Chapter 12. SQL Statement Syntax

Table of Contents

12.1. Data Definition Statements
12.1.1. ALTER DATABASE Syntax
12.1.2. ALTER EVENT Syntax
12.1.3. ALTER FUNCTION Syntax
12.1.4. ALTER PROCEDURE Syntax
12.1.5. ALTER SERVER Syntax
12.1.6. ALTER TABLE Syntax
12.1.7. ALTER VIEW Syntax
12.1.8. CREATE DATABASE Syntax
12.1.9. CREATE EVENT Syntax
12.1.10. The CREATE FUNCTION Statement
12.1.11. CREATE INDEX Syntax
12.1.12. CREATE PROCEDURE and CREATE FUNCTION Syntax
12.1.13. CREATE SERVER Syntax
12.1.14. CREATE TABLE Syntax
12.1.15. CREATE TRIGGER Syntax
12.1.16. CREATE VIEW Syntax
12.1.17. DROP DATABASE Syntax
12.1.18. DROP EVENT Syntax
12.1.19. DROP FUNCTION Syntax
12.1.20. DROP INDEX Syntax
12.1.21. DROP PROCEDURE and DROP FUNCTION Syntax
12.1.22. DROP SERVER Syntax
12.1.23. DROP TABLE Syntax
12.1.24. DROP TRIGGER Syntax
12.1.25. DROP VIEW Syntax
12.1.26. RENAME TABLE Syntax
12.1.27. TRUNCATE TABLE Syntax
12.2. Data Manipulation Statements
12.2.1. CALL Syntax
12.2.2. DELETE Syntax
12.2.3. DO Syntax
12.2.4. HANDLER Syntax
12.2.5. INSERT Syntax
12.2.6. LOAD DATA INFILE Syntax
12.2.7. LOAD XML Syntax
12.2.8. REPLACE Syntax
12.2.9. SELECT Syntax
12.2.10. Subquery Syntax
12.2.11. UPDATE Syntax
12.3. MySQL Transactional and Locking Statements
12.3.1. START TRANSACTION, COMMIT, and ROLLBACK Syntax
12.3.2. Statements That Cannot Be Rolled Back
12.3.3. Statements That Cause an Implicit Commit
12.3.4. SAVEPOINT and ROLLBACK TO SAVEPOINT Syntax
12.3.5. LOCK TABLES and UNLOCK TABLES Syntax
12.3.6. SET TRANSACTION Syntax
12.3.7. XA Transactions
12.4. Database Administration Statements
12.4.1. Account Management Statements
12.4.2. Table Maintenance Statements
12.4.3. Plugin and User-Defined Function Statements
12.4.4. SET Syntax
12.4.5. SHOW Syntax
12.4.6. Other Administrative Statements
12.5. Replication Statements
12.5.1. SQL Statements for Controlling Master Servers
12.5.2. SQL Statements for Controlling Slave Servers
12.6. SQL Syntax for Prepared Statements
12.6.1. PREPARE Syntax
12.6.2. EXECUTE Syntax
12.6.3. DEALLOCATE PREPARE Syntax
12.6.4. Automatic Prepared Statement Repreparation
12.7. MySQL Compound-Statement Syntax
12.7.1. BEGIN ... END Compound Statement Syntax
12.7.2. DECLARE Syntax
12.7.3. Variables in Stored Programs
12.7.4. Conditions and Handlers
12.7.5. Cursors
12.7.6. Flow Control Constructs
12.7.7. RETURN Syntax
12.7.8. SIGNAL and RESIGNAL
12.8. MySQL Utility Statements
12.8.1. DESCRIBE Syntax
12.8.2. EXPLAIN Syntax
12.8.3. HELP Syntax
12.8.4. USE Syntax

This chapter describes the syntax for the SQL statements supported by MySQL.

12.1. Data Definition Statements

12.1.1. ALTER DATABASE Syntax

ALTER {DATABASE | SCHEMA} [db_name]
    alter_specification ...
ALTER {DATABASE | SCHEMA} db_name
    UPGRADE DATA DIRECTORY NAME

alter_specification:
    [DEFAULT] CHARACTER SET [=] charset_name
  | [DEFAULT] COLLATE [=] collation_name

ALTER DATABASE enables you to change the overall characteristics of a database. These characteristics are stored in the db.opt file in the database directory. To use ALTER DATABASE, you need the ALTER privilege on the database. ALTER SCHEMA is a synonym for ALTER DATABASE.

The CHARACTER SET clause changes the default database character set. The COLLATE clause changes the default database collation. Section 9.1, “Character Set Support”, discusses character set and collation names.

You can see what character sets and collations are available using, respectively, the SHOW CHARACTER SET and SHOW COLLATION statements. See Section 12.4.5.4, “SHOW CHARACTER SET Syntax”, and Section 12.4.5.5, “SHOW COLLATION Syntax”, for more information.

The database name can be omitted from the first syntax, in which case the statement applies to the default database.

The syntax that includes the UPGRADE DATA DIRECTORY NAME clause updates the name of the directory associated with the database to use the encoding implemented in MySQL 5.1 for mapping database names to database directory names (see Section 8.2.3, “Mapping of Identifiers to File Names”). This clause is for use under these conditions:

  • It is intended when upgrading MySQL to 5.1 or later from older versions.

  • It is intended to update a database directory name to the current encoding format if the name contains special characters that need encoding.

  • The statement is used by mysqlcheck (as invoked by mysql_upgrade).

For example,if a database in MySQL 5.0 has a name of a-b-c, the name contains instance of the ‘-’ character. In 5.0, the database directory is also named a-b-c, which is not necessarily safe for all file systems. In MySQL 5.1 and up, the same database name is encoded as a@002db@002dc to produce a file system-neutral directory name.

When a MySQL installation is upgraded to MySQL 5.1 or later from an older version,the server displays a name such as a-b-c (which is in the old format) as #mysql50#a-b-c, and you must refer to the name using the #mysql50# prefix. Use UPGRADE DATA DIRECTORY NAME in this case to explicitly tell the server to re-encode the database directory name to the current encoding format:

ALTER DATABASE `#mysql50#a-b-c` UPGRADE DATA DIRECTORY NAME;

After executing this statement, you can refer to the database as a-b-c without the special #mysql50# prefix.

12.1.2. ALTER EVENT Syntax

ALTER
    [DEFINER = { user | CURRENT_USER }]
    EVENT event_name
    [ON SCHEDULE schedule]
    [ON COMPLETION [NOT] PRESERVE]
    [RENAME TO new_event_name]
    [ENABLE | DISABLE | DISABLE ON SLAVE]
    [COMMENT 'comment']
    [DO event_body]

The ALTER EVENT statement is used to change one or more of the characteristics of an existing event without the need to drop and recreate it. The syntax for each of the DEFINER, ON SCHEDULE, ON COMPLETION, COMMENT, ENABLE / DISABLE, and DO clauses is exactly the same as when used with CREATE EVENT. (See Section 12.1.9, “CREATE EVENT Syntax”.)

Any user can alter an event defined on a database for which that user has the EVENT privilege. When a user executes a successful ALTER EVENT statement, that user becomes the definer for the affected event.

ALTER EVENT works only with an existing event:

mysql> ALTER EVENT no_such_event 
     >     ON SCHEDULE 
     >       EVERY '2:3' DAY_HOUR;
ERROR 1517 (HY000): Unknown event 'no_such_event'

In each of the following examples, assume that the event named myevent is defined as shown here:

CREATE EVENT myevent
    ON SCHEDULE
      EVERY 6 HOUR
    COMMENT 'A sample comment.'
    DO
      UPDATE myschema.mytable SET mycol = mycol + 1;

The following statement changes the schedule for myevent from once every six hours starting immediately to once every twelve hours, starting four hours from the time the statement is run:

ALTER EVENT myevent
    ON SCHEDULE
      EVERY 12 HOUR
    STARTS CURRENT_TIMESTAMP + INTERVAL 4 HOUR;

It is possible to change multiple characteristics of an event in a single statement. This example changes the SQL statement executed by myevent to one that deletes all records from mytable; it also changes the schedule for the event such that it executes once, one day after this ALTER EVENT statement is run.

ALTER TABLE myevent
    ON SCHEDULE
      AT CURRENT_TIMESTAMP + INTERVAL 1 DAY
    DO
      TRUNCATE TABLE myschema.mytable;

It is necessary to include only those options in an ALTER EVENT statement which correspond to characteristics that you actually wish to change; options which are omitted retain their existing values. This includes any default values for CREATE EVENT such as ENABLE.

To disable myevent, use this ALTER EVENT statement:

ALTER EVENT myevent
    DISABLE;

The ON SCHEDULE clause may use expressions involving built-in MySQL functions and user variables to obtain any of the timestamp or interval values which it contains. You may not use stored routines or user-defined functions in such expressions, nor may you use any table references; however, you may use SELECT FROM DUAL. This is true for both ALTER EVENT and CREATE EVENT statements. References to stored routines, user-defined functions, and tables in such cases are specifically not permitted, and fail with an error (see Bug#22830).

An ALTER EVENT statement that contains another ALTER EVENT statement in its DO clause appears to succeed; however, when the server attempts to execute the resulting scheduled event, the execution fails with an error.

To rename an event, use the ALTER EVENT statement's RENAME TO clause. This statement renames the event myevent to yourevent:

ALTER EVENT myevent
    RENAME TO yourevent;

You can also move an event to a different database using ALTER EVENT ... RENAME TO ... and db_name.event_name notation, as shown here:

ALTER EVENT olddb.myevent
    RENAME TO newdb.myevent;

To execute the previous statement, the user executing it must have the EVENT privilege on both the olddb and newdb databases.

Note

There is no RENAME EVENT statement.

A third value may also appear in place of ENABLED or DISABLED; DISABLE ON SLAVE is used on a replication slave to indicate an event which was created on the master and replicated to the slave, but which is not executed on the slave. Normally, DISABLE ON SLAVE is set automatically as required; however, there are some circumstances under which you may want or need to change it manually. See Section 17.4.1.8, “Replication of Invoked Features”, for more information.

12.1.3. ALTER FUNCTION Syntax

ALTER FUNCTION func_name [characteristic ...]

characteristic:
    { CONTAINS SQL | NO SQL | READS SQL DATA | MODIFIES SQL DATA }
  | SQL SECURITY { DEFINER | INVOKER }
  | COMMENT 'string'

This statement can be used to change the characteristics of a stored function. More than one change may be specified in an ALTER FUNCTION statement. However, you cannot change the parameters or body of a stored function using this statement; to make such changes, you must drop and re-create the function using DROP FUNCTION and CREATE FUNCTION.

You must have the ALTER ROUTINE privilege for the function. (That privilege is granted automatically to the function creator.) If binary logging is enabled, the ALTER FUNCTION statement might also require the SUPER privilege, as described in Section 19.7, “Binary Logging of Stored Programs”.

12.1.4. ALTER PROCEDURE Syntax

ALTER PROCEDURE proc_name [characteristic ...]

characteristic:
    COMMENT 'string'
  | { CONTAINS SQL | NO SQL | READS SQL DATA | MODIFIES SQL DATA }
  | SQL SECURITY { DEFINER | INVOKER }

This statement can be used to change the characteristics of a stored procedure. More than one change may be specified in an ALTER PROCEDURE statement. However, you cannot change the parameters or body of a stored procedure using this statement; to make such changes, you must drop and re-create the procedure using DROP PROCEDURE and CREATE PROCEDURE.

You must have the ALTER ROUTINE privilege for the procedure. By default, that privilege is granted automatically to the procedure creator. This behavior can be changed by disabling the automatic_sp_privileges system variable. See Section 19.2.2, “Stored Routines and MySQL Privileges”.

12.1.5. ALTER SERVER Syntax

ALTER SERVER  server_name
    OPTIONS (option [, option] ...)

Alters the server information for server_name, adjusting the specified options as per the CREATE SERVER statement. See Section 12.1.13, “CREATE SERVER Syntax”. The corresponding fields in the mysql.servers table are updated accordingly. This statement requires the SUPER privilege.

For example, to update the USER option:

ALTER SERVER s OPTIONS (USER 'sally');

ALTER SERVER does not cause an automatic commit.

12.1.6. ALTER TABLE Syntax

ALTER [IGNORE] TABLE tbl_name
    alter_specification [, alter_specification] ...
    [partition_options]

ALTER [IGNORE] TABLE tbl_name
    partition_options

alter_specification:
    table_options
  | ADD [COLUMN] col_name column_definition
        [FIRST | AFTER col_name ]
  | ADD [COLUMN] (col_name column_definition,...)
  | ADD {INDEX|KEY} [index_name]
        [index_type] (index_col_name,...) [index_option] ...
  | ADD [CONSTRAINT [symbol]] PRIMARY KEY
        [index_type] (index_col_name,...) [index_option] ...
  | ADD [CONSTRAINT [symbol]]
        UNIQUE [INDEX|KEY] [index_name]
        [index_type] (index_col_name,...) [index_option] ...
  | ADD FULLTEXT [INDEX|KEY] [index_name]
        (index_col_name,...) [index_option] ...
  | ADD SPATIAL [INDEX|KEY] [index_name]
        (index_col_name,...) [index_option] ...
  | ADD [CONSTRAINT [symbol]]
        FOREIGN KEY [index_name] (index_col_name,...)
        reference_definition
  | ALTER [COLUMN] col_name {SET DEFAULT literal | DROP DEFAULT}
  | CHANGE [COLUMN] old_col_name new_col_name column_definition
        [FIRST|AFTER col_name]
  | MODIFY [COLUMN] col_name column_definition
        [FIRST | AFTER col_name]
  | DROP [COLUMN] col_name
  | DROP PRIMARY KEY
  | DROP {INDEX|KEY} index_name
  | DROP FOREIGN KEY fk_symbol
  | DISABLE KEYS
  | ENABLE KEYS
  | RENAME [TO] new_tbl_name
  | ORDER BY col_name [, col_name] ...
  | CONVERT TO CHARACTER SET charset_name [COLLATE collation_name]
  | [DEFAULT] CHARACTER SET [=] charset_name [COLLATE [=] collation_name]
  | DISCARD TABLESPACE
  | IMPORT TABLESPACE
  | ADD PARTITION (partition_definition)
  | DROP PARTITION partition_names
  | TRUNCATE PARTITION {partition_names | ALL }
  | COALESCE PARTITION number
  | REORGANIZE PARTITION partition_names INTO (partition_definitions)
  | ANALYZE PARTITION  {partition_names | ALL }
  | CHECK PARTITION  {partition_names | ALL }
  | OPTIMIZE PARTITION  {partition_names | ALL }
  | REBUILD PARTITION  {partition_names | ALL }
  | REPAIR PARTITION  {partition_names | ALL }
  | REMOVE PARTITIONING

index_col_name:
    col_name [(length)] [ASC | DESC]

index_type:
    USING {BTREE | HASH}

index_option:
    KEY_BLOCK_SIZE [=] value
  | index_type
  | WITH PARSER parser_name
  | COMMENT 'string'

table_options:
    table_option [[,] table_option] ...  (see CREATE TABLE options)

partition_options:
    (see CREATE TABLE options)

ALTER TABLE enables you to change the structure of an existing table. For example, you can add or delete columns, create or destroy indexes, change the type of existing columns, or rename columns or the table itself. You can also change the comment for the table and type of the table.

The syntax for many of the permissible alterations is similar to clauses of the CREATE TABLE statement. See Section 12.1.14, “CREATE TABLE Syntax”, for more information.

Some operations may result in warnings if attempted on a table for which the storage engine does not support the operation. These warnings can be displayed with SHOW WARNINGS. See Section 12.4.5.41, “SHOW WARNINGS Syntax”.

In most cases, ALTER TABLE makes a temporary copy of the original table. MySQL incorporates the alteration into the copy, then deletes the original table and renames the new one. While ALTER TABLE is executing, the original table is readable by other sessions. Updates and writes to the table are stalled until the new table is ready, and then are automatically redirected to the new table without any failed updates. The temporary table is created in the database directory of the new table. This can differ from the database directory of the original table for ALTER TABLE operations that rename the table to a different database.

For some operations, a “fastALTER TABLE is possible that does not require a temporary table:

  • Alterations that modify only table metadata and not table data can be made immediately by altering the table's .frm file and not touching table contents. The following changes are fast alterations that can be made this way:

    • Renaming a column.

    • Changing the default value of a column.

    • Changing the definition of an ENUM or SET column by adding new enumeration or set members to the end of the list of valid member values. (Adding members in the middle of the list causes renumbering of existing members, which requires a table copy.)

  • For ALTER TABLE tbl_name RENAME TO new_tbl_name without any other options, MySQL simply renames any files that correspond to the table tbl_name. (You can also use the RENAME TABLE statement to rename tables. See Section 12.1.26, “RENAME TABLE Syntax”.) Any privileges granted specifically for the renamed table are not migrated to the new name. They must be changed manually.

  • ALTER TABLE ... ADD PARTITION creates no temporary table. ADD or DROP operations for RANGE or LIST partitions are immediate operations or nearly so. ADD or COALESCE operations for HASH or KEY partitions copy data between changed partitions; unless LINEAR HASH or LINEAR KEY was used, this is much the same as creating a new table (although the operation is done partition by partition). REORGANIZE operations copy only changed partitions and do not touch unchanged ones.

In other cases, MySQL creates a temporary table, even if the data wouldn't strictly need to be copied. For MyISAM tables, you can speed up the index re-creation operation (which is the slowest part of the alteration process) by setting the myisam_sort_buffer_size system variable to a high value.

You can force an ALTER TABLE operation that would otherwise require a table copy to use the temporary table method (as supported in MySQL 5.0) by setting old-alter-table to ON.

For information on troubleshooting ALTER TABLE, see Section C.5.7.1, “Problems with ALTER TABLE.

  • To use ALTER TABLE, you need ALTER, INSERT, and CREATE privileges for the table.

  • IGNORE is a MySQL extension to standard SQL. It controls how ALTER TABLE works if there are duplicates on unique keys in the new table or if warnings occur when strict mode is enabled. If IGNORE is not specified, the copy is aborted and rolled back if duplicate-key errors occur. If IGNORE is specified, only the first row is used of rows with duplicates on a unique key, The other conflicting rows are deleted. Incorrect values are truncated to the closest matching acceptable value.

  • table_option signifies a table option of the kind that can be used in the CREATE TABLE statement, such as ENGINE, AUTO_INCREMENT, or AVG_ROW_LENGTH. (Section 12.1.14, “CREATE TABLE Syntax”, lists all table options.) However, ALTER TABLE ignores the DATA DIRECTORY and INDEX DIRECTORY table options.

    For example, to convert a table to be an InnoDB table, use this statement:

    ALTER TABLE t1 ENGINE = InnoDB;
    

    The outcome of attempting to change a table's storage engine is affected by whether the desired storage engine is available and the setting of the NO_ENGINE_SUBSTITUTION SQL mode, as described in Section 5.1.7, “Server SQL Modes”.

