flags that are replicated with each row event.
Make sure to set/clear them before executing the main body of
the event.
*/
if (ev->get_flags(Rows_log_event::NO_FOREIGN_KEY_CHECKS_F))
thd->options|= OPTION_NO_FOREIGN_KEY_CHECKS;
else
thd->options&= ~OPTION_NO_FOREIGN_KEY_CHECKS;
if (ev->get_flags(Rows_log_event::RELAXED_UNIQUE_CHECKS_F))
thd->options|= OPTION_RELAXED_UNIQUE_CHECKS;
else
thd->options&= ~OPTION_RELAXED_UNIQUE_CHECKS;
/* A small test to verify that objects have consistent types */
DBUG_ASSERT(sizeof(thd->options) == sizeof(OPTION_RELAXED_UNIQUE_CHECKS));
/*
Now we are in a statement and will stay in a statement until we
see a STMT_END_F.
We set this flag here, before actually applying any rows, in
case the SQL thread is stopped and we need to detect that we're
inside a statement and halting abruptly might cause problems
when restarting.
*/
const_cast<RELAY_LOG_INFO*>(rli)->set_flag(RELAY_LOG_INFO::IN_STMT);
error= do_before_row_operations(table);
while (error == 0 && row_start < ev->m_rows_end)
{
uchar const *row_end= NULL;
if ((error= do_prepare_row(thd, rli, table, row_start, &row_end)))
break; // We should perform the after-row operation even in
// the case of error
DBUG_ASSERT(row_end != NULL); // cannot happen
DBUG_ASSERT(row_end <= ev->m_rows_end);
/* in_use can have been set to NULL in close_tables_for_reopen */
THD* old_thd= table->in_use;
if (!table->in_use)
table->in_use= thd;
error= do_exec_row(table);
table->in_use = old_thd;
switch (error)
{
/* Some recoverable errors */
case HA_ERR_RECORD_CHANGED:
case HA_ERR_KEY_NOT_FOUND: /* Idempotency support: OK if
tuple does not exist */
error= 0;
case 0:
break;
default:
rli->report(ERROR_LEVEL, thd->net.last_errno,
"Error in %s event: row application failed. %s",
ev->get_type_str(),
thd->net.last_error ? thd->net.last_error : "");
thd->query_error= 1;
break;
}
row_start= row_end;
}
DBUG_EXECUTE_IF("STOP_SLAVE_after_first_Rows_event",
const_cast<RELAY_LOG_INFO*>(rli)->abort_slave= 1;);
error= do_after_row_operations(table, error);
if (!ev->cache_stmt)
{
DBUG_PRINT("info", ("Marked that we need to keep log"));
thd->options|= OPTION_KEEP_LOG;
}
}
/*
We need to delay this clear until the table def is no longer needed.
The table def is needed in unpack_row().
*/
if (rli->tables_to_lock && ev->get_flags(Rows_log_event::STMT_END_F))
const_cast<RELAY_LOG_INFO*>(rli)->clear_tables_to_lock();
if (error)
{ /* error has occured during the transaction */
rli->report(ERROR_LEVEL, thd->net.last_errno,
"Error in %s event: error during transaction execution "
"on table %s.%s. %s",
ev->get_type_str(), table->s->db.str,
table->s->table_name.str,
thd->net.last_error ? thd->net.last_error : "");
/*
If one day we honour --skip-slave-errors in row-based replication, and
the error should be skipped, then we would clear mappings, rollback,
close tables, but the slave SQL thread would not stop and then may
assume the mapping is still available, the tables are still open...
So then we should clear mappings/rollback/close here only if this is a
STMT_END_F.
For now we code, knowing that error is not skippable and so slave SQL
thread is certainly going to stop.
rollback at the caller along with sbr.
*/
thd->reset_current_stmt_binlog_row_based();
const_cast<RELAY_LOG_INFO*>(rli)->cleanup_context(thd, error);
thd->query_error= 1;
DBUG_RETURN(error);
}
/*
This code would ideally be placed in do_update_pos() instead, but
since we have no access to table there, we do the setting of
last_event_start_time here instead.
*/
if (table && (table->s->primary_key == MAX_KEY) &&
!ev->cache_stmt &&
ev->get_flags(Rows_log_event::STMT_END_F) == Rows_log_event::RLE_NO_FLAGS)
{
/*
------------ Temporary fix until WL#2975 is implemented ---------
This event is not the last one (no STMT_END_F). If we stop now
(in case of terminate_slave_thread()), how will we restart? We
have to restart from Table_map_log_event, but as this table is
not transactional, the rows already inserted will still be
present, and idempotency is not guaranteed (no PK) so we risk
that repeating leads to double insert. So we desperately try to
continue, hope we'll eventually leave this buggy situation (by
executing the final Rows_log_event). If we are in a hopeless
wait (reached end of last relay log and nothing gets appended
there), we timeout after one minute, and notify DBA about the
problem. When WL#2975 is implemented, just remove the member
st_relay_log_info::last_event_start_time and all its occurences.
*/
const_cast<RELAY_LOG_INFO*>(rli)->last_event_start_time= my_time(0);
}
DBUG_RETURN(0);
}
#endif
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
/*
Check if there are more UNIQUE keys after the given key.