    To prevent inadvertent loss of data, ALTER TABLE cannot be used to change the storage engine of a table to MERGE or BLACKHOLE.

    To change the value of the AUTO_INCREMENT counter to be used for new rows, do this:

    ALTER TABLE t2 AUTO_INCREMENT = value;
    

    You cannot reset the counter to a value less than or equal to any that have already been used. For MyISAM, if the value is less than or equal to the maximum value currently in the AUTO_INCREMENT column, the value is reset to the current maximum plus one. For InnoDB, if the value is less than the current maximum value in the column, no error occurs and the current sequence value is not changed.

  • You can issue multiple ADD, ALTER, DROP, and CHANGE clauses in a single ALTER TABLE statement, separated by commas. This is a MySQL extension to standard SQL, which permits only one of each clause per ALTER TABLE statement. For example, to drop multiple columns in a single statement, do this:

    ALTER TABLE t2 DROP COLUMN c, DROP COLUMN d;
    
  • CHANGE col_name, DROP col_name, and DROP INDEX are MySQL extensions to standard SQL.

  • MODIFY is an Oracle extension to ALTER TABLE.

  • The word COLUMN is optional and can be omitted.

  • column_definition clauses use the same syntax for ADD and CHANGE as for CREATE TABLE. See Section 12.1.14, “CREATE TABLE Syntax”.

  • You can rename a column using a CHANGE old_col_name new_col_name column_definition clause. To do so, specify the old and new column names and the definition that the column currently has. For example, to rename an INTEGER column from a to b, you can do this:

    ALTER TABLE t1 CHANGE a b INTEGER;
    

    If you want to change a column's type but not the name, CHANGE syntax still requires an old and new column name, even if they are the same. For example:

    ALTER TABLE t1 CHANGE b b BIGINT NOT NULL;
    

    You can also use MODIFY to change a column's type without renaming it:

    ALTER TABLE t1 MODIFY b BIGINT NOT NULL;
    

    When you use CHANGE or MODIFY, column_definition must include the data type and all attributes that should apply to the new column, other than index attributes such as PRIMARY KEY or UNIQUE. Attributes present in the original definition but not specified for the new definition are not carried forward. Suppose that a column col1 is defined as INT UNSIGNED DEFAULT 1 COMMENT 'my column' and you modify the column as follows:

    ALTER TABLE t1 MODIFY col1 BIGINT;
    

    The resulting column will be defined as BIGINT, but will not include the attributes UNSIGNED DEFAULT 1 COMMENT 'my column'. To retain them, the statement should be:

    ALTER TABLE t1 MODIFY col1 BIGINT UNSIGNED DEFAULT 1 COMMENT 'my column';
    

  • When you change a data type using CHANGE or MODIFY, MySQL tries to convert existing column values to the new type as well as possible.

    Warning

    This conversion may result in alteration of data. For example, if you shorten a string column, values may be truncated. To prevent the operation from succeeding if conversions to the new data type would result in loss of data, enable strict SQL mode before using ALTER TABLE (see Section 5.1.7, “Server SQL Modes”).

  • To add a column at a specific position within a table row, use FIRST or AFTER col_name. The default is to add the column last. You can also use FIRST and AFTER in CHANGE or MODIFY operations to reorder columns within a table.

  • ALTER ... SET DEFAULT or ALTER ... DROP DEFAULT specify a new default value for a column or remove the old default value, respectively. If the old default is removed and the column can be NULL, the new default is NULL. If the column cannot be NULL, MySQL assigns a default value as described in Section 10.1.4, “Data Type Default Values”.

  • DROP INDEX removes an index. This is a MySQL extension to standard SQL. See Section 12.1.20, “DROP INDEX Syntax”. If you are unsure of the index name, use SHOW INDEX FROM tbl_name.

  • If columns are dropped from a table, the columns are also removed from any index of which they are a part. If all columns that make up an index are dropped, the index is dropped as well. If you use CHANGE or MODIFY to shorten a column for which an index exists on the column, and the resulting column length is less than the index length, MySQL shortens the index automatically.

  • If a table contains only one column, the column cannot be dropped. If what you intend is to remove the table, use DROP TABLE instead.

  • DROP PRIMARY KEY drops the primary key. If there is no primary key, an error occurs.

    If you add a UNIQUE INDEX or PRIMARY KEY to a table, it is stored before any nonunique index so that MySQL can detect duplicate keys as early as possible.

  • Some storage engines permit you to specify an index type when creating an index. The syntax for the index_type specifier is USING type_name. For details about USING, see Section 12.1.11, “CREATE INDEX Syntax”. The preferred position is after the column list. Use of the option before the column list will no longer be recognized in a future MySQL release.

    index_option values specify additional options for an index. USING is one such option. For details about permissible index_option values, see Section 12.1.11, “CREATE INDEX Syntax”.

  • After an ALTER TABLE statement, it may be necessary to run ANALYZE TABLE to update index cardinality information. See Section 12.4.5.23, “SHOW INDEX Syntax”.

  • ORDER BY enables you to create the new table with the rows in a specific order. Note that the table does not remain in this order after inserts and deletes. This option is useful primarily when you know that you are mostly to query the rows in a certain order most of the time. By using this option after major changes to the table, you might be able to get higher performance. In some cases, it might make sorting easier for MySQL if the table is in order by the column that you want to order it by later.

    ORDER BY syntax permits one or more column names to be specified for sorting, each of which optionally can be followed by ASC or DESC to indicate ascending or descending sort order, respectively. The default is ascending order. Only column names are permitted as sort criteria; arbitrary expressions are not permitted.

    ORDER BY does not make sense for InnoDB tables that contain a user-defined clustered index (PRIMARY KEY or NOT NULL UNIQUE index). InnoDB always orders table rows according to such an index if one is present.

    Note

    When used on a partitioned table, ALTER TABLE ... ORDER BY orders rows within each partition only.

  • If you use ALTER TABLE on a MyISAM table, all nonunique indexes are created in a separate batch (as for REPAIR TABLE). This should make ALTER TABLE much faster when you have many indexes.

    This feature can be activated explicitly for a MyISAM table. ALTER TABLE ... DISABLE KEYS tells MySQL to stop updating nonunique indexes. ALTER TABLE ... ENABLE KEYS then should be used to re-create missing indexes. MySQL does this with a special algorithm that is much faster than inserting keys one by one, so disabling keys before performing bulk insert operations should give a considerable speedup. Using ALTER TABLE ... DISABLE KEYS requires the INDEX privilege in addition to the privileges mentioned earlier.

    While the nonunique indexes are disabled, they are ignored for statements such as SELECT and EXPLAIN that otherwise would use them.

  • If ALTER TABLE for an InnoDB table results in changes to column values (for example, because a column is truncated), InnoDB's FOREIGN KEY constraint checks do not notice possible violations caused by changing the values.

  • The FOREIGN KEY and REFERENCES clauses are supported by the InnoDB storage engine, which implements ADD [CONSTRAINT [symbol]] FOREIGN KEY (...) REFERENCES ... (...). See Section 13.6.5.4, “FOREIGN KEY Constraints”. For other storage engines, the clauses are parsed but ignored. The CHECK clause is parsed but ignored by all storage engines. See Section 12.1.14, “CREATE TABLE Syntax”. The reason for accepting but ignoring syntax clauses is for compatibility, to make it easier to port code from other SQL servers, and to run applications that create tables with references. See Section 1.8.5, “MySQL Differences from Standard SQL”.

    Important

    The inline REFERENCES specifications where the references are defined as part of the column specification are silently ignored by InnoDB. InnoDB only accepts REFERENCES clauses defined as part of a separate FOREIGN KEY specification.

    Note

    Partitioned tables do not support foreign keys. See Section 18.5, “Restrictions and Limitations on Partitioning”, for more information.

  • InnoDB supports the use of ALTER TABLE to drop foreign keys:

    ALTER TABLE tbl_name DROP FOREIGN KEY fk_symbol;
    

    For more information, see Section 13.6.5.4, “FOREIGN KEY Constraints”.

  • You cannot add a foreign key and drop a foreign key in separate clauses of a single ALTER TABLE statement. You must use separate statements.

  • For an InnoDB table that is created with its own tablespace in an .ibd file, that file can be discarded and imported. To discard the .ibd file, use this statement:

    ALTER TABLE tbl_name DISCARD TABLESPACE;
    

    This deletes the current .ibd file, so be sure that you have a backup first. Attempting to access the table while the tablespace file is discarded results in an error.

    To import the backup .ibd file back into the table, copy it into the database directory, and then issue this statement:

    ALTER TABLE tbl_name IMPORT TABLESPACE;
    

    The tablespace file must have been created on the server into which it is imported later.

    See Section 13.6.3, “Using Per-Table Tablespaces”.

  • Pending INSERT DELAYED statements are lost if a table is write locked and ALTER TABLE is used to modify the table structure.

  • If you want to change the table default character set and all character columns (CHAR, VARCHAR, TEXT) to a new character set, use a statement like this:

    ALTER TABLE tbl_name CONVERT TO CHARACTER SET charset_name;
    

    For a column that has a data type of VARCHAR or one of the TEXT types, CONVERT TO CHARACTER SET will change the data type as necessary to ensure that the new column is long enough to store as many characters as the original column. For example, a TEXT column has two length bytes, which store the byte-length of values in the column, up to a maximum of 65,535. For a latin1 TEXT column, each character requires a single byte, so the column can store up to 65,535 characters. If the column is converted to utf8, each character might require up to three bytes, for a maximum possible length of 3 × 65,535 = 196,605 bytes. That length will not fit in a TEXT column's length bytes, so MySQL will convert the data type to MEDIUMTEXT, which is the smallest string type for which the length bytes can record a value of 196,605. Similarly, a VARCHAR column might be converted to MEDIUMTEXT.

    To avoid data type changes of the type just described, do not use CONVERT TO CHARACTER SET. Instead, use MODIFY to change individual columns. For example:

    ALTER TABLE t MODIFY latin1_text_col TEXT CHARACTER SET utf8;
    ALTER TABLE t MODIFY latin1_varchar_col VARCHAR(M) CHARACTER SET utf8;
    

    If you specify CONVERT TO CHARACTER SET binary, the CHAR, VARCHAR, and TEXT columns are converted to their corresponding binary string types (BINARY, VARBINARY, BLOB). This means that the columns no longer will have a character set and a subsequent CONVERT TO operation will not apply to them.

    If charset_name is DEFAULT, the database character set is used.

    Warning

    The CONVERT TO operation converts column values between the character sets. This is not what you want if you have a column in one character set (like latin1) but the stored values actually use some other, incompatible character set (like utf8). In this case, you have to do the following for each such column:

    ALTER TABLE t1 CHANGE c1 c1 BLOB;
    ALTER TABLE t1 CHANGE c1 c1 TEXT CHARACTER SET utf8;
    

    The reason this works is that there is no conversion when you convert to or from BLOB columns.

    To change only the default character set for a table, use this statement:

    ALTER TABLE tbl_name DEFAULT CHARACTER SET charset_name;
    

    The word DEFAULT is optional. The default character set is the character set that is used if you do not specify the character set for columns that you add to a table later (for example, with ALTER TABLE ... ADD column).

  • Partitioning-related clauses for ALTER TABLE can be used with partitioned tables for repartitioning, for adding, dropping, merging, and splitting partitions, and for performing partitioning maintenance.

    Simply using a partition_options clause with ALTER TABLE on a partitioned table repartitions the table according to the partitioning scheme defined by the partition_options. This clause always begins with PARTITION BY, and follows the same syntax and other rules as apply to the partition_options clause for CREATE TABLE (see Section 12.1.14, “CREATE TABLE Syntax”, for more detailed information), and can also be used to partition an existing table that is not already partitioned. For example, consider a (nonpartitioned) table defined as shown here:

    CREATE TABLE t1 (
        id INT,
        year_col INT
    );
    

    This table can be partitioned by HASH, using the id column as the partitioning key, into 8 partitions by means of this statement:

    ALTER TABLE t1
        PARTITION BY HASH(id)
        PARTITIONS 8;
    

    The table that results from using an ALTER TABLE ... PARTITION BY statement must follow the same rules as one created using CREATE TABLE ... PARTITION BY. This includes the rules governing the relationship between any unique keys (including any primary key) that the table might have, and the column or columns used in the partitioning expression, as discussed in Section 18.5.1, “Partitioning Keys, Primary Keys, and Unique Keys”. The CREATE TABLE ... PARTITION BY rules for specifying the number of partitions also apply to ALTER TABLE ... PARTITION BY.

    The partition_definition clause for ALTER TABLE ADD PARTITION supports the same options as the clause of the same name for the CREATE TABLE statement. (See Section 12.1.14, “CREATE TABLE Syntax”, for the syntax and description.) Suppose that you have the partitioned table created as shown here:

    CREATE TABLE t1 (
        id INT,
        year_col INT
    )
    PARTITION BY RANGE (year_col) (
        PARTITION p0 VALUES LESS THAN (1991),
        PARTITION p1 VALUES LESS THAN (1995),
        PARTITION p2 VALUES LESS THAN (1999)
    );
    

    You can add a new partition p3 to this table for storing values less than 2002 as follows:

    ALTER TABLE t1 ADD PARTITION (PARTITION p3 VALUES LESS THAN (2002));
    

    DROP PARTITION can be used to drop one or more RANGE or LIST partitions. This statement cannot be used with HASH or KEY partitions; instead, use COALESCE PARTITION (see below). Any data that was stored in the dropped partitions named in the partition_names list is discarded. For example, given the table t1 defined previously, you can drop the partitions named p0 and p1 as shown here:

    ALTER TABLE t1 DROP PARTITION p0, p1;
    

    ADD PARTITION and DROP PARTITION do not currently support IF [NOT] EXISTS. It is also not possible to rename a partition or a partitioned table. Instead, if you wish to rename a partition, you must drop and re-create the partition; if you wish to rename a partitioned table, you must instead drop all partitions, rename the table, and then add back the partitions that were dropped.

    Beginning with MySQL 5.5.0, it is possible to delete rows from selected partitions using the TRUNCATE PARTITION option. This option takes a comma-separated list of one or more partition names. For example, consider the table t1 as defined here:

    CREATE TABLE t1 (
        id INT,
        year_col INT
    )
    PARTITION BY RANGE (year_col) (
        PARTITION p0 VALUES LESS THAN (1991),
        PARTITION p1 VALUES LESS THAN (1995),
        PARTITION p2 VALUES LESS THAN (1999),
        PARTITION p3 VALUES LESS THAN (2003),
        PARTITION p4 VALUES LESS THAN (2007)
    );
    

    To delete all rows from partition p0, you can use the following statement:

    ALTER TABLE t1 TRUNCATE PARTITION p0;
    

    The statement just shown has the same effect as the following DELETE statement:

    DELETE FROM t1 WHERE year_col < 1991;
    

    When truncating multiple partitions, the partitions do not have to be contiguous: This can greatly simplify delete operations on partitioned tables that would otherwise require very complex WHERE conditions if done with DELETE statements. For example, this statement deletes all rows from partitions p1 and p3:

    ALTER TABLE t1 TRUNCATE PARTITION p1, p3;
    

    An equivalent DELETE statement is shown here:

    DELETE FROM t1 WHERE 
        (year_col >= 1991 AND year_col < 1995)
        OR
        (year_col >= 2003 AND year_col < 2007);
    

    You can also use the ALL keyword in place of the list of partition names; in this case, the statement acts on all partitions in the table.

    TRUNCATE PARTITION merely deletes rows; it does not alter the definition of the table itself, or of any of its partitions.

    Note

    TRUNCATE PARTITION does not work with subpartitions.

    You can verify that the rows were dropped by checking the INFORMATION_SCHEMA.PARTITIONS table, using a query such as this one:

    SELECT PARTITION_NAME, TABLE_ROWS 
        FROM INFORMATION_SCHEMA.PARTITIONS 
        WHERE TABLE_NAME = 't1';
    

    TRUNCATE PARTITION is supported only for partitioned tables that use the MyISAM, InnoDB, or MEMORY storage engine. It also works on BLACKHOLE tables (but has no effect). It is not supported for ARCHIVE tables.

    COALESCE PARTITION can be used with a table that is partitioned by HASH or KEY to reduce the number of partitions by number. Suppose that you have created table t2 using the following definition:

    CREATE TABLE t2 (
        name VARCHAR (30),
        started DATE
    )
    PARTITION BY HASH( YEAR(started) )
    PARTITIONS 6;
    

    You can reduce the number of partitions used by t2 from 6 to 4 using the following statement:

    ALTER TABLE t2 COALESCE PARTITION 2;
    

    The data contained in the last number partitions will be merged into the remaining partitions. In this case, partitions 4 and 5 will be merged into the first 4 partitions (the partitions numbered 0, 1, 2, and 3).

    To change some but not all the partitions used by a partitioned table, you can use REORGANIZE PARTITION. This statement can be used in several ways:

    • To merge a set of partitions into a single partition. This can be done by naming several partitions in the partition_names list and supplying a single definition for partition_definition.

    • To split an existing partition into several partitions. You can accomplish this by naming a single partition for partition_names and providing multiple partition_definitions.

    • To change the ranges for a subset of partitions defined using VALUES LESS THAN or the value lists for a subset of partitions defined using VALUES IN.

    Note

    For partitions that have not been explicitly named, MySQL automatically provides the default names p0, p1, p2, and so on. The same is true with regard to subpartitions.

    For more detailed information about and examples of ALTER TABLE ... REORGANIZE PARTITION statements, see Section 18.3.1, “Management of RANGE and LIST Partitions”.

  • Several additional options provide partition maintenance and repair functionality analogous to that implemented for nonpartitioned tables by statements such as CHECK TABLE and REPAIR TABLE (which are also supported for partitioned tables; see Section 12.4.2, “Table Maintenance Statements” for more information). These include ANALYZE PARTITION, CHECK PARTITION, OPTIMIZE PARTITION, REBUILD PARTITION, and REPAIR PARTITION. Each of these options takes a partition_names clause consisting of one or more names of partitions, separated by commas. The partitions must already exist in the table to be altered. You can also use the ALL keyword in place of partition_names, in which case the statement acts on all partitions in the table. For more information and examples, see Section 18.3.3, “Maintenance of Partitions”.

    The ANALYZE PARTITION, CHECK PARTITION, OPTIMIZE PARTITION, and REPAIR PARTITION options are not permitted for tables which are not partitioned.

  • REMOVE PARTITIONING enables you to remove a table's partitioning without otherwise affecting the table or its data. This option can be combined with other ALTER TABLE options such as those used to add, drop, or rename drop columns or indexes.

  • Using the ENGINE option with ALTER TABLE changes the storage engine used by the table without affecting the partitioning.

Important

Only a single instance of any one of the following options can be used in a given ALTER TABLE statement: PARTITION BY, ADD PARTITION, DROP PARTITION, TRUNCATE PARTITION, REORGANIZE PARTITION, or COALESCE PARTITION, ANALYZE PARTITION, CHECK PARTITION, OPTIMIZE PARTITION, REBUILD PARTITION, REMOVE PARTITIONING.

For example, the following two statements are invalid:

ALTER TABLE t1 ANALYZE PARTITION p1, ANALYZE PARTITION p2;

ALTER TABLE t1 ANALYZE PARTITION p1, CHECK PARTITION p2;

In the first case, you can analyze partitions p1 and p2 of table t1 concurrently using a single statement with a single ANALYZE PARTITION option that lists both of the partitions to be analyzed, like this:

ALTER TABLE t1 ANALYZE PARTITION p1, p2;

In the second case, it is not possible to perform ANALYZE and CHECK operations on different partitions of the same table concurrently. Instead, you must issue two separate statements, like this:

ALTER TABLE t1 ANALYZE PARTITION p1;
ALTER TABLE t1 CHECK PARTITION p2;

With the mysql_info() C API function, you can find out how many rows were copied, and (when IGNORE is used) how many rows were deleted due to duplication of unique key values. See Section 22.9.3.35, “mysql_info().

ALTER TABLE Examples

Begin with a table t1 that is created as shown here:

CREATE TABLE t1 (a INTEGER,b CHAR(10));

To rename the table from t1 to t2:

ALTER TABLE t1 RENAME t2;

To change column a from INTEGER to TINYINT NOT NULL (leaving the name the same), and to change column b from CHAR(10) to CHAR(20) as well as renaming it from b to c:

ALTER TABLE t2 MODIFY a TINYINT NOT NULL, CHANGE b c CHAR(20);

To add a new TIMESTAMP column named d:

ALTER TABLE t2 ADD d TIMESTAMP;

To add an index on column d and a UNIQUE index on column a:

ALTER TABLE t2 ADD INDEX (d), ADD UNIQUE (a);

To remove column c:

ALTER TABLE t2 DROP COLUMN c;

To add a new AUTO_INCREMENT integer column named c:

ALTER TABLE t2 ADD c INT UNSIGNED NOT NULL AUTO_INCREMENT,
  ADD PRIMARY KEY (c);

We indexed c (as a PRIMARY KEY) because AUTO_INCREMENT columns must be indexed, and we declare c as NOT NULL because primary key columns cannot be NULL.

When you add an AUTO_INCREMENT column, column values are filled in with sequence numbers automatically. For MyISAM tables, you can set the first sequence number by executing SET INSERT_ID=value before ALTER TABLE or by using the AUTO_INCREMENT=value table option. See Section 5.1.4, “Server System Variables”.

With MyISAM tables, if you do not change the AUTO_INCREMENT column, the sequence number is not affected. If you drop an AUTO_INCREMENT column and then add another AUTO_INCREMENT column, the numbers are resequenced beginning with 1.

When replication is used, adding an AUTO_INCREMENT column to a table might not produce the same ordering of the rows on the slave and the master. This occurs because the order in which the rows are numbered depends on the specific storage engine used for the table and the order in which the rows were inserted. If it is important to have the same order on the master and slave, the rows must be ordered before assigning an AUTO_INCREMENT number. Assuming that you want to add an AUTO_INCREMENT column to the table t1, the following statements produce a new table t2 identical to t1 but with an AUTO_INCREMENT column:

CREATE TABLE t2 (id INT AUTO_INCREMENT PRIMARY KEY)
SELECT * FROM t1 ORDER BY col1, col2;

This assumes that the table t1 has columns col1 and col2.

This set of statements will also produce a new table t2 identical to t1, with the addition of an AUTO_INCREMENT column:

CREATE TABLE t2 LIKE t1;
ALTER TABLE t2 ADD id INT AUTO_INCREMENT PRIMARY KEY;
INSERT INTO t2 SELECT * FROM t1 ORDER BY col1, col2;
Important

To guarantee the same ordering on both master and slave, all columns of t1 must be referenced in the ORDER BY clause.

Regardless of the method used to create and populate the copy having the AUTO_INCREMENT column, the final step is to drop the original table and then rename the copy:

DROP t1;
ALTER TABLE t2 RENAME t1;

12.1.7. ALTER VIEW Syntax

ALTER
    [ALGORITHM = {UNDEFINED | MERGE | TEMPTABLE}]
    [DEFINER = { user | CURRENT_USER }]
    [SQL SECURITY { DEFINER | INVOKER }]
    VIEW view_name [(column_list)]
    AS select_statement
    [WITH [CASCADED | LOCAL] CHECK OPTION]

This statement changes the definition of a view, which must exist. The syntax is similar to that for CREATE VIEW and the effect is the same as for CREATE OR REPLACE VIEW. See Section 12.1.16, “CREATE VIEW Syntax”. This statement requires the CREATE VIEW and DROP privileges for the view, and some privilege for each column referred to in the SELECT statement. ALTER VIEW is permitted only to the definer or users with the SUPER privilege.

12.1.8. CREATE DATABASE Syntax

CREATE {DATABASE | SCHEMA} [IF NOT EXISTS] db_name
    [create_specification] ...

create_specification:
    [DEFAULT] CHARACTER SET [=] charset_name
  | [DEFAULT] COLLATE [=] collation_name

CREATE DATABASE creates a database with the given name. To use this statement, you need the CREATE privilege for the database. CREATE SCHEMA is a synonym for CREATE DATABASE.

An error occurs if the database exists and you did not specify IF NOT EXISTS.

As of MySQL 5.5.3, CREATE DATABASE is not permitted within a session that has an active LOCK TABLES statement.

create_specification options specify database characteristics. Database characteristics are stored in the db.opt file in the database directory. The CHARACTER SET clause specifies the default database character set. The COLLATE clause specifies the default database collation. Section 9.1, “Character Set Support”, discusses character set and collation names.

A database in MySQL is implemented as a directory containing files that correspond to tables in the database. Because there are no tables in a database when it is initially created, the CREATE DATABASE statement creates only a directory under the MySQL data directory and the db.opt file. Rules for permissible database names are given in Section 8.2, “Schema Object Names”. If a database name contains special characters, the name for the database directory contains encoded versions of those characters as described in Section 8.2.3, “Mapping of Identifiers to File Names”.

If you manually create a directory under the data directory (for example, with mkdir), the server considers it a database directory and it shows up in the output of SHOW DATABASES.

You can also use the mysqladmin program to create databases. See Section 4.5.2, “mysqladmin — Client for Administering a MySQL Server”.

12.1.9. CREATE EVENT Syntax

CREATE
    [DEFINER = { user | CURRENT_USER }]
    EVENT
    [IF NOT EXISTS]
    event_name
    ON SCHEDULE schedule
    [ON COMPLETION [NOT] PRESERVE]
    [ENABLE | DISABLE | DISABLE ON SLAVE]
    [COMMENT 'comment']
    DO event_body;

schedule:
    AT timestamp [+ INTERVAL interval] ...
  | EVERY interval
    [STARTS timestamp [+ INTERVAL interval] ...]
    [ENDS timestamp [+ INTERVAL interval] ...]

interval:
    quantity {YEAR | QUARTER | MONTH | DAY | HOUR | MINUTE |
              WEEK | SECOND | YEAR_MONTH | DAY_HOUR | DAY_MINUTE |
              DAY_SECOND | HOUR_MINUTE | HOUR_SECOND | MINUTE_SECOND}

This statement creates and schedules a new event. The event will not run unless the Event Scheduler is enabled. For information about checking Event Scheduler status and enabling it if necessary, see Section 19.4.2, “Event Scheduler Configuration”.

CREATE EVENT requires the EVENT privilege for the schema in which the event is to be created. It might also require the SUPER privilege, depending on the DEFINER value, as described later in this section.

The minimum requirements for a valid CREATE EVENT statement are as follows:

  • The keywords CREATE EVENT plus an event name, which uniquely identifies the event in a database schema.

  • An ON SCHEDULE clause, which determines when and how often the event executes.

  • A DO clause, which contains the SQL statement to be executed by an event.

This is an example of a minimal CREATE EVENT statement:

CREATE EVENT myevent
    ON SCHEDULE AT CURRENT_TIMESTAMP + INTERVAL 1 HOUR
    DO
      UPDATE myschema.mytable SET mycol = mycol + 1;

The previous statement creates an event named myevent. This event executes once—one hour following its creation—by running an SQL statement that increments the value of the myschema.mytable table's mycol column by 1.

The event_name must be a valid MySQL identifier with a maximum length of 64 characters. Event names are not case sensitive, so you cannot have two events named myevent and MyEvent in the same schema. In general, the rules governing event names are the same as those for names of stored routines. See Section 8.2, “Schema Object Names”.

An event is associated with a schema. If no schema is indicated as part of event_name, the default (current) schema is assumed. To create an event in a specific schema, qualify the event name with a schema using schema_name.event_name syntax.

The DEFINER clause specifies the MySQL account to be used when checking access privileges at event execution time. If a user value is given, it should be a MySQL account specified as 'user_name'@'host_name' (the same format used in the GRANT statement), CURRENT_USER, or CURRENT_USER(). The default DEFINER value is the user who executes the CREATE EVENT statement. This is the same as specifying DEFINER = CURRENT_USER explicitly.

If you specify the DEFINER clause, these rules determine the legal DEFINER user values:

  • If you do not have the SUPER privilege, the only legal user value is your own account, either specified literally or by using CURRENT_USER. You cannot set the definer to some other account.

  • If you have the SUPER privilege, you can specify any syntactically legal account name. If the account does not actually exist, a warning is generated.

  • Although it is possible to create an event with a nonexistent DEFINER account, an error occurs at event execution time if the account does not exist.

For more information about event security, see Section 19.6, “Access Control for Stored Programs and Views”.

Within an event, the CURRENT_USER() function returns the account used to check privileges at event execution time, which is the DEFINER user. For information about user auditing within events, see Section 5.5.10, “Auditing MySQL Account Activity”.

IF NOT EXISTS has the same meaning for CREATE EVENT as for CREATE TABLE: If an event named event_name already exists in the same schema, no action is taken, and no error results. (However, a warning is generated in such cases.)

The ON SCHEDULE clause determines when, how often, and for how long the event_body defined for the event repeats. This clause takes one of two forms:

  • AT timestamp is used for a one-time event. It specifies that the event executes one time only at the date and time given by timestamp, which must include both the date and time, or must be an expression that resolves to a datetime value. You may use a value of either the DATETIME or TIMESTAMP type for this purpose. If the date is in the past, a warning occurs, as shown here:

    mysql> SELECT NOW();
    +---------------------+
    | NOW()               |
    +---------------------+
    | 2006-02-10 23:59:01 |
    +---------------------+
    1 row in set (0.04 sec)
    
    mysql> CREATE EVENT e_totals
        ->     ON SCHEDULE AT '2006-02-10 23:59:00'
        ->     DO INSERT INTO test.totals VALUES (NOW());
    Query OK, 0 rows affected, 1 warning (0.00 sec)
    
    mysql> SHOW WARNINGS\G
    *************************** 1. row ***************************
      Level: Note
       Code: 1588
    Message: Event execution time is in the past and ON COMPLETION NOT
             PRESERVE is set. The event was dropped immediately after
             creation.
    

    CREATE EVENT statements which are themselves invalid—for whatever reason—fail with an error.

    You may use CURRENT_TIMESTAMP to specify the current date and time. In such a case, the event acts as soon as it is created.

    To create an event which occurs at some point in the future relative to the current date and time—such as that expressed by the phrase “three weeks from now”—you can use the optional clause + INTERVAL interval. The interval portion consists of two parts, a quantity and a unit of time, and follows the same syntax rules that govern intervals used in the DATE_ADD() function (see Section 11.7, “Date and Time Functions”. The units keywords are also the same, except that you cannot use any units involving microseconds when defining an event. With some interval types, complex time units may be used. For example, “two minutes and ten seconds” can be expressed as + INTERVAL '2:10' MINUTE_SECOND.

    You can also combine intervals. For example, AT CURRENT_TIMESTAMP + INTERVAL 3 WEEK + INTERVAL 2 DAY is equivalent to “three weeks and two days from now”. Each portion of such a clause must begin with + INTERVAL.

  • To repeat actions at a regular interval, use an EVERY clause. The EVERY keyword is followed by an interval as described in the previous dicussion of the AT keyword. (+ INTERVAL is not used with EVERY.) For example, EVERY 6 WEEK means “every six weeks”.

    Although + INTERVAL clauses are not permitted in an EVERY clause, you can use the same complex time units permitted in a + INTERVAL.

    An EVERY clause may contain an optional STARTS clause. STARTS is followed by a timestamp value that indicates when the action should begin repeating, and may also use + INTERVAL interval to specify an amount of time “from now”. For example, EVERY 3 MONTH STARTS CURRENT_TIMESTAMP + INTERVAL 1 WEEK means “every three months, beginning one week from now”. Similarly, you can express “every two weeks, beginning six hours and fifteen minutes from now” as EVERY 2 WEEK STARTS CURRENT_TIMESTAMP + INTERVAL '6:15' HOUR_MINUTE. Not specifying STARTS is the same as using STARTS CURRENT_TIMESTAMP—that is, the action specified for the event begins repeating immediately upon creation of the event.

    An EVERY clause may contain an optional ENDS clause. The ENDS keyword is followed by a timestamp value that tells MySQL when the event should stop repeating. You may also use + INTERVAL interval with ENDS; for instance, EVERY 12 HOUR STARTS CURRENT_TIMESTAMP + INTERVAL 30 MINUTE ENDS CURRENT_TIMESTAMP + INTERVAL 4 WEEK is equivalent to “every twelve hours, beginning thirty minutes from now, and ending four weeks from now”. Not using ENDS means that the event continues executing indefinitely.

    ENDS supports the same syntax for complex time units as STARTS does.

    You may use STARTS, ENDS, both, or neither in an EVERY clause.

    If a repeating event does not terminate within its scheduling interval, the result may be multiple instances of the event executing simultaneously. If this is undesirable, you should institute a mechanism to prevent simultaneous instances. For example, you could use the GET_LOCK() function, or row or table locking.

The ON SCHEDULE clause may use expressions involving built-in MySQL functions and user variables to obtain any of the timestamp or interval values which it contains. You may not use stored functions or user-defined functions in such expressions, nor may you use any table references; however, you may use SELECT FROM DUAL. This is true for both CREATE EVENT and ALTER EVENT statements. References to stored functions, user-defined functions, and tables in such cases are specifically not permitted, and fail with an error (see Bug#22830).

Times in the ON SCHEDULE clause are interpreted using the current session time_zone value. This becomes the event time zone; that is, the time zone that is used for event scheduling and is in effect within the event as it executes. These times are converted to UTC and stored along with the event time zone in the mysql.event table. This enables event execution to proceed as defined regardless of any subsequent changes to the server time zone or daylight saving time effects. For additional information about representation of event times, see Section 19.4.4, “Event Metadata”. See also Section 12.4.5.19, “SHOW EVENTS Syntax”, and Section 20.20, “The INFORMATION_SCHEMA EVENTS Table”.

Normally, once an event has expired, it is immediately dropped. You can override this behavior by specifying ON COMPLETION PRESERVE. Using ON COMPLETION NOT PRESERVE merely makes the default nonpersistent behavior explicit.

You can create an event but prevent it from being active using the DISABLE keyword. Alternatively, you can use ENABLE to make explicit the default status, which is active. This is most useful in conjunction with ALTER EVENT (see Section 12.1.2, “ALTER EVENT Syntax”).

A third value may also appear in place of ENABLED or DISABLED; DISABLE ON SLAVE is set for the status of an event on a replication slave to indicate that the event was created on the master and replicated to the slave, but is not executed on the slave. See Section 17.4.1.8, “Replication of Invoked Features”.

You may supply a comment for an event using a COMMENT clause. comment may be any string of up to 64 characters that you wish to use for describing the event. The comment text, being a string literal, must be surrounded by quotation marks.

The DO clause specifies an action carried by the event, and consists of an SQL statement. Nearly any valid MySQL statement that can be used in a stored routine can also be used as the action statement for a scheduled event. (See Section E.1, “Restrictions on Stored Routines, Triggers, and Events”.) For example, the following event e_hourly deletes all rows from the sessions table once per hour, where this table is part of the site_activity schema:

CREATE EVENT e_hourly
    ON SCHEDULE
      EVERY 1 HOUR
    COMMENT 'Clears out sessions table each hour.'
    DO
      DELETE FROM site_activity.sessions;

MySQL stores the sql_mode system variable setting that is in effect at the time an event is created, and always executes the event with this setting in force, regardless of the current server SQL mode.

A CREATE EVENT statement that contains an ALTER EVENT statement in its DO clause appears to succeed; however, when the server attempts to execute the resulting scheduled event, the execution fails with an error.

Note

Statements such as SELECT or SHOW that merely return a result set have no effect when used in an event; the output from these is not sent to the MySQL Monitor, nor is it stored anywhere. However, you can use statements such as SELECT ... INTO and INSERT INTO ... SELECT that store a result. (See the next example in this section for an instance of the latter.)

The schema to which an event belongs is the default schema for table references in the DO clause. Any references to tables in other schemas must be qualified with the proper schema name.

As with stored routines, you can use compound-statement syntax in the DO clause by using the BEGIN and END keywords, as shown here:

delimiter |

CREATE EVENT e_daily
    ON SCHEDULE
      EVERY 1 DAY
    COMMENT 'Saves total number of sessions then clears the table each day'
    DO
      BEGIN
        INSERT INTO site_activity.totals (time, total)
          SELECT CURRENT_TIMESTAMP, COUNT(*)
            FROM site_activity.sessions;
        DELETE FROM site_activity.sessions;
      END |

delimiter ;

This example uses the delimiter command to change the statement delimiter. See Section 19.1, “Defining Stored Programs”.

More complex compound statements, such as those used in stored routines, are possible in an event. This example uses local variables, an error handler, and a flow control construct:

delimiter |

CREATE EVENT e
    ON SCHEDULE
      EVERY 5 SECOND
    DO
      BEGIN
        DECLARE v INTEGER;
        DECLARE CONTINUE HANDLER FOR SQLEXCEPTION BEGIN END;

        SET v = 0;

        WHILE v < 5 DO
          INSERT INTO t1 VALUES (0);
          UPDATE t2 SET s1 = s1 + 1;
          SET v = v + 1;
        END WHILE;
    END |

delimiter ;

There is no way to pass parameters directly to or from events; however, it is possible to invoke a stored routine with parameters within an event:

CREATE EVENT e_call_myproc
    ON SCHEDULE
      AT CURRENT_TIMESTAMP + INTERVAL 1 DAY
    DO CALL myproc(5, 27);

If an event's definer has the SUPER privilege, the event can read and write global variables. As granting this privilege entails a potential for abuse, extreme care must be taken in doing so.