*/
static int
last_uniq_key(TABLE *table, uint keyno)
{
while (++keyno < table->s->keys)
if (table->key_info[keyno].flags & HA_NOSAME)
return 0;
return 1;
}
/*
Compares table->record[0] and table->record[1]
Returns TRUE if different.
*/
static bool record_compare(TABLE *table)
{
/*
Need to set the X bit and the filler bits in both records since
there are engines that do not set it correctly.
In addition, since MyISAM checks that one hasn't tampered with the
record, it is necessary to restore the old bytes into the record
after doing the comparison.
TODO[record format ndb]: Remove it once NDB returns correct
records. Check that the other engines also return correct records.
*/
bool result= FALSE;
uchar saved_x[2], saved_filler[2];
if (table->s->null_bytes > 0)
{
for (int i = 0 ; i < 2 ; ++i)
{
saved_x[i]= table->record[i][0];
saved_filler[i]= table->record[i][table->s->null_bytes - 1];
table->record[i][0]|= 1U;
table->record[i][table->s->null_bytes - 1]|=
256U - (1U << table->s->last_null_bit_pos);
}
}
if (table->s->blob_fields + table->s->varchar_fields == 0)
{
result= cmp_record(table,record[1]);
goto record_compare_exit;
}
/* Compare null bits */
if (memcmp(table->null_flags,
table->null_flags+table->s->rec_buff_length,
table->s->null_bytes))
{
result= TRUE; // Diff in NULL value
goto record_compare_exit;
}
/* Compare updated fields */
for (Field **ptr=table->field ; *ptr ; ptr++)
{
if ((*ptr)->cmp_binary_offset(table->s->rec_buff_length))
{
result= TRUE;
goto record_compare_exit;
}
}
record_compare_exit:
/*
Restore the saved bytes.
TODO[record format ndb]: Remove this code once NDB returns the
correct record format.
*/
if (table->s->null_bytes > 0)
{
for (int i = 0 ; i < 2 ; ++i)
{
table->record[i][0]= saved_x[i];
table->record[i][table->s->null_bytes - 1]= saved_filler[i];
}
}
return result;
}
/*
Copy "extra" columns from record[1] to record[0].
Copy the extra fields that are not present on the master but are
present on the slave from record[1] to record[0]. This is used
after fetching a record that are to be updated, either inside
replace_record() or as part of executing an update_row().
*/
static int
copy_extra_record_fields(TABLE *table,
size_t master_reclength,
my_ptrdiff_t master_fields)
{
DBUG_ENTER("copy_extra_record_fields(table, master_reclen, master_fields)");
DBUG_PRINT("info", ("Copying to 0x%lx "
"from field %lu at offset %lu "
"to field %d at offset %lu",
(long) table->record[0],
(ulong) master_fields, (ulong) master_reclength,
table->s->fields, table->s->reclength));
/*
Copying the extra fields of the slave that does not exist on
master into record[0] (which are basically the default values).
*/
if (table->s->fields < (uint) master_fields)
DBUG_RETURN(0);
DBUG_ASSERT(master_reclength <= table->s->reclength);
if (master_reclength < table->s->reclength)
bmove_align(table->record[0] + master_reclength,
table->record[1] + master_reclength,
table->s->reclength - master_reclength);
/*
Bit columns are special. We iterate over all the remaining
columns and copy the "extra" bits to the new record. This is
not a very good solution: it should be refactored on
opportunity.
REFACTORING SUGGESTION (Matz). Introduce a member function
similar to move_field_offset() called copy_field_offset() to
copy field values and implement it for all Field subclasses. Use
this function to copy data from the found record to the record
that are going to be inserted.
The copy_field_offset() function need to be a virtual function,
which in this case will prevent copying an entire range of
fields efficiently.
*/
{
Field **field_ptr= table->field + master_fields;
for ( ; *field_ptr ; ++field_ptr)
{
/*
Set the null bit according to the values in record[1]
*/
if ((*field_ptr)->maybe_null() &&
(*field_ptr)->is_null_in_record(reinterpret_cast<uchar*>(table->record[1])))
(*field_ptr)->set_null();
else
(*field_ptr)->set_notnull();
/*
Do the extra work for special columns.
*/
switch ((*field_ptr)->real_type())
{
default:
/* Nothing to do */
break;
case MYSQL_TYPE_BIT:
Field_bit *f= static_cast<Field_bit*>(*field_ptr);
if (f->bit_len > 0)
{
my_ptrdiff_t const offset= table->record[1] - table->record[0];
uchar const bits=
get_rec_bits(f->bit_ptr + offset, f->bit_ofs, f->bit_len);
set_rec_bits(bits, f->bit_ptr, f->bit_ofs, f->bit_len);
}
break;
}
}
}
DBUG_RETURN(0); // All OK
}
/*
Replace the provided record in the database.
SYNOPSIS
replace_record()
thd Thread context for writing the record.
table Table to which record should be written.
master_reclength
Offset to first column that is not present on the master,
alternatively the length of the record on the master
side.
RETURN VALUE
Error code on failure, 0 on success.
DESCRIPTION
Similar to how it is done in mysql_insert(), we first try to do
a ha_write_row() and of that fails due to duplicated keys (or
indices), we do an ha_update_row() or a ha_delete_row() instead.