Generally, any statements that are valid in stored routines may be used for action statements executed by events. For more information about statements permissible within stored routines, see Section 19.2.1, “Stored Routine Syntax”. You can create an event as part of a stored routine, but an event cannot be created by another event.

12.1.10. The CREATE FUNCTION Statement

The CREATE FUNCTION statement is used to create stored functions and user-defined functions (UDFs):

12.1.11. CREATE INDEX Syntax

CREATE [UNIQUE|FULLTEXT|SPATIAL] INDEX index_name
    [index_type]
    ON tbl_name (index_col_name,...)
    [index_option] ...

index_col_name:
    col_name [(length)] [ASC | DESC]

index_type:
    USING {BTREE | HASH}

index_option:
    KEY_BLOCK_SIZE [=] value
  | index_type
  | WITH PARSER parser_name
  | COMMENT 'string'

CREATE INDEX is mapped to an ALTER TABLE statement to create indexes. See Section 12.1.6, “ALTER TABLE Syntax”. CREATE INDEX cannot be used to create a PRIMARY KEY; use ALTER TABLE instead. For more information about indexes, see Section 7.3.1, “How MySQL Uses Indexes”.

Normally, you create all indexes on a table at the time the table itself is created with CREATE TABLE. See Section 12.1.14, “CREATE TABLE Syntax”. CREATE INDEX enables you to add indexes to existing tables.

A column list of the form (col1,col2,...) creates a multiple-column index. Index values are formed by concatenating the values of the given columns.

Indexes can be created that use only the leading part of column values, using col_name(length) syntax to specify an index prefix length:

  • Prefixes can be specified for CHAR, VARCHAR, BINARY, and VARBINARY columns.

  • BLOB and TEXT columns also can be indexed, but a prefix length must be given.

  • Prefix lengths are given in characters for nonbinary string types and in bytes for binary string types. That is, index entries consist of the first length characters of each column value for CHAR, VARCHAR, and TEXT columns, and the first length bytes of each column value for BINARY, VARBINARY, and BLOB columns.

  • For spatial columns, prefix values cannot be given, as described later in this section.

The statement shown here creates an index using the first 10 characters of the name column:

CREATE INDEX part_of_name ON customer (name(10));

If names in the column usually differ in the first 10 characters, this index should not be much slower than an index created from the entire name column. Also, using column prefixes for indexes can make the index file much smaller, which could save a lot of disk space and might also speed up INSERT operations.

Prefix support and lengths of prefixes (where supported) are storage engine dependent. For example, a prefix can be up to 1000 bytes long for MyISAM tables, and 767 bytes for InnoDB tables.

Note

Prefix limits are measured in bytes, whereas the prefix length in CREATE INDEX statements is interpreted as number of characters for nonbinary data types (CHAR, VARCHAR, TEXT). Take this into account when specifying a prefix length for a column that uses a multi-byte character set.

A UNIQUE index creates a constraint such that all values in the index must be distinct. An error occurs if you try to add a new row with a key value that matches an existing row. For all engines, a UNIQUE index permits multiple NULL values for columns that can contain NULL. If you specify a prefix value for a column in a UNIQUE index, the column values must be unique within the prefix.

FULLTEXT indexes are supported only for MyISAM tables and can include only CHAR, VARCHAR, and TEXT columns. Indexing always happens over the entire column; column prefix indexing is not supported and any prefix length is ignored if specified. See Section 11.9, “Full-Text Search Functions”, for details of operation.

The MyISAM, InnoDB, NDB, and ARCHIVE storage engines support spatial columns such as (POINT and GEOMETRY. (Section 11.17, “Spatial Extensions”, describes the spatial data types.) However, support for spatial column indexing varies among engines. Spatial and nonspatial indexes are available according to the following rules.

Characteristics of spatial indexes (created using SPATIAL INDEX):

  • Available only for MyISAM tables. Specifying SPATIAL INDEX for other storage engines results in an error.

  • Indexed columns must be NOT NULL.

  • In MySQL 5.5, column prefix lengths are prohibited. The full width of each column is indexed.

Characteristics of nonspatial indexes (created with INDEX, UNIQUE, or PRIMARY KEY):

  • Permitted for any storage engine that supports spatial columns except ARCHIVE.

  • Columns can be NULL unless the index is a primary key.

  • For each spatial column in a non-SPATIAL index except POINT columns, a column prefix length must be specified. (This is the same requirement as for indexed BLOB columns.) The prefix length is given in bytes.

  • The index type for a non-SPATIAL index depends on the storage engine. Currently, B-tree is used.

In MySQL 5.5:

  • You can add an index on a column that can have NULL values only if you are using the MyISAM, InnoDB, or MEMORY storage engine.

  • You can add an index on a BLOB or TEXT column only if you are using the MyISAM, or InnoDB storage engine.

An index_col_name specification can end with ASC or DESC. These keywords are permitted for future extensions for specifying ascending or descending index value storage. Currently, they are parsed but ignored; index values are always stored in ascending order.

Following the index column list, index options can be given. An index_option value can be any of the following:

  • KEY_BLOCK_SIZE [=] value

    This option provides a hint to the storage engine about the size in bytes to use for index key blocks. The engine is permitted to change the value if necessary. A value of 0 indicates that the default value should be used.

  • index_type

    Some storage engines permit you to specify an index type when creating an index. The permissible index type values supported by different storage engines are shown in the following table. Where multiple index types are listed, the first one is the default when no index type specifier is given.

    Storage EnginePermissible Index Types
    MyISAMBTREE
    InnoDBBTREE
    MEMORY/HEAPHASH, BTREE
    NDBHASH, BTREE (see note in text)

    Example:

    CREATE TABLE lookup (id INT) ENGINE = MEMORY;
    CREATE INDEX id_index USING BTREE ON lookup (id);
    

    The index_type clause cannot be used together with SPATIAL INDEX.

    If you specify an index type that is not legal for a given storage engine, but there is another index type available that the engine can use without affecting query results, the engine uses the available type. The parser recognizes RTREE as a type name, but currently this cannot be specfied for any storage engine.

    Use of this option before the ON tbl_name clause is deprecated; support for use of the option in this position is to be dropped in a future MySQL release. If an index_type option is given in both the earlier and later positions, the final option applies.

    TYPE type_name is recognized as a synonym for USING type_name. However, USING is the preferred form.

  • WITH PARSER parser_name

    This option can be used only with FULLTEXT indexes. It associates a parser plugin with the index if full-text indexing and searching operations need special handling. See Section 23.2, “The MySQL Plugin API”, for details on creating plugins.

  • As of MySQL 5.5.3, index definitions can include an optional comment of up to 1024 characters.

12.1.12. CREATE PROCEDURE and CREATE FUNCTION Syntax

CREATE
    [DEFINER = { user | CURRENT_USER }]
    PROCEDURE sp_name ([proc_parameter[,...]])
    [characteristic ...] routine_body

CREATE
    [DEFINER = { user | CURRENT_USER }]
    FUNCTION sp_name ([func_parameter[,...]])
    RETURNS type
    [characteristic ...] routine_body

proc_parameter:
    [ IN | OUT | INOUT ] param_name type

func_parameter:
    param_name type

type:
    Any valid MySQL data type

characteristic:
    COMMENT 'string'
  | LANGUAGE SQL
  | [NOT] DETERMINISTIC
  | { CONTAINS SQL | NO SQL | READS SQL DATA | MODIFIES SQL DATA }
  | SQL SECURITY { DEFINER | INVOKER }

routine_body:
    Valid SQL routine statement

These statements create stored routines. By default, a routine is associated with the default database. To associate the routine explicitly with a given database, specify the name as db_name.sp_name when you create it.

The CREATE FUNCTION statement is also used in MySQL to support UDFs (user-defined functions). See Section 23.3, “Adding New Functions to MySQL”. A UDF can be regarded as an external stored function. Stored functions share their namespace with UDFs. See Section 8.2.4, “Function Name Parsing and Resolution”, for the rules describing how the server interprets references to different kinds of functions.

To invoke a stored procedure, use the CALL statement (see Section 12.2.1, “CALL Syntax”). To invoke a stored function, refer to it in an expression. The function returns a value during expression evaluation.

CREATE PROCEDURE and CREATE FUNCTION require the CREATE ROUTINE privilege. They might also require the SUPER privilege, depending on the DEFINER value, as described later in this section. If binary logging is enabled, CREATE FUNCTION might require the SUPER privilege, as described in Section 19.7, “Binary Logging of Stored Programs”.

By default, MySQL automatically grants the ALTER ROUTINE and EXECUTE privileges to the routine creator. This behavior can be changed by disabling the automatic_sp_privileges system variable. See Section 19.2.2, “Stored Routines and MySQL Privileges”.

The DEFINER and SQL SECURITY clauses specify the security context to be used when checking access privileges at routine execution time, as described later in this section.

If the routine name is the same as the name of a built-in SQL function, a syntax error occurs unless you use a space between the name and the following parenthesis when defining the routine or invoking it later. For this reason, avoid using the names of existing SQL functions for your own stored routines.

The IGNORE_SPACE SQL mode applies to built-in functions, not to stored routines. It is always permissible to have spaces after a stored routine name, regardless of whether IGNORE_SPACE is enabled.

The parameter list enclosed within parentheses must always be present. If there are no parameters, an empty parameter list of () should be used. Parameter names are not case sensitive.

Each parameter is an IN parameter by default. To specify otherwise for a parameter, use the keyword OUT or INOUT before the parameter name.

Note

Specifying a parameter as IN, OUT, or INOUT is valid only for a PROCEDURE. For a FUNCTION, parameters are always regarded as IN parameters.

An IN parameter passes a value into a procedure. The procedure might modify the value, but the modification is not visible to the caller when the procedure returns. An OUT parameter passes a value from the procedure back to the caller. Its initial value is NULL within the procedure, and its value is visible to the caller when the procedure returns. An INOUT parameter is initialized by the caller, can be modified by the procedure, and any change made by the procedure is visible to the caller when the procedure returns.

For each OUT or INOUT parameter, pass a user-defined variable in the CALL statement that invokes the procedure so that you can obtain its value when the procedure returns. If you are calling the procedure from within another stored procedure or function, you can also pass a routine parameter or local routine variable as an IN or INOUT parameter.

The following example shows a simple stored procedure that uses an OUT parameter:

mysql> delimiter //

mysql> CREATE PROCEDURE simpleproc (OUT param1 INT)
    -> BEGIN
    ->   SELECT COUNT(*) INTO param1 FROM t;
    -> END//
Query OK, 0 rows affected (0.00 sec)

mysql> delimiter ;

mysql> CALL simpleproc(@a);
Query OK, 0 rows affected (0.00 sec)

mysql> SELECT @a;
+------+
| @a   |
+------+
| 3    |
+------+
1 row in set (0.00 sec)

The example uses the mysql client delimiter command to change the statement delimiter from ; to // while the procedure is being defined. This enables the ; delimiter used in the procedure body to be passed through to the server rather than being interpreted by mysql itself. See Section 19.1, “Defining Stored Programs”.

The RETURNS clause may be specified only for a FUNCTION, for which it is mandatory. It indicates the return type of the function, and the function body must contain a RETURN value statement. If the RETURN statement returns a value of a different type, the value is coerced to the proper type. For example, if a function specifies an ENUM or SET value in the RETURNS clause, but the RETURN statement returns an integer, the value returned from the function is the string for the corresponding ENUM member of set of SET members.

The following example function takes a parameter, performs an operation using an SQL function, and returns the result. In this case, it is unnecessary to use delimiter because the function definition contains no internal ; statement delimiters:

mysql> CREATE FUNCTION hello (s CHAR(20))
mysql> RETURNS CHAR(50) DETERMINISTIC
    -> RETURN CONCAT('Hello, ',s,'!');
Query OK, 0 rows affected (0.00 sec)

mysql> SELECT hello('world');
+----------------+
| hello('world') |
+----------------+
| Hello, world!  |
+----------------+
1 row in set (0.00 sec)

Parameter types and function return types can be declared to use any valid data type, except that the COLLATE attribute cannot be used prior to MySQL 5.5.3. As of 5.5.3, COLLATE can be used if preceded by the CHARACTER SET attribute.

The routine_body consists of a valid SQL routine statement. This can be a simple statement such as SELECT or INSERT, or a compound statement written using BEGIN and END. Compound statements can contain declarations, loops, and other control structure statements. The syntax for these statements is described in Section 12.7, “MySQL Compound-Statement Syntax”.

MySQL permits routines to contain DDL statements, such as CREATE and DROP. MySQL also permits stored procedures (but not stored functions) to contain SQL transaction statements such as COMMIT. Stored functions may not contain statements that perform explicit or implicit commit or rollback. Support for these statements is not required by the SQL standard, which states that each DBMS vendor may decide whether to permit them.

Statements that return a result set can be used within a stored procedcure but not within a stored function. This prohibition includes SELECT statements that do not have an INTO var_list clause and other statements such as SHOW, EXPLAIN, and CHECK TABLE. For statements that can be determined at function definition time to return a result set, a Not allowed to return a result set from a function error occurs (ER_SP_NO_RETSET). For statements that can be determined only at runtime to return a result set, a PROCEDURE %s can't return a result set in the given context error occurs (ER_SP_BADSELECT).

USE statements within stored routines are not permitted. When a routine is invoked, an implicit USE db_name is performed (and undone when the routine terminates). The causes the routine to have the given default database while it executes. References to objects in databases other than the routine default database should be qualified with the appropriate database name.

For additional information about statements that are not permitted in stored routines, see Section E.1, “Restrictions on Stored Routines, Triggers, and Events”.

For information about invoking stored procedures from within programs written in a language that has a MySQL interface, see Section 12.2.1, “CALL Syntax”.

MySQL stores the sql_mode system variable setting that is in effect at the time a routine is created, and always executes the routine with this setting in force, regardless of the server SQL mode in effect when the routine is invoked.

The switch from the SQL mode of the invoker to that of the routine occurs after evaluation of arguments and assignment of the resulting values to routine parameters. If you define a routine in strict SQL mode but invoke it in nonstrict mode, assignment of arguments to routine parameters does not take place in strict mode. If you require that expressions passed to a routine be assigned in strict SQL mode, you should invoke the routine with strict mode in effect.

The COMMENT characteristic is a MySQL extension, and may be used to describe the stored routine. This information is displayed by the SHOW CREATE PROCEDURE and SHOW CREATE FUNCTION statements.

The LANGUAGE characteristic indicates the language in which the routine is written. The server ignores this characteristic; only SQL routines are supported.

A routine is considered “deterministic” if it always produces the same result for the same input parameters, and “not deterministic” otherwise. If neither DETERMINISTIC nor NOT DETERMINISTIC is given in the routine definition, the default is NOT DETERMINISTIC. To declare that a function is deterministic, you must specify DETERMINISTIC explicitly.

Assessment of the nature of a routine is based on the “honesty” of the creator: MySQL does not check that a routine declared DETERMINISTIC is free of statements that produce nondeterministic results. However, misdeclaring a routine might affect results or affect performance. Declaring a nondeterministic routine as DETERMINISTIC might lead to unexpected results by causing the optimizer to make incorrect execution plan choices. Declaring a deterministic routine as NONDETERMINISTIC might diminish performance by causing available optimizations not to be used.

If binary logging is enabled, the DETERMINISTIC characteristic affects which routine definitions MySQL accepts. See Section 19.7, “Binary Logging of Stored Programs”.

A routine that contains the NOW() function (or its synonyms) or RAND() is nondeterministic, but it might still be replication-safe. For NOW(), the binary log includes the timestamp and replicates correctly. RAND() also replicates correctly as long as it is called only a single time during the execution of a routine. (You can consider the routine execution timestamp and random number seed as implicit inputs that are identical on the master and slave.)

Several characteristics provide information about the nature of data use by the routine. In MySQL, these characteristics are advisory only. The server does not use them to constrain what kinds of statements a routine will be permitted to execute.

  • CONTAINS SQL indicates that the routine does not contain statements that read or write data. This is the default if none of these characteristics is given explicitly. Examples of such statements are SET @x = 1 or DO RELEASE_LOCK('abc'), which execute but neither read nor write data.

  • NO SQL indicates that the routine contains no SQL statements.

  • READS SQL DATA indicates that the routine contains statements that read data (for example, SELECT), but not statements that write data.

  • MODIFIES SQL DATA indicates that the routine contains statements that may write data (for example, INSERT or DELETE).

The SQL SECURITY characteristic can be DEFINER or INVOKER to specify the security context; that is, whether the routine executes using the privileges of the account named in the routine DEFINER clause or the user who invokes it. This account must have permission to access the database with which the routine is associated. The default value is DEFINER. The user who invokes the routine must have the EXECUTE privilege for it, as must the DEFINER account if the routine executes in definer security context.

The DEFINER clause specifies the MySQL account to be used when checking access privileges at routine execution time for routines that have the SQL SECURITY DEFINER characteristic.

If a user value is given for the DEFINER clause, it should be a MySQL account specified as 'user_name'@'host_name' (the same format used in the GRANT statement), CURRENT_USER, or CURRENT_USER(). The default DEFINER value is the user who executes the CREATE PROCEDURE or CREATE FUNCTION or statement. This is the same as specifying DEFINER = CURRENT_USER explicitly.

If you specify the DEFINER clause, these rules determine the legal DEFINER user values:

  • If you do not have the SUPER privilege, the only legal user value is your own account, either specified literally or by using CURRENT_USER. You cannot set the definer to some other account.

  • If you have the SUPER privilege, you can specify any syntactically legal account name. If the account does not actually exist, a warning is generated.

  • Although it is possible to create a routine with a nonexistent DEFINER account, an error occurs at routine execution time if the SQL SECURITY value is DEFINER but the definer account does not exist.

For more information about stored routine security, see Section 19.6, “Access Control for Stored Programs and Views”.

Within a stored routine that is defined with the SQL SECURITY DEFINER characteristic, CURRENT_USER returns the routine's DEFINER value. For information about user auditing within stored routines, see Section 5.5.10, “Auditing MySQL Account Activity”.

Consider the following procedure, which displays a count of the number of MySQL accounts listed in the mysql.user table:

CREATE DEFINER = 'admin'@'localhost' PROCEDURE account_count()
BEGIN
  SELECT 'Number of accounts:', COUNT(*) FROM mysql.user;
END;

The procedure is assigned a DEFINER account of 'admin'@'localhost' no matter which user defines it. It executes with the privileges of that account no matter which user invokes it (because the default security characteristic is DEFINER). The procedure succeeds or fails depending on whether invoker has the EXECUTE privilege for it and 'admin'@'localhost' has the SELECT privilege for the mysql.user table.

Now suppose that the procedure is defined with the SQL SECURITY INVOKER characteristic:

CREATE DEFINER = 'admin'@'localhost' PROCEDURE account_count()
SQL SECURITY INVOKER
BEGIN
  SELECT 'Number of accounts:', COUNT(*) FROM mysql.user;
END;

The procedure still has a DEFINER of 'admin'@'localhost', but in this case, it executes with the privileges of the invoking user. Thus, the procedure succeeds or fails depending on whether the invoker has the EXECUTE privilege for it and the SELECT privilege for the mysql.user table.