*/
static int
replace_record(THD *thd, TABLE *table,
ulong const master_reclength,
uint const master_fields)
{
DBUG_ENTER("replace_record");
DBUG_ASSERT(table != NULL && thd != NULL);
int error;
int keynum;
auto_afree_ptr<char> key(NULL);
#ifndef DBUG_OFF
DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
DBUG_PRINT_BITSET("debug", "write_set = %s", table->write_set);
DBUG_PRINT_BITSET("debug", "read_set = %s", table->read_set);
#endif
while ((error= table->file->ha_write_row(table->record[0])))
{
if (error == HA_ERR_LOCK_DEADLOCK || error == HA_ERR_LOCK_WAIT_TIMEOUT)
{
table->file->print_error(error, MYF(0)); /* to check at exec_relay_log_event */
DBUG_RETURN(error);
}
if ((keynum= table->file->get_dup_key(error)) < 0)
{
table->file->print_error(error, MYF(0));
/*
We failed to retrieve the duplicate key
- either because the error was not "duplicate key" error
- or because the information which key is not available
*/
DBUG_RETURN(error);
}
/*
We need to retrieve the old row into record[1] to be able to
either update or delete the offending record. We either:
- use rnd_pos() with a row-id (available as dupp_row) to the
offending row, if that is possible (MyISAM and Blackhole), or else
- use index_read_idx() with the key that is duplicated, to
retrieve the offending row.
*/
if (table->file->ha_table_flags() & HA_DUPLICATE_POS)
{
error= table->file->rnd_pos(table->record[1], table->file->dup_ref);
if (error)
{
table->file->print_error(error, MYF(0));
DBUG_RETURN(error);
}
}
else
{
if (table->file->extra(HA_EXTRA_FLUSH_CACHE))
{
DBUG_RETURN(my_errno);
}
if (key.get() == NULL)
{
key.assign(static_cast<char*>(my_alloca(table->s->max_unique_length)));
if (key.get() == NULL)
DBUG_RETURN(ENOMEM);
}
key_copy((uchar*)key.get(), table->record[0], table->key_info + keynum,
0);
error= table->file->index_read_idx_map(table->record[1], keynum,
(const uchar*)key.get(),
HA_WHOLE_KEY,
HA_READ_KEY_EXACT);
if (error)
{
table->file->print_error(error, MYF(0));
DBUG_RETURN(error);
}
}
/*
Now, table->record[1] should contain the offending row. That
will enable us to update it or, alternatively, delete it (so
that we can insert the new row afterwards).
First we copy the columns into table->record[0] that are not
present on the master from table->record[1], if there are any.
*/
copy_extra_record_fields(table, master_reclength, master_fields);
/*
REPLACE is defined as either INSERT or DELETE + INSERT. If
possible, we can replace it with an UPDATE, but that will not
work on InnoDB if FOREIGN KEY checks are necessary.
I (Matz) am not sure of the reason for the last_uniq_key()
check as, but I'm guessing that it's something along the
following lines.
Suppose that we got the duplicate key to be a key that is not
the last unique key for the table and we perform an update:
then there might be another key for which the unique check will
fail, so we're better off just deleting the row and inserting
the correct row.
*/
if (last_uniq_key(table, keynum) &&
!table->file->referenced_by_foreign_key())
{
error=table->file->ha_update_row(table->record[1],
table->record[0]);
if (error && error != HA_ERR_RECORD_IS_THE_SAME)
table->file->print_error(error, MYF(0));
else
error= 0;
DBUG_RETURN(error);
}
else
{
if ((error= table->file->ha_delete_row(table->record[1])))
{
table->file->print_error(error, MYF(0));
DBUG_RETURN(error);
}
/* Will retry ha_write_row() with the offending row removed. */
}
}
DBUG_RETURN(error);
}
/**
Find the row given by 'key', if the table has keys, or else use a table scan
to find (and fetch) the row.
If the engine allows random access of the records, a combination of
position() and rnd_pos() will be used.
@param table Pointer to table to search
@param key Pointer to key to use for search, if table has key
@pre <code>table->record[0]</code> shall contain the row to locate
and <code>key</code> shall contain a key to use for searching, if
the engine has a key.
@post If the return value is zero, <code>table->record[1]</code>
will contain the fetched row and the internal "cursor" will refer to
the row. If the return value is non-zero,
<code>table->record[1]</code> is undefined. In either case,
<code>table->record[0]</code> is undefined.
@return Zero if the row was successfully fetched into
<code>table->record[1]</code>, error code otherwise.
*/
static int find_and_fetch_row(TABLE *table, uchar *key)
{
DBUG_ENTER("find_and_fetch_row(TABLE *table, uchar *key, uchar *record)");
DBUG_PRINT("enter", ("table: 0x%lx, key: 0x%lx record: 0x%lx",
(long) table, (long) key, (long) table->record[1]));
DBUG_ASSERT(table->in_use != NULL);
DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
if ((table->file->ha_table_flags() & HA_PRIMARY_KEY_REQUIRED_FOR_POSITION) &&
table->s->primary_key < MAX_KEY)
{
/*
Use a more efficient method to fetch the record given by
table->record[0] if the engine allows it. We first compute a
row reference using the position() member function (it will be
stored in table->file->ref) and the use rnd_pos() to position
the "cursor" (i.e., record[0] in this case) at the correct row.
TODO: Add a check that the correct record has been fetched by
comparing with the original record. Take into account that the
record on the master and slave can be of different
length. Something along these lines should work:
ADD>>> store_record(table,record[1]);
int error= table->file->rnd_pos(table->record[0], table->file->ref);
ADD>>> DBUG_ASSERT(memcmp(table->record[1], table->record[0],
table->s->reclength) == 0);
*/
table->file->position(table->record[0]);
int error= table->file->rnd_pos(table->record[0], table->file->ref);
/*
rnd_pos() returns the record in table->record[0], so we have to
move it to table->record[1].
*/
bmove_align(table->record[1], table->record[0], table->s->reclength);
DBUG_RETURN(error);
}
/* We need to retrieve all fields */
/* TODO: Move this out from this function to main loop */
table->use_all_columns();
if (table->s->keys > 0)
{
int error;
/* We have a key: search the table using the index */
if (!table->file->inited && (error= table->file->ha_index_init(0, FALSE)))
DBUG_RETURN(error);
/*
Don't print debug messages when running valgrind since they can
trigger false warnings.
*/
#ifndef HAVE_purify
DBUG_DUMP("table->record[0]", table->record[0], table->s->reclength);
DBUG_DUMP("table->record[1]", table->record[1], table->s->reclength);
#endif
/*
We need to set the null bytes to ensure that the filler bit are
all set when returning. There are storage engines that just set
the necessary bits on the bytes and don't set the filler bits
correctly.
*/
my_ptrdiff_t const pos=
table->s->null_bytes > 0 ? table->s->null_bytes - 1 : 0;
table->record[1][pos]= 0xFF;
if ((error= table->file->index_read_map(table->record[1], key, HA_WHOLE_KEY,
HA_READ_KEY_EXACT)))
{
table->file->print_error(error, MYF(0));
table->file->ha_index_end();
DBUG_RETURN(error);
}
/*
Don't print debug messages when running valgrind since they can
trigger false warnings.
*/
#ifndef HAVE_purify
DBUG_DUMP("table->record[0]", table->record[0], table->s->reclength);
DBUG_DUMP("table->record[1]", table->record[1], table->s->reclength);
#endif
/*
Below is a minor "optimization". If the key (i.e., key number
0) has the HA_NOSAME flag set, we know that we have found the
correct record (since there can be no duplicates); otherwise, we
have to compare the record with the one found to see if it is
the correct one.
CAVEAT! This behaviour is essential for the replication of,
e.g., the mysql.proc table since the correct record *shall* be
found using the primary key *only*. There shall be no
comparison of non-PK columns to decide if the correct record is
found. I can see no scenario where it would be incorrect to
chose the row to change only using a PK or an UNNI.
*/
if (table->key_info->flags & HA_NOSAME)
{
table->file->ha_index_end();
DBUG_RETURN(0);
}
while (record_compare(table))
{
int error;
/*
We need to set the null bytes to ensure that the filler bit
are all set when returning. There are storage engines that
just set the necessary bits on the bytes and don't set the
filler bits correctly.
TODO[record format ndb]: Remove this code once NDB returns the
correct record format.
*/
if (table->s->null_bytes > 0)
{
table->record[1][table->s->null_bytes - 1]|=
256U - (1U << table->s->last_null_bit_pos);
}
if ((error= table->file->index_next(table->record[1])))
{
table->file->print_error(error, MYF(0));
table->file->ha_index_end();
DBUG_RETURN(error);
}
}
/*
Have to restart the scan to be able to fetch the next row.
*/
table->file->ha_index_end();
}
else
{
int restart_count= 0; // Number of times scanning has restarted from top
int error;
/* We don't have a key: search the table using rnd_next() */
if ((error= table->file->ha_rnd_init(1)))
return error;
/* Continue until we find the right record or have made a full loop */
do
{
error= table->file->rnd_next(table->record[1]);
DBUG_DUMP("record[0]", table->record[0], table->s->reclength);
DBUG_DUMP("record[1]", table->record[1], table->s->reclength);
switch (error) {
case 0:
case HA_ERR_RECORD_DELETED:
break;
case HA_ERR_END_OF_FILE:
if (++restart_count < 2)
table->file->ha_rnd_init(1);
break;
default:
table->file->print_error(error, MYF(0));
DBUG_PRINT("info", ("Record not found"));
table->file->ha_rnd_end();
DBUG_RETURN(error);
}
}
while (restart_count < 2 && record_compare(table));
/*
Have to restart the scan to be able to fetch the next row.
*/
DBUG_PRINT("info", ("Record %sfound", restart_count == 2 ? "not " : ""));
table->file->ha_rnd_end();
DBUG_ASSERT(error == HA_ERR_END_OF_FILE || error == 0);
DBUG_RETURN(error);
}
DBUG_RETURN(0);
}
/**********************************************************
Row handling primitives for Write_rows_log_event_old
**********************************************************/
int Write_rows_log_event_old::do_before_row_operations(TABLE *table)
{
int error= 0;
/*
We are using REPLACE semantics and not INSERT IGNORE semantics
when writing rows, that is: new rows replace old rows. We need to
inform the storage engine that it should use this behaviour.
*/
/* Tell the storage engine that we are using REPLACE semantics. */
thd->lex->duplicates= DUP_REPLACE;
/*
Pretend we're executing a REPLACE command: this is needed for
InnoDB and NDB Cluster since they are not (properly) checking the
lex->duplicates flag.