The server handles the data type of a routine parameter, local routine variable created with DECLARE, or function return value as follows:

  • Assignments are checked for data type mismatches and overflow. Conversion and overflow problems result in warnings, or errors in strict SQL mode.

  • Only scalar values can be assigned. For example, a statement such as SET x = (SELECT 1, 2) is invalid.

  • For character data types, if there is a CHARACTER SET attribute in the declaration, the specified character set and its default collation is used. If the COLLATE attribute is also present, that collation is used rather than the default collation. If there is no CHARACTER SET attribute, the database character set and collation in effect at routine creation time are used. (The database character set and collation are given by the value of the character_set_database and collation_database system variables.)

    Prior to MySQL 5.5.3, if there is a CHARACTER SET attribute in the declaration, the COLLATE attribute is not supported, and the character set's default collation is used. (This includes use of BINARY, which in this context specifies the binary collation of the character set.) If there is no CHARACTER SET attribute, the database character set and its default collation (rather than the database collation) are used.

12.1.13. CREATE SERVER Syntax

CREATE SERVER server_name
    FOREIGN DATA WRAPPER wrapper_name
    OPTIONS (option [, option] ...)

option:
  { HOST character-literal
  | DATABASE character-literal
  | USER character-literal
  | PASSWORD character-literal
  | SOCKET character-literal
  | OWNER character-literal
  | PORT numeric-literal }

This statement creates the definition of a server for use with the FEDERATED storage engine. The CREATE SERVER statement creates a new row within the servers table within the mysql database. This statement requires the SUPER privilege.

The server_name should be a unique reference to the server. Server definitions are global within the scope of the server, it is not possible to qualify the server definition to a specific database. server_name has a maximum length of 64 characters (names longer than 64 characters are silently truncated), and is case insensitive. You may specify the name as a quoted string.

The wrapper_name should be mysql, and may be quoted with single quotation marks. Other values for wrapper_name are not currently supported.

For each option you must specify either a character literal or numeric literal. Character literals are UTF-8, support a maximum length of 64 characters and default to a blank (empty) string. String literals are silently truncated to 64 characters. Numeric literals must be a number between 0 and 9999, default value is 0.

Note

Note that the OWNER option is currently not applied, and has no effect on the ownership or operation of the server connection that is created.

The CREATE SERVER statement creates an entry in the mysql.servers table that can later be used with the CREATE TABLE statement when creating a FEDERATED table. The options that you specify will be used to populate the columns in the mysql.servers table. The table columns are Server_name, Host, Db, Username, Password, Port and Socket.

For example:

CREATE SERVER s
FOREIGN DATA WRAPPER mysql
OPTIONS (USER 'Remote', HOST '192.168.1.106', DATABASE 'test');

The data stored in the table can be used when creating a connection to a FEDERATED table:

CREATE TABLE t (s1 INT) ENGINE=FEDERATED CONNECTION='s';

For more information, see Section 13.11, “The FEDERATED Storage Engine”.

CREATE SERVER does not cause an automatic commit.

12.1.14. CREATE TABLE Syntax

CREATE [TEMPORARY] TABLE [IF NOT EXISTS] tbl_name
    (create_definition,...)
    [table_options]
    [partition_options]

Or:

CREATE [TEMPORARY] TABLE [IF NOT EXISTS] tbl_name
    [(create_definition,...)]
    [table_options]
    [partition_options]
    select_statement

Or:

CREATE [TEMPORARY] TABLE [IF NOT EXISTS] tbl_name
    { LIKE old_tbl_name | (LIKE old_tbl_name) }
create_definition:
    col_name column_definition
  | [CONSTRAINT [symbol]] PRIMARY KEY [index_type] (index_col_name,...)
      [index_option] ...
  | {INDEX|KEY} [index_name] [index_type] (index_col_name,...)
      [index_option] ...
  | [CONSTRAINT [symbol]] UNIQUE [INDEX|KEY]
      [index_name] [index_type] (index_col_name,...)
      [index_option] ...
  | {FULLTEXT|SPATIAL} [INDEX|KEY] [index_name] (index_col_name,...)
      [index_option] ...
  | [CONSTRAINT [symbol]] FOREIGN KEY
      [index_name] (index_col_name,...) reference_definition
  | CHECK (expr)

column_definition:
    data_type [NOT NULL | NULL] [DEFAULT default_value]
      [AUTO_INCREMENT] [UNIQUE [KEY] | [PRIMARY] KEY]
      [COMMENT 'string']
      [COLUMN_FORMAT {FIXED|DYNAMIC|DEFAULT}]
      [reference_definition]

data_type:
    BIT[(length)]
  | TINYINT[(length)] [UNSIGNED] [ZEROFILL]
  | SMALLINT[(length)] [UNSIGNED] [ZEROFILL]
  | MEDIUMINT[(length)] [UNSIGNED] [ZEROFILL]
  | INT[(length)] [UNSIGNED] [ZEROFILL]
  | INTEGER[(length)] [UNSIGNED] [ZEROFILL]
  | BIGINT[(length)] [UNSIGNED] [ZEROFILL]
  | REAL[(length,decimals)] [UNSIGNED] [ZEROFILL]
  | DOUBLE[(length,decimals)] [UNSIGNED] [ZEROFILL]
  | FLOAT[(length,decimals)] [UNSIGNED] [ZEROFILL]
  | DECIMAL[(length[,decimals])] [UNSIGNED] [ZEROFILL]
  | NUMERIC[(length[,decimals])] [UNSIGNED] [ZEROFILL]
  | DATE
  | TIME
  | TIMESTAMP
  | DATETIME
  | YEAR
  | CHAR[(length)]
      [CHARACTER SET charset_name] [COLLATE collation_name]
  | VARCHAR(length)
      [CHARACTER SET charset_name] [COLLATE collation_name]
  | BINARY[(length)]
  | VARBINARY(length)
  | TINYBLOB
  | BLOB
  | MEDIUMBLOB
  | LONGBLOB
  | TINYTEXT [BINARY]
      [CHARACTER SET charset_name] [COLLATE collation_name]
  | TEXT [BINARY]
      [CHARACTER SET charset_name] [COLLATE collation_name]
  | MEDIUMTEXT [BINARY]
      [CHARACTER SET charset_name] [COLLATE collation_name]
  | LONGTEXT [BINARY]
      [CHARACTER SET charset_name] [COLLATE collation_name]
  | ENUM(value1,value2,value3,...)
      [CHARACTER SET charset_name] [COLLATE collation_name]
  | SET(value1,value2,value3,...)
      [CHARACTER SET charset_name] [COLLATE collation_name]
  | spatial_type

index_col_name:
    col_name [(length)] [ASC | DESC]

index_type:
    USING {BTREE | HASH}

index_option:
    KEY_BLOCK_SIZE [=] value
  | index_type
  | WITH PARSER parser_name
  | COMMENT 'string'

reference_definition:
    REFERENCES tbl_name (index_col_name,...)
      [MATCH FULL | MATCH PARTIAL | MATCH SIMPLE]
      [ON DELETE reference_option]
      [ON UPDATE reference_option]

reference_option:
    RESTRICT | CASCADE | SET NULL | NO ACTION

table_options:
    table_option [[,] table_option] ...

table_option:
    ENGINE [=] engine_name
  | AUTO_INCREMENT [=] value
  | AVG_ROW_LENGTH [=] value
  | [DEFAULT] CHARACTER SET [=] charset_name
  | CHECKSUM [=] {0 | 1}
  | [DEFAULT] COLLATE [=] collation_name
  | COMMENT [=] 'string'
  | CONNECTION [=] 'connect_string'
  | DATA DIRECTORY [=] 'absolute path to directory'
  | DELAY_KEY_WRITE [=] {0 | 1}
  | INDEX DIRECTORY [=] 'absolute path to directory'
  | INSERT_METHOD [=] { NO | FIRST | LAST }
  | KEY_BLOCK_SIZE [=] value
  | MAX_ROWS [=] value
  | MIN_ROWS [=] value
  | PACK_KEYS [=] {0 | 1 | DEFAULT}
  | PASSWORD [=] 'string'
  | ROW_FORMAT [=] {DEFAULT|DYNAMIC|FIXED|COMPRESSED|REDUNDANT|COMPACT}
  | UNION [=] (tbl_name[,tbl_name]...)

partition_options:
    PARTITION BY
        { [LINEAR] HASH(expr)
        | [LINEAR] KEY(column_list)
        | RANGE{(expr) | COLUMNS(column_list)}
        | LIST{(expr) | COLUMNS(column_list)} }
    [PARTITIONS num]
    [SUBPARTITION BY
        { [LINEAR] HASH(expr)
        | [LINEAR] KEY(column_list) }
      [SUBPARTITIONS num]
    ]
    [(partition_definition [, partition_definition] ...)]

partition_definition:
    PARTITION partition_name
        [VALUES 
            {LESS THAN {(expr | value_list) | MAXVALUE} 
            | 
            IN (value_list)}]
        [[STORAGE] ENGINE [=] engine_name]
        [COMMENT [=] 'comment_text' ]
        [DATA DIRECTORY [=] 'data_dir']
        [INDEX DIRECTORY [=] 'index_dir']
        [MAX_ROWS [=] max_number_of_rows]
        [MIN_ROWS [=] min_number_of_rows]
        [(subpartition_definition [, subpartition_definition] ...)]

subpartition_definition:
    SUBPARTITION logical_name
        [[STORAGE] ENGINE [=] engine_name]
        [COMMENT [=] 'comment_text' ]
        [DATA DIRECTORY [=] 'data_dir']
        [INDEX DIRECTORY [=] 'index_dir']
        [MAX_ROWS [=] max_number_of_rows]
        [MIN_ROWS [=] min_number_of_rows]

select_statement:
    [IGNORE | REPLACE] [AS] SELECT ...   (Some legal select statement)

CREATE TABLE creates a table with the given name. You must have the CREATE privilege for the table.

Rules for permissible table names are given in Section 8.2, “Schema Object Names”. By default, the table is created in the default database. An error occurs if the table exists, if there is no default database, or if the database does not exist.

The table name can be specified as db_name.tbl_name to create the table in a specific database. This works regardless of whether there is a default database, assuming that the database exists. If you use quoted identifiers, quote the database and table names separately. For example, write `mydb`.`mytbl`, not `mydb.mytbl`.

You can use the TEMPORARY keyword when creating a table. A TEMPORARY table is visible only to the current connection, and is dropped automatically when the connection is closed. This means that two different connections can use the same temporary table name without conflicting with each other or with an existing non-TEMPORARY table of the same name. (The existing table is hidden until the temporary table is dropped.) To create temporary tables, you must have the CREATE TEMPORARY TABLES privilege.

Note

CREATE TABLE does not automatically commit the current active transaction if you use the TEMPORARY keyword.

The keywords IF NOT EXISTS prevent an error from occurring if the table exists. However, there is no verification that the existing table has a structure identical to that indicated by the CREATE TABLE statement.

MySQL represents each table by an .frm table format (definition) file in the database directory. The storage engine for the table might create other files as well. In the case of MyISAM tables, the storage engine creates data and index files. Thus, for each MyISAM table tbl_name, there are three disk files.

FilePurpose
tbl_name.frmTable format (definition) file
tbl_name.MYDData file
tbl_name.MYIIndex file

Chapter 13, Storage Engines, describes what files each storage engine creates to represent tables. If a table name contains special characters, the names for the table files contain encoded versions of those characters as described in Section 8.2.3, “Mapping of Identifiers to File Names”.

data_type represents the data type in a column definition. spatial_type represents a spatial data type. The data type syntax shown is representative only. For a full description of the syntax available for specifying column data types, as well as information about the properties of each type, see Chapter 10, Data Types, and Section 11.17, “Spatial Extensions”.

Some attributes do not apply to all data types. AUTO_INCREMENT applies only to integer and floating-point types. DEFAULT does not apply to the BLOB or TEXT types.

  • If neither NULL nor NOT NULL is specified, the column is treated as though NULL had been specified.

  • An integer or floating-point column can have the additional attribute AUTO_INCREMENT. When you insert a value of NULL (recommended) or 0 into an indexed AUTO_INCREMENT column, the column is set to the next sequence value. Typically this is value+1, where value is the largest value for the column currently in the table. AUTO_INCREMENT sequences begin with 1.

    To retrieve an AUTO_INCREMENT value after inserting a row, use the LAST_INSERT_ID() SQL function or the mysql_insert_id() C API function. See Section 11.14, “Information Functions”, and Section 22.9.3.37, “mysql_insert_id().

    If the NO_AUTO_VALUE_ON_ZERO SQL mode is enabled, you can store 0 in AUTO_INCREMENT columns as 0 without generating a new sequence value. See Section 5.1.7, “Server SQL Modes”.

    Note

    There can be only one AUTO_INCREMENT column per table, it must be indexed, and it cannot have a DEFAULT value. An AUTO_INCREMENT column works properly only if it contains only positive values. Inserting a negative number is regarded as inserting a very large positive number. This is done to avoid precision problems when numbers “wrap” over from positive to negative and also to ensure that you do not accidentally get an AUTO_INCREMENT column that contains 0.

    For MyISAM tables, you can specify an AUTO_INCREMENT secondary column in a multiple-column key. See Section 3.6.9, “Using AUTO_INCREMENT.

    To make MySQL compatible with some ODBC applications, you can find the AUTO_INCREMENT value for the last inserted row with the following query:

    SELECT * FROM tbl_name WHERE auto_col IS NULL
    

    For information about InnoDB and AUTO_INCREMENT, see Section 13.6.5.3, “AUTO_INCREMENT Handling in InnoDB. For information about AUTO_INCREMENT and MySQL Replication, see Section 17.4.1.1, “Replication and AUTO_INCREMENT.

  • Character data types (CHAR, VARCHAR, TEXT) can include CHARACTER SET and COLLATE attributes to specify the character set and collation for the column. For details, see Section 9.1, “Character Set Support”. CHARSET is a synonym for CHARACTER SET. Example:

    CREATE TABLE t (c CHAR(20) CHARACTER SET utf8 COLLATE utf8_bin);
    

    MySQL 5.5 interprets length specifications in character column definitions in characters. (Versions before MySQL 4.1 interpreted them in bytes.) Lengths for BINARY and VARBINARY are in bytes.

  • The DEFAULT clause specifies a default value for a column. With one exception, the default value must be a constant; it cannot be a function or an expression. This means, for example, that you cannot set the default for a date column to be the value of a function such as NOW() or CURRENT_DATE. The exception is that you can specify CURRENT_TIMESTAMP as the default for a TIMESTAMP column. See Section 10.3.1.1, “TIMESTAMP Properties”.

    If a column definition includes no explicit DEFAULT value, MySQL determines the default value as described in Section 10.1.4, “Data Type Default Values”.

    BLOB and TEXT columns cannot be assigned a default value.

    CREATE TABLE fails if a date-valued default is not correct according to the NO_ZERO_IN_DATE SQL mode, even if strict SQL mode is not enabled. For example, c1 DATE DEFAULT '2010-00-00' causes CREATE TABLE to fail with Invalid default value for 'c1'.

  • A comment for a column can be specified with the COMMENT option, up to 1024 characters long (255 characters before MySQL 5.5.3). The comment is displayed by the SHOW CREATE TABLE and SHOW FULL COLUMNS statements.

  • KEY is normally a synonym for INDEX. The key attribute PRIMARY KEY can also be specified as just KEY when given in a column definition. This was implemented for compatibility with other database systems.

  • A UNIQUE index creates a constraint such that all values in the index must be distinct. An error occurs if you try to add a new row with a key value that matches an existing row. For all engines, a UNIQUE index permits multiple NULL values for columns that can contain NULL.

  • A PRIMARY KEY is a unique index where all key columns must be defined as NOT NULL. If they are not explicitly declared as NOT NULL, MySQL declares them so implicitly (and silently). A table can have only one PRIMARY KEY. If you do not have a PRIMARY KEY and an application asks for the PRIMARY KEY in your tables, MySQL returns the first UNIQUE index that has no NULL columns as the PRIMARY KEY.

    In InnoDB tables, having a long PRIMARY KEY wastes a lot of space. (See Section 13.6.11, “InnoDB Table and Index Structures”.)

  • In the created table, a PRIMARY KEY is placed first, followed by all UNIQUE indexes, and then the nonunique indexes. This helps the MySQL optimizer to prioritize which index to use and also more quickly to detect duplicated UNIQUE keys.

  • A PRIMARY KEY can be a multiple-column index. However, you cannot create a multiple-column index using the PRIMARY KEY key attribute in a column specification. Doing so only marks that single column as primary. You must use a separate PRIMARY KEY(index_col_name, ...) clause.

  • If a PRIMARY KEY or UNIQUE index consists of only one column that has an integer type, you can also refer to the column as _rowid in SELECT statements.

  • In MySQL, the name of a PRIMARY KEY is PRIMARY. For other indexes, if you do not assign a name, the index is assigned the same name as the first indexed column, with an optional suffix (_2, _3, ...) to make it unique. You can see index names for a table using SHOW INDEX FROM tbl_name. See Section 12.4.5.23, “SHOW INDEX Syntax”.

  • Some storage engines permit you to specify an index type when creating an index. The syntax for the index_type specifier is USING type_name.

    Example:

    CREATE TABLE lookup
      (id INT, INDEX USING BTREE (id))
      ENGINE = MEMORY;
    

    The preferred position is after the column list. Use of the option before the column list will no longer be recognized in a future MySQL release.

    index_option values specify additional options for an index. USING is one such option. For details about permissible index_option values, see Section 12.1.11, “CREATE INDEX Syntax”.

    For more information about indexes, see Section 7.3.1, “How MySQL Uses Indexes”.

  • In MySQL 5.5, only the MyISAM, InnoDB, and MEMORY storage engines support indexes on columns that can have NULL values. In other cases, you must declare indexed columns as NOT NULL or an error results.

  • For CHAR, VARCHAR, BINARY, and VARBINARY columns, indexes can be created that use only the leading part of column values, using col_name(length) syntax to specify an index prefix length. BLOB and TEXT columns also can be indexed, but a prefix length must be given. Prefix lengths are given in characters for nonbinary string types and in bytes for binary string types. That is, index entries consist of the first length characters of each column value for CHAR, VARCHAR, and TEXT columns, and the first length bytes of each column value for BINARY, VARBINARY, and BLOB columns. Indexing only a prefix of column values like this can make the index file much smaller. See Section 7.3.4, “Column Indexes”.