*/
thd->lex->sql_command= SQLCOM_REPLACE;
/*
Do not raise the error flag in case of hitting to an unique attribute
*/
table->file->extra(HA_EXTRA_IGNORE_DUP_KEY);
/*
NDB specific: update from ndb master wrapped as Write_rows
*/
/*
so that the event should be applied to replace slave's row
*/
table->file->extra(HA_EXTRA_WRITE_CAN_REPLACE);
/*
NDB specific: if update from ndb master wrapped as Write_rows
does not find the row it's assumed idempotent binlog applying
is taking place; don't raise the error.
*/
table->file->extra(HA_EXTRA_IGNORE_NO_KEY);
/*
TODO: the cluster team (Tomas?) says that it's better if the engine knows
how many rows are going to be inserted, then it can allocate needed memory
from the start.
*/
table->file->ha_start_bulk_insert(0);
/*
We need TIMESTAMP_NO_AUTO_SET otherwise ha_write_row() will not use fill
any TIMESTAMP column with data from the row but instead will use
the event's current time.
As we replicate from TIMESTAMP to TIMESTAMP and slave has no extra
columns, we know that all TIMESTAMP columns on slave will receive explicit
data from the row, so TIMESTAMP_NO_AUTO_SET is ok.
When we allow a table without TIMESTAMP to be replicated to a table having
more columns including a TIMESTAMP column, or when we allow a TIMESTAMP
column to be replicated into a BIGINT column and the slave's table has a
TIMESTAMP column, then the slave's TIMESTAMP column will take its value
from set_time() which we called earlier (consistent with SBR). And then in
some cases we won't want TIMESTAMP_NO_AUTO_SET (will require some code to
analyze if explicit data is provided for slave's TIMESTAMP columns).
*/
table->timestamp_field_type= TIMESTAMP_NO_AUTO_SET;
return error;
}
int Write_rows_log_event_old::do_after_row_operations(TABLE *table, int error)
{
int local_error= 0;
table->file->extra(HA_EXTRA_NO_IGNORE_DUP_KEY);
table->file->extra(HA_EXTRA_WRITE_CANNOT_REPLACE);
/*
reseting the extra with
table->file->extra(HA_EXTRA_NO_IGNORE_NO_KEY);
fires bug#27077
todo: explain or fix
*/
if ((local_error= table->file->ha_end_bulk_insert()))
{
table->file->print_error(local_error, MYF(0));
}
return error? error : local_error;
}
int
Write_rows_log_event_old::do_prepare_row(THD *thd_arg,
RELAY_LOG_INFO const *rli,
......@@ -23,6 +981,65 @@ Write_rows_log_event_old::do_prepare_row(THD *thd_arg,
return error;
}
int Write_rows_log_event_old::do_exec_row(TABLE *table)
{
DBUG_ASSERT(table != NULL);
int error= replace_record(thd, table, m_master_reclength, m_width);
return error;
}
/**********************************************************
Row handling primitives for Delete_rows_log_event_old
**********************************************************/
int Delete_rows_log_event_old::do_before_row_operations(TABLE *table)
{
DBUG_ASSERT(m_memory == NULL);
if ((table->file->ha_table_flags() & HA_PRIMARY_KEY_REQUIRED_FOR_POSITION) &&
table->s->primary_key < MAX_KEY)
{
/*
We don't need to allocate any memory for m_after_image and
m_key since they are not used.
*/
return 0;
}
int error= 0;
if (table->s->keys > 0)
{
m_memory= (uchar*) my_multi_malloc(MYF(MY_WME),
&m_after_image,
(uint) table->s->reclength,
&m_key,
(uint) table->key_info->key_length,
NullS);
}
else
{
m_after_image= (uchar*) my_malloc(table->s->reclength, MYF(MY_WME));
m_memory= (uchar*)m_after_image;
m_key= NULL;
}
if (!m_memory)
return HA_ERR_OUT_OF_MEM;
return error;
}
int Delete_rows_log_event_old::do_after_row_operations(TABLE *table, int error)
{
/*error= ToDo:find out what this should really be, this triggers close_scan in nbd, returning error?*/
table->file->ha_index_or_rnd_end();
my_free(m_memory, MYF(MY_ALLOW_ZERO_PTR)); // Free for multi_malloc
m_memory= NULL;
m_after_image= NULL;
m_key= NULL;
return error;
}
int
Delete_rows_log_event_old::do_prepare_row(THD *thd_arg,
......@@ -57,6 +1074,67 @@ Delete_rows_log_event_old::do_prepare_row(THD *thd_arg,
return error;
}
int Delete_rows_log_event_old::do_exec_row(TABLE *table)
{
int error;
DBUG_ASSERT(table != NULL);
if (!(error= ::find_and_fetch_row(table, m_key)))
{
/*
Now we should have the right row to delete. We are using
record[0] since it is guaranteed to point to a record with the
correct value.
*/
error= table->file->ha_delete_row(table->record[0]);
}
return error;
}
/**********************************************************
Row handling primitives for Update_rows_log_event_old
**********************************************************/
int Update_rows_log_event_old::do_before_row_operations(TABLE *table)
{
DBUG_ASSERT(m_memory == NULL);
int error= 0;
if (table->s->keys > 0)
{
m_memory= (uchar*) my_multi_malloc(MYF(MY_WME),
&m_after_image,
(uint) table->s->reclength,
&m_key,
(uint) table->key_info->key_length,
NullS);
}
else
{
m_after_image= (uchar*) my_malloc(table->s->reclength, MYF(MY_WME));
m_memory= m_after_image;
m_key= NULL;
}
if (!m_memory)
return HA_ERR_OUT_OF_MEM;
table->timestamp_field_type= TIMESTAMP_NO_AUTO_SET;
return error;
}
int Update_rows_log_event_old::do_after_row_operations(TABLE *table, int error)
{
/*error= ToDo:find out what this should really be, this triggers close_scan in nbd, returning error?*/
table->file->ha_index_or_rnd_end();
my_free(m_memory, MYF(MY_ALLOW_ZERO_PTR));
m_memory= NULL;
m_after_image= NULL;
m_key= NULL;
return error;
}
int Update_rows_log_event_old::do_prepare_row(THD *thd_arg,
RELAY_LOG_INFO const *rli,
......@@ -101,4 +1179,41 @@ int Update_rows_log_event_old::do_prepare_row(THD *thd_arg,
return error;
}
int Update_rows_log_event_old::do_exec_row(TABLE *table)
{
DBUG_ASSERT(table != NULL);
int error= ::find_and_fetch_row(table, m_key);
if (error)
return error;
/*
We have to ensure that the new record (i.e., the after image) is
in record[0] and the old record (i.e., the before image) is in
record[1]. This since some storage engines require this (for
example, the partition engine).
Since find_and_fetch_row() puts the fetched record (i.e., the old
record) in record[1], we can keep it there. We put the new record
(i.e., the after image) into record[0], and copy the fields that
are on the slave (i.e., in record[1]) into record[0], effectively
overwriting the default values that where put there by the
unpack_row() function.
*/
bmove_align(table->record[0], m_after_image, table->s->reclength);
copy_extra_record_fields(table, m_master_reclength, m_width);
/*
Now we have the right row to update. The old row (the one we're
looking for) is in record[1] and the new row has is in record[0].
We also have copied the original values already in the slave's
database into the after image delivered from the master.
*/
error= table->file->ha_update_row(table->record[1], table->record[0]);
if (error == HA_ERR_RECORD_IS_THE_SAME)
error= 0;
return error;
}
#endif
sql/log_event_old.h
View file @ f8b64e17
......@@ -21,8 +21,89 @@
*/
class Write_rows_log_event_old : public Write_rows_log_event
class Old_rows_log_event
{
public:
virtual ~Old_rows_log_event() {}
protected:
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
int do_apply_event(Rows_log_event*,const RELAY_LOG_INFO*);
/*
Primitive to prepare for a sequence of row executions.
DESCRIPTION
Before doing a sequence of do_prepare_row() and do_exec_row()
calls, this member function should be called to prepare for the
entire sequence. Typically, this member function will allocate
space for any buffers that are needed for the two member
functions mentioned above.
RETURN VALUE
The member function will return 0 if all went OK, or a non-zero
error code otherwise.
*/
virtual int do_before_row_operations(TABLE *table) = 0;
/*
Primitive to clean up after a sequence of row executions.
DESCRIPTION
After doing a sequence of do_prepare_row() and do_exec_row(),
this member function should be called to clean up and release
any allocated buffers.
*/
virtual int do_after_row_operations(TABLE *table, int error) = 0;
/*
Primitive to prepare for handling one row in a row-level event.
DESCRIPTION
The member function prepares for execution of operations needed for one
row in a row-level event by reading up data from the buffer containing
the row. No specific interpretation of the data is normally done here,
since SQL thread specific data is not available: that data is made
available for the do_exec function.
A pointer to the start of the next row, or NULL if the preparation
failed. Currently, preparation cannot fail, but don't rely on this
behavior.
RETURN VALUE
Error code, if something went wrong, 0 otherwise.
*/
virtual int do_prepare_row(THD*, RELAY_LOG_INFO const*, TABLE*,
uchar const *row_start,
uchar const **row_end) = 0;
/*
Primitive to do the actual execution necessary for a row.
DESCRIPTION
The member function will do the actual execution needed to handle a row.
RETURN VALUE
0 if execution succeeded, 1 if execution failed.