    Only the MyISAM and InnoDB storage engines support indexing on BLOB and TEXT columns. For example:

    CREATE TABLE test (blob_col BLOB, INDEX(blob_col(10)));
    

    Prefixes can be up to 1000 bytes long (767 bytes for InnoDB tables). Note that prefix limits are measured in bytes, whereas the prefix length in CREATE TABLE statements is interpreted as number of characters for nonbinary data types (CHAR, VARCHAR, TEXT). Take this into account when specifying a prefix length for a column that uses a multi-byte character set.

  • An index_col_name specification can end with ASC or DESC. These keywords are permitted for future extensions for specifying ascending or descending index value storage. Currently, they are parsed but ignored; index values are always stored in ascending order.

  • When you use ORDER BY or GROUP BY on a TEXT or BLOB column in a SELECT, the server sorts values using only the initial number of bytes indicated by the max_sort_length system variable. See Section 10.4.3, “The BLOB and TEXT Types”.

  • You can create special FULLTEXT indexes, which are used for full-text searches. Only the MyISAM storage engine supports FULLTEXT indexes. They can be created only from CHAR, VARCHAR, and TEXT columns. Indexing always happens over the entire column; column prefix indexing is not supported and any prefix length is ignored if specified. See Section 11.9, “Full-Text Search Functions”, for details of operation. A WITH PARSER clause can be specified as an index_option value to associate a parser plugin with the index if full-text indexing and searching operations need special handling. This clause is legal only for FULLTEXT indexes. See Section 23.2, “The MySQL Plugin API”, for details on creating plugins.

  • You can create SPATIAL indexes on spatial data types. Spatial types are supported only for MyISAM tables and indexed columns must be declared as NOT NULL. See Section 11.17, “Spatial Extensions”.

  • As of MySQL 5.5.3, index definitions can include an optional comment of up to 1024 characters.

  • InnoDB tables support checking of foreign key constraints. See Section 13.6, “The InnoDB Storage Engine”. Note that the FOREIGN KEY syntax in InnoDB is more restrictive than the syntax presented for the CREATE TABLE statement at the beginning of this section: The columns of the referenced table must always be explicitly named. InnoDB supports both ON DELETE and ON UPDATE actions on foreign keys. For the precise syntax, see Section 13.6.5.4, “FOREIGN KEY Constraints”.

    For other storage engines, MySQL Server parses and ignores the FOREIGN KEY and REFERENCES syntax in CREATE TABLE statements. The CHECK clause is parsed but ignored by all storage engines. See Section 1.8.5.4, “Foreign Keys”.

    Important

    For users familiar with the ANSI/ISO SQL Standard, please note that no storage engine, including InnoDB, recognizes or enforces the MATCH clause used in referential integrity constraint definitions. Use of an explicit MATCH clause will not have the specified effect, and also causes ON DELETE and ON UPDATE clauses to be ignored. For these reasons, specifying MATCH should be avoided.

    The MATCH clause in the SQL standard controls how NULL values in a composite (multiple-column) foreign key are handled when comparing to a primary key. InnoDB essentially implements the semantics defined by MATCH SIMPLE, which permit a foreign key to be all or partially NULL. In that case, the (child table) row containing such a foreign key is permitted to be inserted, and does not match any row in the referenced (parent) table. It is possible to implement other semantics using triggers.

    Additionally, MySQL and InnoDB require that the referenced columns be indexed for performance. However, the system does not enforce a requirement that the referenced columns be UNIQUE or be declared NOT NULL. The handling of foreign key references to nonunique keys or keys that contain NULL values is not well defined for operations such as UPDATE or DELETE CASCADE. You are advised to use foreign keys that reference only UNIQUE and NOT NULL keys.

    Furthermore, InnoDB does not recognize or support “inline REFERENCES specifications” (as defined in the SQL standard) where the references are defined as part of the column specification. InnoDB accepts REFERENCES clauses only when specified as part of a separate FOREIGN KEY specification. For other storage engines, MySQL Server parses and ignores foreign key specifications.

    Note

    Partitioned tables do not support foreign keys. See Section 18.5, “Restrictions and Limitations on Partitioning”, for more information.

  • There is a hard limit of 4096 columns per table, but the effective maximum may be less for a given table and depends on the factors discussed in Section E.9.2, “The Maximum Number of Columns Per Table”.

The ENGINE table option specifies the storage engine for the table.

The ENGINE table option takes the storage engine names shown in the following table.

Storage EngineDescription
ARCHIVEThe archiving storage engine. See Section 13.12, “The ARCHIVE Storage Engine”.
CSVTables that store rows in comma-separated values format. See Section 13.13, “The CSV Storage Engine”.
EXAMPLEAn example engine. See Section 13.10, “The EXAMPLE Storage Engine”.
FEDERATEDStorage engine that accesses remote tables. See Section 13.11, “The FEDERATED Storage Engine”.
HEAPThis is a synonym for MEMORY.
ISAM (OBSOLETE)Not available in MySQL 5.5. If you are upgrading to MySQL 5.5 from a previous version, you should convert any existing ISAM tables to MyISAM before performing the upgrade.
InnoDBTransaction-safe tables with row locking and foreign keys. See Section 13.6, “The InnoDB Storage Engine”.
MEMORYThe data for this storage engine is stored only in memory. See Section 13.9, “The MEMORY Storage Engine”.
MERGEA collection of MyISAM tables used as one table. Also known as MRG_MyISAM. See Section 13.8, “The MERGE Storage Engine”.
MyISAMThe binary portable storage engine that is the default storage engine used by MySQL. See Section 13.5, “The MyISAM Storage Engine”.

If a storage engine is specified that is not available, MySQL uses the default engine instead. Normally, this is MyISAM. For example, if a table definition includes the ENGINE=INNODB option but the MySQL server does not support INNODB tables, the table is created as a MyISAM table. This makes it possible to have a replication setup where you have transactional tables on the master but tables created on the slave are nontransactional (to get more speed). In MySQL 5.5, a warning occurs if the storage engine specification is not honored.

Engine substitution can be controlled by the setting of the NO_ENGINE_SUBSTITUTION SQL mode, as described in Section 5.1.7, “Server SQL Modes”.

Note

The older TYPE option was synonymous with ENGINE. TYPE was deprecated in MySQL 4.0 and removed in MySQL 5.5. When upgrading to MySQL 5.5 or later, you must convert existing applications that rely on TYPE to use ENGINE instead.

The other table options are used to optimize the behavior of the table. In most cases, you do not have to specify any of them. These options apply to all storage engines unless otherwise indicated. Options that do not apply to a given storage engine may be accepted and remembered as part of the table definition. Such options then apply if you later use ALTER TABLE to convert the table to use a different storage engine.

  • AUTO_INCREMENT

    The initial AUTO_INCREMENT value for the table. In MySQL 5.5, this works for MyISAM, MEMORY, InnoDB, and ARCHIVE tables. To set the first auto-increment value for engines that do not support the AUTO_INCREMENT table option, insert a “dummy” row with a value one less than the desired value after creating the table, and then delete the dummy row.

    For engines that support the AUTO_INCREMENT table option in CREATE TABLE statements, you can also use ALTER TABLE tbl_name AUTO_INCREMENT = N to reset the AUTO_INCREMENT value. The value cannot be set lower than the maximum value currently in the column.

  • AVG_ROW_LENGTH

    An approximation of the average row length for your table. You need to set this only for large tables with variable-size rows.

    When you create a MyISAM table, MySQL uses the product of the MAX_ROWS and AVG_ROW_LENGTH options to decide how big the resulting table is. If you don't specify either option, the maximum size for MyISAM data and index files is 256TB by default. (If your operating system does not support files that large, table sizes are constrained by the file size limit.) If you want to keep down the pointer sizes to make the index smaller and faster and you don't really need big files, you can decrease the default pointer size by setting the myisam_data_pointer_size system variable. (See Section 5.1.4, “Server System Variables”.) If you want all your tables to be able to grow above the default limit and are willing to have your tables slightly slower and larger than necessary, you can increase the default pointer size by setting this variable. Setting the value to 7 permits table sizes up to 65,536TB.

  • [DEFAULT] CHARACTER SET

    Specify a default character set for the table. CHARSET is a synonym for CHARACTER SET. If the character set name is DEFAULT, the database character set is used.

  • CHECKSUM

    Set this to 1 if you want MySQL to maintain a live checksum for all rows (that is, a checksum that MySQL updates automatically as the table changes). This makes the table a little slower to update, but also makes it easier to find corrupted tables. The CHECKSUM TABLE statement reports the checksum. (MyISAM only.)

  • [DEFAULT] COLLATE

    Specify a default collation for the table.

  • COMMENT

    A comment for the table, up to 2048 characters long (60 characters before MySQL 5.5.3).

  • CONNECTION

    The connection string for a FEDERATED table.

    Note

    Older versions of MySQL used a COMMENT option for the connection string.

  • DATA DIRECTORY, INDEX DIRECTORY

    By using DATA DIRECTORY='directory' or INDEX DIRECTORY='directory' you can specify where the MyISAM storage engine should put a table's data file and index file. The directory must be the full path name to the directory, not a relative path.

    Important

    Table-level DATA DIRECTORY and INDEX DIRECTORY options are ignored for partitioned tables. (Bug#32091)

    These options work only when you are not using the --skip-symbolic-links option. Your operating system must also have a working, thread-safe realpath() call. See Section 7.11.3.1.2, “Using Symbolic Links for Tables on Unix”, for more complete information.

    If a MyISAM table is created with no DATA DIRECTORY option, the .MYD file is created in the database directory. By default, if MyISAM finds an existing .MYD file in this case, it overwrites it. The same applies to .MYI files for tables created with no INDEX DIRECTORY option. To suppress this behavior, start the server with the --keep_files_on_create option, in which case MyISAM will not overwrite existing files and returns an error instead.

    If a MyISAM table is created with a DATA DIRECTORY or INDEX DIRECTORY option and an existing .MYD or .MYI file is found, MyISAM always returns an error. It will not overwrite a file in the specified directory.

    Important

    You cannot use path names that contain the MySQL data directory with DATA DIRECTORY or INDEX DIRECTORY. This includes partitioned tables and individual table partitions. (See Bug#32167.)

  • DELAY_KEY_WRITE

    Set this to 1 if you want to delay key updates for the table until the table is closed. See the description of the delay_key_write system variable in Section 5.1.4, “Server System Variables”. (MyISAM only.)

  • INSERT_METHOD

    If you want to insert data into a MERGE table, you must specify with INSERT_METHOD the table into which the row should be inserted. INSERT_METHOD is an option useful for MERGE tables only. Use a value of FIRST or LAST to have inserts go to the first or last table, or a value of NO to prevent inserts. See Section 13.8, “The MERGE Storage Engine”.

  • KEY_BLOCK_SIZE

    This option provides a hint to the storage engine about the size in bytes to use for index key blocks. The engine is permitted to change the value if necessary. A value of 0 indicates that the default value should be used. Individual index definitions can specify a KEY_BLOCK_SIZE value of their own to override the table value.

  • MAX_ROWS

    The maximum number of rows you plan to store in the table. This is not a hard limit, but rather a hint to the storage engine that the table must be able to store at least this many rows.

    The maximum MAX_ROWS value is 4294967295; larger values are truncated to this limit.

  • MIN_ROWS

    The minimum number of rows you plan to store in the table. The MEMORY storage engine uses this option as a hint about memory use.

  • PACK_KEYS

    PACK_KEYS takes effect only with MyISAM tables. Set this option to 1 if you want to have smaller indexes. This usually makes updates slower and reads faster. Setting the option to 0 disables all packing of keys. Setting it to DEFAULT tells the storage engine to pack only long CHAR, VARCHAR, BINARY, or VARBINARY columns.

    If you do not use PACK_KEYS, the default is to pack strings, but not numbers. If you use PACK_KEYS=1, numbers are packed as well.

    When packing binary number keys, MySQL uses prefix compression:

    • Every key needs one extra byte to indicate how many bytes of the previous key are the same for the next key.

    • The pointer to the row is stored in high-byte-first order directly after the key, to improve compression.

    This means that if you have many equal keys on two consecutive rows, all following “same” keys usually only take two bytes (including the pointer to the row). Compare this to the ordinary case where the following keys takes storage_size_for_key + pointer_size (where the pointer size is usually 4). Conversely, you get a significant benefit from prefix compression only if you have many numbers that are the same. If all keys are totally different, you use one byte more per key, if the key is not a key that can have NULL values. (In this case, the packed key length is stored in the same byte that is used to mark if a key is NULL.)

  • PASSWORD

    This option is unused. If you have a need to scramble your .frm files and make them unusable to any other MySQL server, please contact our sales department.

  • RAID_TYPE

    RAID support has been removed as of MySQL 5.0. For information on RAID, see CREATE TABLE Syntax.

  • ROW_FORMAT

    Defines how the rows should be stored. For MyISAM tables, the option value can be FIXED or DYNAMIC for static or variable-length row format. myisampack sets the type to COMPRESSED. See Section 13.5.3, “MyISAM Table Storage Formats”.

    For InnoDB tables, rows are stored in compact format (ROW_FORMAT=COMPACT) by default. The noncompact format used in older versions of MySQL can still be requested by specifying ROW_FORMAT=REDUNDANT.

    Note

    When executing a CREATE TABLE statement, if you specify a row format which is not supported by the storage engine that is used for the table, the table is created using that storage engine's default row format. The information reported in this column in response to SHOW TABLE STATUS is the actual row format used. This may differ from the value in the Create_options column because the original CREATE TABLE definition is retained during creation.

  • UNION

    UNION is used when you want to access a collection of identical MyISAM tables as one. This works only with MERGE tables. See Section 13.8, “The MERGE Storage Engine”.

    You must have SELECT, UPDATE, and DELETE privileges for the tables you map to a MERGE table.

    Note

    Formerly, all tables used had to be in the same database as the MERGE table itself. This restriction no longer applies.

partition_options can be used to control partitioning of the table created with CREATE TABLE.

Important

Not all options shown in the syntax for partition_options at the beginning of this section are available for all partitioning types. Please see the listings for the following individual types for information specific to each type, and see Chapter 18, Partitioning, for more complete information about the workings of and uses for partitioning in MySQL, as well as additional examples of table creation and other statements relating to MySQL partitioning.

If used, a partition_options clause begins with PARTITION BY. This clause contains the function that is used to determine the partition; the function returns an integer value ranging from 1 to num, where num is the number of partitions. (The maximum number of user-defined partitions which a table may contain is 1024; the number of subpartitions—discussed later in this section—is included in this maximum.) The choices that are available for this function in MySQL 5.5 are shown in the following list:

  • HASH(expr): Hashes one or more columns to create a key for placing and locating rows. expr is an expression using one or more table columns. This can be any legal MySQL expression (including MySQL functions) that yields a single integer value. For example, these are both valid CREATE TABLE statements using PARTITION BY HASH:

    CREATE TABLE t1 (col1 INT, col2 CHAR(5))
        PARTITION BY HASH(col1);
    
    CREATE TABLE t1 (col1 INT, col2 CHAR(5), col3 DATETIME)
        PARTITION BY HASH ( YEAR(col3) );
    

    You may not use either VALUES LESS THAN or VALUES IN clauses with PARTITION BY HASH.

    PARTITION BY HASH uses the remainder of expr divided by the number of partitions (that is, the modulus). For examples and additional information, see Section 18.2.4, “HASH Partitioning”.

    The LINEAR keyword entails a somewhat different algorithm. In this case, the number of the partition in which a row is stored is calculated as the result of one or more logical AND operations. For discussion and examples of linear hashing, see Section 18.2.4.1, “LINEAR HASH Partitioning”.

  • KEY(column_list): This is similar to HASH, except that MySQL supplies the hashing function so as to guarantee an even data distribution. The column_list argument is simply a list of table columns. This example shows a simple table partitioned by key, with 4 partitions:

    CREATE TABLE tk (col1 INT, col2 CHAR(5), col3 DATE)
        PARTITION BY KEY(col3)
        PARTITIONS 4;
    

    For tables that are partitioned by key, you can employ linear partitioning by using the LINEAR keyword. This has the same effect as with tables that are partitioned by HASH. That is, the partition number is found using the & operator rather than the modulus (see Section 18.2.4.1, “LINEAR HASH Partitioning”, and Section 18.2.5, “KEY Partitioning”, for details). This example uses linear partitioning by key to distribute data between 5 partitions:

    CREATE TABLE tk (col1 INT, col2 CHAR(5), col3 DATE)
        PARTITION BY LINEAR KEY(col3)
        PARTITIONS 5;
    

    You may not use either VALUES LESS THAN or VALUES IN clauses with PARTITION BY KEY.

  • RANGE: In this case, expr shows a range of values using a set of VALUES LESS THAN operators. When using range partitioning, you must define at least one partition using VALUES LESS THAN. You cannot use VALUES IN with range partitioning.

    When used with a table partitioned by RANGE, VALUES LESS THAN must be used with either an integer literal value or an expression that evaluates to a single integer value. In MySQL 5.5, this limitation can be overcome in a table that is defined using PARTITION BY RANGE COLUMNS, as described later in this section.

    Suppose that you have a table that you wish to partition on a column containing year values, according to the following scheme.

    Partition Number:Years Range:
    01990 and earlier
    11991 to 1994
    21995 to 1998
    31999 to 2002
    42003 to 2005
    52006 and later

    A table implementing such a partitioning scheme can be realized by the CREATE TABLE statement shown here:

    CREATE TABLE t1 (
        year_col  INT,
        some_data INT
    )
    PARTITION BY RANGE (year_col) (
        PARTITION p0 VALUES LESS THAN (1991),
        PARTITION p1 VALUES LESS THAN (1995),
        PARTITION p2 VALUES LESS THAN (1999),
        PARTITION p3 VALUES LESS THAN (2002),
        PARTITION p4 VALUES LESS THAN (2006),
        PARTITION p5 VALUES LESS THAN MAXVALUE
    );
    

    PARTITION ... VALUES LESS THAN ... statements work in a consecutive fashion. VALUES LESS THAN MAXVALUE works to specify “leftover” values that are greater than the maximum value otherwise specified.

    Note that VALUES LESS THAN clauses work sequentially in a manner similar to that of the case portions of a switch ... case block (as found in many programming languages such as C, Java, and PHP). That is, the clauses must be arranged in such a way that the upper limit specified in each successive VALUES LESS THAN is greater than that of the previous one, with the one referencing MAXVALUE coming last of all in the list.

  • RANGE COLUMNS(column_list): This variant on RANGE was introduced in MySQL 5.5.0 to facilitate partition pruning for queries using range conditions on multiple columns (that is, having conditions such as WHERE a = 1 AND b < 10 or WHERE a = 1 AND b = 10 AND c < 10). It enables you to specify value ranges in multiple columns by using a list of columns in the COLUMNS clause and a set of column values in each PARTITION ... VALUES LESS THAN (value_list) partition definition clause. (In the simplest case, this set consists of a single column.) The maximum number of columns that can be referenced in the column_list and value_list is 16.