*/
virtual int do_exec_row(TABLE *table) = 0;
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
};
class Write_rows_log_event_old
: public Write_rows_log_event, public Old_rows_log_event
{
public:
enum
{
......@@ -49,14 +130,26 @@ private:
virtual Log_event_type get_type_code() { return (Log_event_type)TYPE_CODE; }
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
// use old definition of do_apply_event()
virtual int do_apply_event(const RELAY_LOG_INFO *rli)
{ return Old_rows_log_event::do_apply_event(this,rli); }
// primitives for old version of do_apply_event()
virtual int do_before_row_operations(TABLE *table);
virtual int do_after_row_operations(TABLE *table, int error);
virtual int do_prepare_row(THD*, RELAY_LOG_INFO const*, TABLE*,
uchar const *row_start, uchar const **row_end);
virtual int do_exec_row(TABLE *table);
#endif
};
class Update_rows_log_event_old : public Update_rows_log_event
class Update_rows_log_event_old
: public Update_rows_log_event, public Old_rows_log_event
{
uchar *m_after_image, *m_memory;
public:
enum
{
......@@ -67,14 +160,16 @@ public:
#if !defined(MYSQL_CLIENT)
Update_rows_log_event_old(THD *thd, TABLE *table, ulong table_id,
MY_BITMAP const *cols, bool is_transactional)
: Update_rows_log_event(thd, table, table_id, cols, is_transactional)
: Update_rows_log_event(thd, table, table_id, cols, is_transactional),
m_after_image(NULL), m_memory(NULL)
{
}
#endif
#if defined(HAVE_REPLICATION)
Update_rows_log_event_old(const char *buf, uint event_len,
const Format_description_log_event *descr)
: Update_rows_log_event(buf, event_len, descr)
: Update_rows_log_event(buf, event_len, descr),
m_after_image(NULL), m_memory(NULL)
{
}
#endif
......@@ -83,14 +178,25 @@ private:
virtual Log_event_type get_type_code() { return (Log_event_type)TYPE_CODE; }
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
// use old definition of do_apply_event()
virtual int do_apply_event(const RELAY_LOG_INFO *rli)
{ return Old_rows_log_event::do_apply_event(this,rli); }
// primitives for old version of do_apply_event()
virtual int do_before_row_operations(TABLE *table);
virtual int do_after_row_operations(TABLE *table, int error);
virtual int do_prepare_row(THD*, RELAY_LOG_INFO const*, TABLE*,
uchar const *row_start, uchar const **row_end);
virtual int do_exec_row(TABLE *table);
#endif /* !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION) */
};
class Delete_rows_log_event_old : public Delete_rows_log_event
class Delete_rows_log_event_old
: public Delete_rows_log_event, public Old_rows_log_event
{
uchar *m_after_image, *m_memory;
public:
enum
{
......@@ -101,14 +207,16 @@ public:
#if !defined(MYSQL_CLIENT)
Delete_rows_log_event_old(THD *thd, TABLE *table, ulong table_id,
MY_BITMAP const *cols, bool is_transactional)
: Delete_rows_log_event(thd, table, table_id, cols, is_transactional)
: Delete_rows_log_event(thd, table, table_id, cols, is_transactional),
m_after_image(NULL), m_memory(NULL)
{
}
#endif
#if defined(HAVE_REPLICATION)
Delete_rows_log_event_old(const char *buf, uint event_len,
const Format_description_log_event *descr)
: Delete_rows_log_event(buf, event_len, descr)
: Delete_rows_log_event(buf, event_len, descr),
m_after_image(NULL), m_memory(NULL)
{
}
#endif
......@@ -117,8 +225,16 @@ private:
virtual Log_event_type get_type_code() { return (Log_event_type)TYPE_CODE; }
#if !defined(MYSQL_CLIENT) && defined(HAVE_REPLICATION)
// use old definition of do_apply_event()
virtual int do_apply_event(const RELAY_LOG_INFO *rli)
{ return Old_rows_log_event::do_apply_event(this,rli); }
// primitives for old version of do_apply_event()
virtual int do_before_row_operations(TABLE *table);
virtual int do_after_row_operations(TABLE *table, int error);
virtual int do_prepare_row(THD*, RELAY_LOG_INFO const*, TABLE*,
uchar const *row_start, uchar const **row_end);
virtual int do_exec_row(TABLE *table);
#endif
};
......
sql/rpl_record.cc
View file @ f8b64e17
......@@ -157,9 +157,8 @@ pack_row(TABLE *table, MY_BITMAP const* cols,
the various member functions of Field and subclasses expect to
write.
The row is assumed to only consist of the fields for which the
bitset represented by @c arr and @c bits; the other parts of the
record are left alone.
The row is assumed to only consist of the fields for which the corresponding
bit in bitset @c cols is set; the other parts of the record are left alone.
At most @c colcnt columns are read: if the table is larger than
that, the remaining fields are not filled in.
......@@ -169,15 +168,12 @@ pack_row(TABLE *table, MY_BITMAP const* cols,
@param colcnt Number of columns to read from record
@param row_data
Packed row data
@param cols Pointer to columns data to fill in
@param cols Pointer to bitset describing columns to fill in
@param row_end Pointer to variable that will hold the value of the
one-after-end position for the row
@param master_reclength
Pointer to variable that will be set to the length of the
record on the master side
@param rw_set Pointer to bitmap that holds either the read_set or the
write_set of the table
@retval 0 No error
......@@ -191,8 +187,7 @@ int
unpack_row(RELAY_LOG_INFO const *rli,
TABLE *table, uint const colcnt,
uchar const *const row_data, MY_BITMAP const *cols,
uchar const **const row_end, ulong *const master_reclength,
MY_BITMAP* const rw_set, Log_event_type const event_type)
uchar const **const row_end, ulong *const master_reclength)
{
DBUG_ENTER("unpack_row");
DBUG_ASSERT(row_data);
......@@ -202,10 +197,6 @@ unpack_row(RELAY_LOG_INFO const *rli,
uchar const *null_ptr= row_data;
uchar const *pack_ptr= row_data + master_null_byte_count;
bitmap_clear_all(rw_set);
empty_record(table);
Field **const begin_ptr = table->field;
Field **field_ptr;
Field **const end_ptr= begin_ptr + colcnt;
......@@ -265,7 +256,6 @@ unpack_row(RELAY_LOG_INFO const *rli,
#endif
}
bitmap_set_bit(rw_set, f->field_index);
null_mask <<= 1;
}
i++;
......@@ -308,29 +298,57 @@ unpack_row(RELAY_LOG_INFO const *rli,
*master_reclength = table->s->reclength;
}
/*
Set properties for remaining columns, if there are any. We let the
corresponding bit in the write_set be set, to write the value if
it was not there already. We iterate over all remaining columns,
even if there were an error, to get as many error messages as
possible. We are still able to return a pointer to the next row,
so redo that.
This generation of error messages is only relevant when inserting
new rows.
DBUG_RETURN(error);
}
/**
Fills @c table->record[0] with default values.
First @c empty_record() is called and then, additionally, fields are
initialized explicitly with a call to @c set_default().
For optimization reasons, the explicit initialization can be skipped for
first @c skip fields. This is useful if later we are going to fill these
fields from other source (e.g. from a Rows replication event).
If @c check is true, fields are explicitly initialized only if they have
default value or can be NULL. Otherwise error is reported.
@param log Used to report errors.
@param table Table whose record[0] buffer is prepared.
@param skip Number of columns for which default value initialization
should be skipped.
@param check Indicates if errors should be checked when setting default
values.
@returns 0 on success.
*/
for ( ; *field_ptr ; ++field_ptr)
int prepare_record(const Slave_reporting_capability *const log,
TABLE *const table,
const uint skip, const bool check)
{
DBUG_ENTER("prepare_record");
int error= 0;
empty_record(table);
if (skip >= table->s->fields) // nothing to do
DBUG_RETURN(0);
/* Explicit initialization of fields */
for (Field **field_ptr= table->field+skip ; *field_ptr ; ++field_ptr)
{
uint32 const mask= NOT_NULL_FLAG | NO_DEFAULT_VALUE_FLAG;
Field *const f= *field_ptr;
if (event_type == WRITE_ROWS_EVENT &&
((*field_ptr)->flags & mask) == mask)
if (check && ((f->flags & mask) == mask))
{
rli->report(ERROR_LEVEL, ER_NO_DEFAULT_FOR_FIELD,
DBUG_ASSERT(log);
log->report(ERROR_LEVEL, ER_NO_DEFAULT_FOR_FIELD,
"Field `%s` of table `%s`.`%s` "
"has no default value and cannot be NULL",
(*field_ptr)->field_name, table->s->db.str,
f->field_name, table->s->db.str,
table->s->table_name.str);
error = ER_NO_DEFAULT_FOR_FIELD;
}
......
sql/rpl_record.h
View file @ f8b64e17
......@@ -16,6 +16,8 @@
#ifndef RPL_RECORD_H
#define RPL_RECORD_H
#include <rpl_reporting.h>
#if !defined(MYSQL_CLIENT)
size_t pack_row(TABLE* table, MY_BITMAP const* cols,
uchar *row_data, const uchar *data);
......@@ -25,9 +27,11 @@ size_t pack_row(TABLE* table, MY_BITMAP const* cols,
int unpack_row(RELAY_LOG_INFO const *rli,
TABLE *table, uint const colcnt,
uchar const *const row_data, MY_BITMAP const *cols,
uchar const **const row_end, ulong *const master_reclength,
MY_BITMAP* const rw_set,
Log_event_type const event_type);
uchar const **const row_end, ulong *const master_reclength);
// Fill table's record[0] with default values.
int prepare_record(const Slave_reporting_capability *const, TABLE *const,
const uint =0, const bool =FALSE);
#endif
#endif
sql/rpl_utility.cc
View file @ f8b64e17
......@@ -24,7 +24,7 @@
This function returns the field size in raw bytes based on the type
and the encoded field data from the master's raw data.
*/
uint32 table_def::calc_field_size(uint col, uchar *master_data)
uint32 table_def::calc_field_size(uint col, uchar *master_data) const
{
uint32 length;
......
sql/rpl_utility.h
View file @ f8b64e17
......@@ -218,7 +218,7 @@ public:
WL#3915) or needs to advance the pointer for the fields in the raw
data from the master to a specific column.
*/
uint32 calc_field_size(uint col, uchar *master_data);
uint32 calc_field_size(uint col, uchar *master_data) const;
/**
Decide if the table definition is compatible with a table.
......@@ -258,4 +258,44 @@ struct RPL_TABLE_LIST
table_def m_tabledef;
};
/* Anonymous namespace for template functions/classes */
namespace {
/*
Smart pointer that will automatically call my_afree (a macro) when
the pointer goes out of scope. This is used so that I do not have
to remember to call my_afree() before each return. There is no
overhead associated with this, since all functions are inline.
I (Matz) would prefer to use the free function as a template
parameter, but that is not possible when the "function" is a
macro.
*/
template <class Obj>
class auto_afree_ptr
{
Obj* m_ptr;
public:
auto_afree_ptr(Obj* ptr) : m_ptr(ptr) { }
~auto_afree_ptr() { if (m_ptr) my_afree(m_ptr); }
void assign(Obj* ptr) {
/* Only to be called if it hasn't been given a value before. */
DBUG_ASSERT(m_ptr == NULL);
m_ptr= ptr;
}
Obj* get() { return m_ptr; }
};
}
#define DBUG_PRINT_BITSET(N,FRM,BS) \
do { \
char buf[256]; \
for (uint i = 0 ; i < (BS)->n_bits ; ++i) \
buf[i] = bitmap_is_set((BS), i) ? '1' : '0'; \
buf[(BS)->n_bits] = '\0'; \
DBUG_PRINT((N), ((FRM), buf)); \
} while (0)
#endif /* RPL_UTILITY_H */
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7