    The column_list used in the COLUMNS clause may contain only names of columns; each column in the list must be one of the following MySQL data types: the integer types; the string types; and time or date column types. Columns using BLOB, TEXT, SET, ENUM, BIT, or spatial data types are not permitted; columns that use floating-point number types are also not permitted. You also may not use functions or arithmetic expressions in the COLUMNS clause.

    The VALUES LESS THAN clause used in a partition definition must specify a literal value for each column that appears in the COLUMNS() clause; that is, the list of values used for each VALUES LESS THAN clause must contain the same number of values as there are columns listed in the COLUMNS clause. An attempt to use more or fewer values in a VALUES LESS THAN clause than there are in the COLUMNS clause causes the statement to fail with the error Inconsistency in usage of column lists for partitioning.... You cannot use NULL for any value appearing in VALUES LESS THAN. It is possible to use MAXVALUE more than once for a given column other than the first, as shonw in this example:

    CREATE TABLE rc (
        a INT NOT NULL, 
        b INT NOT NULL
    )
    PARTITION BY RANGE COLUMNS(a,b) (
        PARTITION p0 VALUES LESS THAN (10,5),
        PARTITION p1 VALUES LESS THAN (20,10),
        PARTITION p2 VALUES LESS THAN (MAXVALUE,15),
        PARTITION p3 VALUES LESS THAN (MAXVALUE,MAXVALUE)
    );
    

    Each value used in a VALUES LESS THAN value list must match the type of the corresponding column exactly; no conversion is made. For example, you cannot use the string "1" for a value that matches a column that uses an integer type (you must use the numeral 1 instead), nor can you use the numeral 1 for a value that matches a column that uses a string type (in such a case, you must use a quoted string: "1").

    For more information, see Section 18.2.1, “RANGE Partitioning”, and Section 18.4, “Partition Pruning”.

  • LIST(expr): This is useful when assigning partitions based on a table column with a restricted set of possible values, such as a state or country code. In such a case, all rows pertaining to a certain state or country can be assigned to a single partition, or a partition can be reserved for a certain set of states or countries. It is similar to RANGE, except that only VALUES IN may be used to specify permissible values for each partition.

    VALUES IN is used with a list of values to be matched. For instance, you could create a partitioning scheme such as the following:

    CREATE TABLE client_firms (
        id   INT,
        name VARCHAR(35)
    )
    PARTITION BY LIST (id) (
        PARTITION r0 VALUES IN (1, 5, 9, 13, 17, 21),
        PARTITION r1 VALUES IN (2, 6, 10, 14, 18, 22),
        PARTITION r2 VALUES IN (3, 7, 11, 15, 19, 23),
        PARTITION r3 VALUES IN (4, 8, 12, 16, 20, 24)
    );
    

    When using list partitioning, you must define at least one partition using VALUES IN. You cannot use VALUES LESS THAN with PARTITION BY LIST.

    Note

    For tables partitioned by LIST, the value list used with VALUES IN must consist of integer values only. In MySQL 5.5, you can overcome this limitation using partitioning by LIST COLUMNS, which is described later in this section.

  • LIST COLUMNS(column_list): This variant on LIST was introduced in MySQL 5.5.0 to facilitate partition pruning for queries using comparison conditions on multiple columns (that is, having conditions such as WHERE a = 5 AND b = 5 or WHERE a = 1 AND b = 10 AND c = 5). It enables you to specify values in multiple columns by using a list of columns in the COLUMNS clause and a set of column values in each PARTITION ... VALUES IN (value_list) partition definition clause.

    The rules governing regarding data types for the column list used in LIST COLUMNS(column_list) and the value list used in VALUES IN(value_list) are the as those for the column list used in RANGE COLUMNS(column_list) and the value list used in VALUES LESS THAN(value_list), respectively, except that in the VALUES IN clause, MAXVALUE is not permitted, and you may use NULL.

    There is one important difference between the list of values used for VALUES IN with PARTITION BY LIST COLUMNS as opposed to when it is used with PARTITION BY LIST. When used with PARTITION BY LIST COLUMNS, each element in the VALUES IN clause must be a set of column values; the number of values in each set must be the same as the number of columns used in the COLUMNS clause, and the data types of these values must match those of the columns (and occur in the same order). In the simplest case, the set consists of a single column. The maximum number of columns that can be used in the column_list and in the elements making up the value_list is 16.

    The table defined by the following CREATE TABLE statement provides an example of a table using LIST COLUMNS partitioning:

    CREATE TABLE lc (
        a INT NULL, 
        b INT NULL
    )
    PARTITION BY LIST COLUMNS(a,b) (
        PARTITION p0 VALUES IN( (0,0), (NULL,NULL) ),
        PARTITION p1 VALUES IN( (0,1), (0,2), (0,3), (1,1), (1,2) ),
        PARTITION p2 VALUES IN( (1,0), (2,0), (2,1), (3,0), (3,1) ),
        PARTITION p3 VALUES IN( (1,3), (2,2), (2,3), (3,2), (3,3) )
    );
    
  • The number of partitions may optionally be specified with a PARTITIONS num clause, where num is the number of partitions. If both this clause and any PARTITION clauses are used, num must be equal to the total number of any partitions that are declared using PARTITION clauses.

    Note

    Whether or not you use a PARTITIONS clause in creating a table that is partitioned by RANGE or LIST, you must still include at least one PARTITION VALUES clause in the table definition (see below).

  • A partition may optionally be divided into a number of subpartitions. This can be indicated by using the optional SUBPARTITION BY clause. Subpartitioning may be done by HASH or KEY. Either of these may be LINEAR. These work in the same way as previously described for the equivalent partitioning types. (It is not possible to subpartition by LIST or RANGE.)

    The number of subpartitions can be indicated using the SUBPARTITIONS keyword followed by an integer value.

  • Rigorous checking of the value used in PARTITIONS or SUBPARTITIONS clauses is applied and this value must adhere to the following rules:

    • The value must be a positive, nonzero integer.

    • No leading zeros are permitted.

    • The value must be an integer literal, and cannot not be an expression. For example, PARTITIONS 0.2E+01 is not permitted, even though 0.2E+01 evaluates to 2. (Bug#15890)

Note

The expression (expr) used in a PARTITION BY clause cannot refer to any columns not in the table being created; such references are specifically not permitted and cause the statement to fail with an error. (Bug#29444)

Each partition may be individually defined using a partition_definition clause. The individual parts making up this clause are as follows:

  • PARTITION partition_name: This specifies a logical name for the partition.

  • A VALUES clause: For range partitioning, each partition must include a VALUES LESS THAN clause; for list partitioning, you must specify a VALUES IN clause for each partition. This is used to determine which rows are to be stored in this partition. See the discussions of partitioning types in Chapter 18, Partitioning, for syntax examples.

  • An optional COMMENT clause may be used to specify a string that describes the partition. Example:

    COMMENT = 'Data for the years previous to 1999'
    
  • DATA DIRECTORY and INDEX DIRECTORY may be used to indicate the directory where, respectively, the data and indexes for this partition are to be stored. Both the data_dir and the index_dir must be absolute system path names. Example:

    CREATE TABLE th (id INT, name VARCHAR(30), adate DATE)
    PARTITION BY LIST(YEAR(adate))
    (
      PARTITION p1999 VALUES IN (1995, 1999, 2003)
        DATA DIRECTORY = '/var/appdata/95/data'
        INDEX DIRECTORY = '/var/appdata/95/idx',
      PARTITION p2000 VALUES IN (1996, 2000, 2004)
        DATA DIRECTORY = '/var/appdata/96/data'
        INDEX DIRECTORY = '/var/appdata/96/idx',
      PARTITION p2001 VALUES IN (1997, 2001, 2005)
        DATA DIRECTORY = '/var/appdata/97/data'
        INDEX DIRECTORY = '/var/appdata/97/idx',
      PARTITION p2000 VALUES IN (1998, 2002, 2006)
        DATA DIRECTORY = '/var/appdata/98/data'
        INDEX DIRECTORY = '/var/appdata/98/idx'
    );
    

    DATA DIRECTORY and INDEX DIRECTORY behave in the same way as in the CREATE TABLE statement's table_option clause as used for MyISAM tables.

    One data directory and one index directory may be specified per partition. If left unspecified, the data and indexes are stored by default in the table's database directory.

    On Windows, the DATA DIRECTORY and INDEX DIRECTORY options are not supported for individual partitions or subpartitions. These options are ignored on Windows, except that a warning is generated. (Bug#30459)

    Note

    The DATA DIRECTORY and INDEX DIRECTORY options are ignored for creating partitioned tables if NO_DIR_IN_CREATE is in effect. (Bug#24633)

  • MAX_ROWS and MIN_ROWS may be used to specify, respectively, the maximum and minimum number of rows to be stored in the partition. The values for max_number_of_rows and min_number_of_rows must be positive integers. As with the table-level options with the same names, these act only as “suggestions” to the server and are not hard limits.

  • The partitioning handler accepts a [STORAGE] ENGINE option for both PARTITION and SUBPARTITION. Currently, the only way in which this can be used is to set all partitions or all subpartitions to the same storage engine, and an attempt to set different storage engines for partitions or subpartitions in the same table will give rise to the error ERROR 1469 (HY000): The mix of handlers in the partitions is not permitted in this version of MySQL. We expect to lift this restriction on partitioning in a future MySQL release.

  • The partition definition may optionally contain one or more subpartition_definition clauses. Each of these consists at a minimum of the SUBPARTITION name, where name is an identifier for the subpartition. Except for the replacement of the PARTITION keyword with SUBPARTITION, the syntax for a subpartition definition is identical to that for a partition definition.

    Subpartitioning must be done by HASH or KEY, and can be done only on RANGE or LIST partitions. See Section 18.2.6, “Subpartitioning”.

Partitions can be modified, merged, added to tables, and dropped from tables. For basic information about the MySQL statements to accomplish these tasks, see Section 12.1.6, “ALTER TABLE Syntax”. For more detailed descriptions and examples, see Section 18.3, “Partition Management”.

Important

The original CREATE TABLE statement, including all specifications and table options are stored by MySQL when the table is created. The information is retained so that if you change storage engines, collations or other settings using an ALTER TABLE statement, the original table options specified are retained. This enables you to change between InnoDB and MyISAM table types even though the row formats supported by the two engines are different.

Because the text of the original statement is retained, but due to the way that certain values and options may be silently reconfigured (such as the ROW_FORMAT), the active table definition (accessible through DESCRIBE or with SHOW TABLE STATUS) and the table creation string (accessible through SHOW CREATE TABLE) will report different values.

You can create one table from another by adding a SELECT statement at the end of the CREATE TABLE statement:

CREATE TABLE new_tbl SELECT * FROM orig_tbl;

For more information, see Section 12.1.14.1, “CREATE TABLE ... SELECT Syntax”.

Use LIKE to create an empty table based on the definition of another table, including any column attributes and indexes defined in the original table:

CREATE TABLE new_tbl LIKE orig_tbl;

The copy is created using the same version of the table storage format as the original table. The SELECT privilege is required on the original table.

LIKE works only for base tables, not for views.

Important

Beginning with MySQL 5.5.3, you cannot execute CREATE TABLE or CREATE TABLE ... LIKE while a LOCK TABLES statement is in effect.

CREATE TABLE ... LIKE does not preserve any DATA DIRECTORY or INDEX DIRECTORY table options that were specified for the original table, or any foreign key definitions.

If the original table is a TEMPORARY table, CREATE TABLE ... LIKE does not preserve TEMPORARY. To create a TEMPORARY destination table, use CREATE TEMPORARY TABLE ... LIKE.

12.1.14.1. CREATE TABLE ... SELECT Syntax

You can create one table from another by adding a SELECT statement at the end of the CREATE TABLE statement:

CREATE TABLE new_tbl SELECT * FROM orig_tbl;

MySQL creates new columns for all elements in the SELECT. For example:

mysql> CREATE TABLE test (a INT NOT NULL AUTO_INCREMENT,
    ->        PRIMARY KEY (a), KEY(b))
    ->        ENGINE=MyISAM SELECT b,c FROM test2;

This creates a MyISAM table with three columns, a, b, and c. Notice that the columns from the SELECT statement are appended to the right side of the table, not overlapped onto it. Take the following example:

mysql> SELECT * FROM foo;
+---+
| n |
+---+
| 1 |
+---+

mysql> CREATE TABLE bar (m INT) SELECT n FROM foo;
Query OK, 1 row affected (0.02 sec)
Records: 1  Duplicates: 0  Warnings: 0

mysql> SELECT * FROM bar;
+------+---+
| m    | n |
+------+---+
| NULL | 1 |
+------+---+
1 row in set (0.00 sec)

For each row in table foo, a row is inserted in bar with the values from foo and default values for the new columns.

In a table resulting from CREATE TABLE ... SELECT, columns named only in the CREATE TABLE part come first. Columns named in both parts or only in the SELECT part come after that. The data type of SELECT columns can be overridden by also specifying the column in the CREATE TABLE part.

If any errors occur while copying the data to the table, it is automatically dropped and not created.

You can precede the SELECT by IGNORE or REPLACE to indicate how to handle rows that duplicate unique key values. With IGNORE, new rows that duplicate an existing row on a unique key value are discarded. With REPLACE, new rows replace rows that have the same unique key value. If neither IGNORE nor REPLACE is specified, duplicate unique key values result in an error.

CREATE TABLE ... SELECT does not automatically create any indexes for you. This is done intentionally to make the statement as flexible as possible. If you want to have indexes in the created table, you should specify these before the SELECT statement:

mysql> CREATE TABLE bar (UNIQUE (n)) SELECT n FROM foo;

Some conversion of data types might occur. For example, the AUTO_INCREMENT attribute is not preserved, and VARCHAR columns can become CHAR columns. Retrained attributes are NULL (or NOT NULL) and, for those columns that have them, CHARACTER SET, COLLATION, COMMENT, and the DEFAULT clause.

When creating a table with CREATE TABLE ... SELECT, make sure to alias any function calls or expressions in the query. If you do not, the CREATE statement might fail or result in undesirable column names.

CREATE TABLE artists_and_works
  SELECT artist.name, COUNT(work.artist_id) AS number_of_works
  FROM artist LEFT JOIN work ON artist.id = work.artist_id
  GROUP BY artist.id;

You can also explicitly specify the data type for a generated column:

CREATE TABLE foo (a TINYINT NOT NULL) SELECT b+1 AS a FROM bar;

For CREATE TABLE ... SELECT, if IF NOT EXISTS is given and the destination table already exists, the result is version dependent. Before MySQL 5.5.6, MySQL handles the statement as follows:

  • The table definition given in the CREATE TABLE part is ignored. No error occurs, even if the definition does not match that of the existing table. MySQL attempts to insert the rows from the SELECT part anyway.

  • If there is a mismatch between the number of columns in the table and the number of columns produced by the SELECT part, the selected values are assigned to the rightmost columns. For example, if the table contains n columns and the SELECT produces m columns, where m < n, the selected values are assigned to the m rightmost columns in the table. Each of the initial nm columns is assigned its default value, either that specified explicitly in the column definition or the implicit column data type default if the definition contains no default. If the SELECT part produces too many columns (m > n), an error occurs.

  • If strict SQL mode is enabled and any of these initial columns do not have an explicit default value, the statement fails with an error.

The following example illustrates IF NOT EXISTS handling:

mysql> CREATE TABLE t1 (i1 INT DEFAULT 0, i2 INT, i3 INT, i4 INT);
Query OK, 0 rows affected (0.05 sec)

mysql> CREATE TABLE IF NOT EXISTS t1 (c1 CHAR(10)) SELECT 1, 2;
Query OK, 1 row affected, 1 warning (0.01 sec)
Records: 1  Duplicates: 0  Warnings: 0

mysql> SELECT * FROM t1;
+------+------+------+------+
| i1   | i2   | i3   | i4   |
+------+------+------+------+
|    0 | NULL |    1 |    2 |
+------+------+------+------+
1 row in set (0.00 sec)

As of MySQL 5.5.6, handling of CREATE TABLE IF NOT EXISTS ... SELECT statements was changed for the case that the destination table already exists. This change also involves a change in MySQL 5.1 beginning with 5.1.51.

  • Previously, for CREATE TABLE IF NOT EXISTS ... SELECT, MySQL produced a warning that the table exists, but inserted the rows and wrote the statement to the binary log anyway. By contrast, CREATE TABLE ... SELECT (without IF NOT EXISTS) failed with an error, but MySQL inserted no rows and did not write the statement to the binary log.

  • MySQL now handles both statements the same way when the destination table exists, in that neither statement inserts rows or is written to the binary log. The difference between them is that MySQL produces a warning when IF NOT EXISTS is present and an error when it is not.

This change means that, for the preceding example, the CREATE TABLE IF NOT EXISTS ... SELECT statement inserts nothing into the destination table as of MySQL 5.5.6.

This change in handling of IF NOT EXISTS results in an incompatibility for statement-based replication from a MySQL 5.1 master with the original behavior and a MySQL 5.5 slave with the new behavior. Suppose that CREATE TABLE IF NOT EXISTS ... SELECT is executed on the master and the destination table exists. The result is that rows are inserted on the master but not on the slave. (Row-based replication does not have this problem.)

To address this issue, statement-based binary logging for CREATE TABLE IF NOT EXISTS ... SELECT is changed in MySQL 5.1 as of 5.1.51:

This change provides forward compatibility for statement-based replication from MySQL 5.1 to 5.5 because when the destination table exists, the rows will be inserted on both the master and slave. To take advantage of this compatibility measure, the 5.1 server must be at least 5.1.51 and the 5.5 server must be at least 5.5.6.

To upgrade an existing 5.1-to-5.5 replication scenario, upgrade the master first to 5.1.51 or higher. Note that this differs from the usual replication upgrade advice of upgrading the slave first.

A workaround for applications that wish to achieve the original effect (rows inserted regardless of whether the destination table exists) is to use CREATE TABLE IF NOT EXISTS and INSERT ... SELECT statements rather than CREATE TABLE IF NOT EXISTS ... SELECT statements.

Along with the change just described, the following related change was made: Previously, if an existing view was named as the destination table for CREATE TABLE IF NOT EXISTS ... SELECT, rows were inserted into the underlying base table and the statement was written to the binary log. As of MySQL 5.1.51 and 5.5.6, nothing is inserted or logged.

To ensure that the binary log can be used to re-create the original tables, MySQL does not permit concurrent inserts during CREATE TABLE ... SELECT.

12.1.14.2. Silent Column Specification Changes

In some cases, MySQL silently changes column specifications from those given in a CREATE TABLE or ALTER TABLE statement. These might be changes to a data type, to attributes associated with a data type, or to an index specification.

  • TIMESTAMP display sizes are discarded.

    Also note that TIMESTAMP columns are NOT NULL by default.

  • Columns that are part of a PRIMARY KEY are made NOT NULL even if not declared that way.

  • Trailing spaces are automatically deleted from ENUM and SET member values when the table is created.

  • MySQL maps certain data types used by other SQL database vendors to MySQL types. See Section 10.8, “Using Data Types from Other Database Engines”.

  • If you include a USING clause to specify an index type that is not legal for a given storage engine, but there is another index type available that the engine can use without affecting query results, the engine uses the available type.

  • If strict SQL mode is not enabled, a VARCHAR column with a length specification greater than 65535 is converted to TEXT, and a VARBINARY column with a length specification greater than 65535 is converted to BLOB. Otherwise, an error occurs in either of these cases.

  • Specifying the CHARACTER SET binary attribute for a character data type causes the column to be created as the corresponding binary data type: CHAR becomes BINARY, VARCHAR becomes VARBINARY, and TEXT becomes BLOB. For the ENUM and SET data types, this does not occur; they are created as declared. Suppose that you specify a table using this definition:

    CREATE TABLE t
    (
      c1 VARCHAR(10) CHARACTER SET binary,
      c2 TEXT CHARACTER SET binary,
      c3 ENUM('a','b','c') CHARACTER SET binary
    );
    

    The resulting table has this definition:

    CREATE TABLE t
    (
      c1 VARBINARY(10),
      c2 BLOB,
      c3 ENUM('a','b','c') CHARACTER SET binary
    );
    

To see whether MySQL used a data type other than the one you specified, issue a DESCRIBE or SHOW CREATE TABLE statement after creating or altering the table.

Certain other data type changes can occur if you compress a table using myisampack. See Section 13.5.3.3, “Compressed Table Characteristics”.

12.1.15. CREATE TRIGGER Syntax

CREATE
    [DEFINER = { user | CURRENT_USER }]
    TRIGGER trigger_name trigger_time trigger_event
    ON tbl_name FOR EACH ROW trigger_body

This statement creates a new trigger. A trigger is a named database object that is associated with a table, and that activates when a particular event occurs for the table. The trigger becomes associated with the table named tbl_name, which must refer to a permanent table. You cannot associate a trigger with a TEMPORARY table or a view.

CREATE TRIGGER requires the TRIGGER privilege for the table associated with the trigger. The statement might also require the SUPER privilege, depending on the DEFINER value, as described later in this section. If binary logging is enabled, CREATE TRIGGER might require the SUPER privilege, as described in Section 19.7, “Binary Logging of Stored Programs”.

The DEFINER clause determines the security context to be used when checking access privileges at trigger activation time. See later in this section for more information.

trigger_time is the trigger action time. It can be BEFORE or AFTER to indicate that the trigger activates before or after each row to be modified.

trigger_event indicates the kind of statement that activates the trigger. The trigger_event can be one of the following:

It is important to understand that the trigger_event does not represent a literal type of SQL statement that activates the trigger so much as it represents a type of table operation. For example, an INSERT trigger is activated by not only INSERT statements but also LOAD DATA statements because both statements insert rows into a table.

A potentially confusing example of this is the INSERT INTO ... ON DUPLICATE KEY UPDATE ... syntax: a BEFORE INSERT trigger will activate for every row, followed by either an AFTER INSERT trigger or both the BEFORE UPDATE and AFTER UPDATE triggers, depending on whether there was a duplicate key for the row.

There cannot be two triggers for a given table that have the same trigger action time and event. For example, you cannot have two BEFORE UPDATE triggers for a table. But you can have a BEFORE UPDATE and a BEFORE INSERT trigger, or a BEFORE UPDATE and an AFTER UPDATE trigger.

trigger_body is the statement to execute when the trigger activates. If you want to execute multiple statements, use the BEGIN ... END compound statement construct. This also enables you to use the same statements that are permissible within stored routines. See Section 12.7.1, “BEGIN ... END Compound Statement Syntax”. Some statements are not permitted in triggers; see Section E.1, “Restrictions on Stored Routines, Triggers, and Events”.

You can refer to columns in the subject table (the table associated with the trigger) by using the aliases OLD and NEW. OLD.col_name refers to a column of an existing row before it is updated or deleted. NEW.col_name refers to the column of a new row to be inserted or an existing row after it is updated.

MySQL stores the sql_mode system variable setting that is in effect at the time a trigger is created, and always executes the trigger with this setting in force, regardless of the server SQL mode in effect when the event begins executing.

Note

Currently, cascaded foreign key actions do not activate triggers.

The DEFINER clause specifies the MySQL account to be used when checking access privileges at trigger activation time. If a user value is given, it should be a MySQL account specified as 'user_name'@'host_name' (the same format used in the GRANT statement), CURRENT_USER, or CURRENT_USER(). The default DEFINER value is the user who executes the CREATE TRIGGER statement. This is the same as specifying DEFINER = CURRENT_USER explicitly.

If you specify the DEFINER clause, these rules determine the legal DEFINER user values:

  • If you do not have the SUPER privilege, the only legal user value is your own account, either specified literally or by using CURRENT_USER. You cannot set the definer to some other account.

  • If you have the SUPER privilege, you can specify any syntactically legal account name. If the account does not actually exist, a warning is generated.

  • Although it is possible to create a trigger with a nonexistent DEFINER account, it is not a good idea for such triggers to be activated until the account actually does exist. Otherwise, the behavior with respect to privilege checking is undefined.

MySQL takes the DEFINER user into account when checking trigger privileges as follows:

  • At CREATE TRIGGER time, the user who issues the statement must have the TRIGGER privilege.

  • At trigger activation time, privileges are checked against the DEFINER user. This user must have these privileges:

    • The TRIGGER privilege.

    • The SELECT privilege for the subject table if references to table columns occur using OLD.col_name or NEW.col_name in the trigger definition.

    • The UPDATE privilege for the subject table if table columns are targets of SET NEW.col_name = value assignments in the trigger definition.

    • Whatever other privileges normally are required for the statements executed by the trigger.

For more information about trigger security, see Section 19.6, “Access Control for Stored Programs and Views”.

Within a trigger, the CURRENT_USER() function returns the account used to check privileges at trigger activation time. This is the DEFINER user, not the user whose actions caused the trigger to be activated. For information about user auditing within triggers, see Section 5.5.10, “Auditing MySQL Account Activity”.

If you use LOCK TABLES to lock a table that has triggers, the tables used within the trigger are also locked, as described in Section 12.3.5.2, “LOCK TABLES and Triggers”.

In MySQL 5.5, you can write triggers containing direct references to tables by name, such as the trigger named testref shown in this example:

CREATE TABLE test1(a1 INT);
CREATE TABLE test2(a2 INT);
CREATE TABLE test3(a3 INT NOT NULL AUTO_INCREMENT PRIMARY KEY);
CREATE TABLE test4(
  a4 INT NOT NULL AUTO_INCREMENT PRIMARY KEY,
  b4 INT DEFAULT 0
);

delimiter |

CREATE TRIGGER testref BEFORE INSERT ON test1
  FOR EACH ROW BEGIN
    INSERT INTO test2 SET a2 = NEW.a1;
    DELETE FROM test3 WHERE a3 = NEW.a1;
    UPDATE test4 SET b4 = b4 + 1 WHERE a4 = NEW.a1;
  END;
|

delimiter ;

INSERT INTO test3 (a3) VALUES
  (NULL), (NULL), (NULL), (NULL), (NULL),
  (NULL), (NULL), (NULL), (NULL), (NULL);

INSERT INTO test4 (a4) VALUES
  (0), (0), (0), (0), (0), (0), (0), (0), (0), (0);

Suppose that you insert the following values into table test1 as shown here:

mysql> INSERT INTO test1 VALUES 
    -> (1), (3), (1), (7), (1), (8), (4), (4);
Query OK, 8 rows affected (0.01 sec)
Records: 8  Duplicates: 0  Warnings: 0

As a result, the data in the four tables will be as follows:

mysql> SELECT * FROM test1;
+------+
| a1   |
+------+
|    1 |
|    3 |
|    1 |
|    7 |
|    1 |
|    8 |
|    4 |
|    4 |
+------+
8 rows in set (0.00 sec)

mysql> SELECT * FROM test2;
+------+
| a2   |
+------+
|    1 |
|    3 |
|    1 |
|    7 |
|    1 |
|    8 |
|    4 |
|    4 |
+------+
8 rows in set (0.00 sec)

mysql> SELECT * FROM test3;
+----+
| a3 |
+----+
|  2 |
|  5 |
|  6 |
|  9 |
| 10 |
+----+
5 rows in set (0.00 sec)

mysql> SELECT * FROM test4;
+----+------+
| a4 | b4   |
+----+------+
|  1 |    3 |
|  2 |    0 |
|  3 |    1 |
|  4 |    2 |
|  5 |    0 |
|  6 |    0 |
|  7 |    1 |
|  8 |    1 |
|  9 |    0 |
| 10 |    0 |
+----+------+
10 rows in set (0.00 sec)

12.1.16. CREATE VIEW Syntax

CREATE
    [OR REPLACE]
    [ALGORITHM = {UNDEFINED | MERGE | TEMPTABLE}]
    [DEFINER = { user | CURRENT_USER }]
    [SQL SECURITY { DEFINER | INVOKER }]
    VIEW view_name [(column_list)]
    AS select_statement
    [WITH [CASCADED | LOCAL] CHECK OPTION]

The CREATE VIEW statement creates a new view, or replaces an existing one if the OR REPLACE clause is given. If the view does not exist, CREATE OR REPLACE VIEW is the same as CREATE VIEW. If the view does exist, CREATE OR REPLACE VIEW is the same as ALTER VIEW.

The select_statement is a SELECT statement that provides the definition of the view. (When you select from the view, you select in effect using the SELECT statement.) select_statement can select from base tables or other views.

The view definition is “frozen” at creation time, so changes to the underlying tables afterward do not affect the view definition. For example, if a view is defined as SELECT * on a table, new columns added to the table later do not become part of the view.

The ALGORITHM clause affects how MySQL processes the view. The DEFINER and SQL SECURITY clauses specify the security context to be used when checking access privileges at view invocation time. The WITH CHECK OPTION clause can be given to constrain inserts or updates to rows in tables referenced by the view. These clauses are described later in this section.

The CREATE VIEW statement requires the CREATE VIEW privilege for the view, and some privilege for each column selected by the SELECT statement. For columns used elsewhere in the SELECT statement you must have the SELECT privilege. If the OR REPLACE clause is present, you must also have the DROP privilege for the view. CREATE VIEW might also require the SUPER privilege, depending on the DEFINER value, as described later in this section.

When a view is referenced, privilege checking occurs as described later in this section.

A view belongs to a database. By default, a new view is created in the default database. To create the view explicitly in a given database, specify the name as db_name.view_name when you create it:

mysql> CREATE VIEW test.v AS SELECT * FROM t;

Within a database, base tables and views share the same namespace, so a base table and a view cannot have the same name.

Columns retrieved by the SELECT statement can be simple references to table columns. They can also be expressions that use functions, constant values, operators, and so forth.

Views must have unique column names with no duplicates, just like base tables. By default, the names of the columns retrieved by the SELECT statement are used for the view column names. To define explicit names for the view columns, the optional column_list clause can be given as a list of comma-separated identifiers. The number of names in column_list must be the same as the number of columns retrieved by the SELECT statement.

Unqualified table or view names in the SELECT statement are interpreted with respect to the default database. A view can refer to tables or views in other databases by qualifying the table or view name with the proper database name.

A view can be created from many kinds of SELECT statements. It can refer to base tables or other views. It can use joins, UNION, and subqueries. The SELECT need not even refer to any tables. The following example defines a view that selects two columns from another table, as well as an expression calculated from those columns:

mysql> CREATE TABLE t (qty INT, price INT);
mysql> INSERT INTO t VALUES(3, 50);
mysql> CREATE VIEW v AS SELECT qty, price, qty*price AS value FROM t;
mysql> SELECT * FROM v;
+------+-------+-------+
| qty  | price | value |
+------+-------+-------+
|    3 |    50 |   150 |
+------+-------+-------+

A view definition is subject to the following restrictions:

  • The SELECT statement cannot contain a subquery in the FROM clause.

  • The SELECT statement cannot refer to system or user variables.

  • Within a stored program, the definition cannot refer to program parameters or local variables.

  • The SELECT statement cannot refer to prepared statement parameters.

  • Any table or view referred to in the definition must exist. However, after a view has been created, it is possible to drop a table or view that the definition refers to. In this case, use of the view results in an error. To check a view definition for problems of this kind, use the CHECK TABLE statement.

  • The definition cannot refer to a TEMPORARY table, and you cannot create a TEMPORARY view.

  • Any tables named in the view definition must exist at definition time.

  • You cannot associate a trigger with a view.

  • Aliases for column names in the SELECT statement are checked against the maximum column length of 64 characters (not the maximum alias length of 256 characters).

ORDER BY is permitted in a view definition, but it is ignored if you select from a view using a statement that has its own ORDER BY.

For other options or clauses in the definition, they are added to the options or clauses of the statement that references the view, but the effect is undefined. For example, if a view definition includes a LIMIT clause, and you select from the view using a statement that has its own LIMIT clause, it is undefined which limit applies. This same principle applies to options such as ALL, DISTINCT, or SQL_SMALL_RESULT that follow the SELECT keyword, and to clauses such as INTO, FOR UPDATE, LOCK IN SHARE MODE, and PROCEDURE.

If you create a view and then change the query processing environment by changing system variables, that may affect the results that you get from the view:

mysql> CREATE VIEW v (mycol) AS SELECT 'abc';
Query OK, 0 rows affected (0.01 sec)

mysql> SET sql_mode = '';
Query OK, 0 rows affected (0.00 sec)

mysql> SELECT "mycol" FROM v;
+-------+
| mycol |
+-------+
| mycol |
+-------+
1 row in set (0.01 sec)

mysql> SET sql_mode = 'ANSI_QUOTES';
Query OK, 0 rows affected (0.00 sec)

mysql> SELECT "mycol" FROM v;
+-------+
| mycol |
+-------+
| abc   |
+-------+
1 row in set (0.00 sec)

The DEFINER and SQL SECURITY clauses determine which MySQL account to use when checking access privileges for the view when a statement is executed that references the view. The legal SQL SECURITY characteristic values are DEFINER and INVOKER. These indicate that the required privileges must be held by the user who defined or invoked the view, respectively. The default SQL SECURITY value is DEFINER.

If a user value is given for the DEFINER clause, it should be a MySQL account specified as 'user_name'@'host_name' (the same format used in the GRANT statement), CURRENT_USER, or CURRENT_USER(). The default DEFINER value is the user who executes the CREATE VIEW statement. This is the same as specifying DEFINER = CURRENT_USER explicitly.

If you specify the DEFINER clause, these rules determine the legal DEFINER user values:

  • If you do not have the SUPER privilege, the only legal user value is your own account, either specified literally or by using CURRENT_USER. You cannot set the definer to some other account.

  • If you have the SUPER privilege, you can specify any syntactically legal account name. If the account does not actually exist, a warning is generated.

  • Although it is possible to create a view with a nonexistent DEFINER account, an error occurs when the view is referenced if the SQL SECURITY value is DEFINER but the definer account does not exist.

For more information about view security, see Section 19.6, “Access Control for Stored Programs and Views”.

Within a view definition, CURRENT_USER returns the view's DEFINER value by default. For views defined with the SQL SECURITY INVOKER characteristic, CURRENT_USER returns the account for the view's invoker. For information about user auditing within views, see Section 5.5.10, “Auditing MySQL Account Activity”.

Within a stored routine that is defined with the SQL SECURITY DEFINER characteristic, CURRENT_USER returns the routine's DEFINER value. This also affects a view defined within such a routine, if the view definition contains a DEFINER value of CURRENT_USER.

View privileges are checked like this:

  • At view definition time, the view creator must have the privileges needed to use the top-level objects accessed by the view. For example, if the view definition refers to table columns, the creator must have some privilege for each column in the select list of the definition, and the SELECT privilege for each column used elsewhere in the definition. If the definition refers to a stored function, only the privileges needed to invoke the function can be checked. The privileges required at function invocation time can be checked only as it executes: For different invocations, different execution paths within the function might be taken.

  • The user who references a view must have appropriate privileges to access it (SELECT to select from it, INSERT to insert into it, and so forth.)

  • When a view has been referenced, privileges for objects accessed by the view are checked against the privileges held by the view DEFINER account or invoker, depending on whether the SQL SECURITY characteristic is DEFINER or INVOKER, respectively.

  • If reference to a view causes execution of a stored function, privilege checking for statements executed within the function depend on whether the function SQL SECURITY characteristic is DEFINER or INVOKER. If the security characteristic is DEFINER, the function runs with the privileges of the DEFINER account. If the characteristic is INVOKER, the function runs with the privileges determined by the view's SQL SECURITY characteristic.

Example: A view might depend on a stored function, and that function might invoke other stored routines. For example, the following view invokes a stored function f():

CREATE VIEW v AS SELECT * FROM t WHERE t.id = f(t.name);

Suppose that f() contains a statement such as this:

IF name IS NULL then
  CALL p1();
ELSE
  CALL p2();
END IF;

The privileges required for executing statements within f() need to be checked when f() executes. This might mean that privileges are needed for p1() or p2(), depending on the execution path within f(). Those privileges must be checked at runtime, and the user who must possess the privileges is determined by the SQL SECURITY values of the view v and the function f().

The DEFINER and SQL SECURITY clauses for views are extensions to standard SQL. In standard SQL, views are handled using the rules for SQL SECURITY DEFINER. The standard says that the definer of the view, which is the same as the owner of the view's schema, gets applicable privileges on the view (for example, SELECT) and may grant them. MySQL has no concept of a schema “owner”, so MySQL adds a clause to identify the definer. The DEFINER clause is an extension where the intent is to have what the standard has; that is, a permanent record of who defined the view. This is why the default DEFINER value is the account of the view creator.

The optional ALGORITHM clause is a MySQL extension to standard SQL. It affects how MySQL processes the view. ALGORITHM takes three values: MERGE, TEMPTABLE, or UNDEFINED. The default algorithm is UNDEFINED if no ALGORITHM clause is present. For more information, see Section 19.5.2, “View Processing Algorithms”.

Some views are updatable. That is, you can use them in statements such as UPDATE, DELETE, or INSERT to update the contents of the underlying table. For a view to be updatable, there must be a one-to-one relationship between the rows in the view and the rows in the underlying table. There are also certain other constructs that make a view nonupdatable.

The WITH CHECK OPTION clause can be given for an updatable view to prevent inserts or updates to rows except those for which the WHERE clause in the select_statement is true.

In a WITH CHECK OPTION clause for an updatable view, the LOCAL and CASCADED keywords determine the scope of check testing when the view is defined in terms of another view. The LOCAL keyword restricts the CHECK OPTION only to the view being defined. CASCADED causes the checks for underlying views to be evaluated as well. When neither keyword is given, the default is CASCADED.

For more information about updatable views and the WITH CHECK OPTION clause, see Section 19.5.3, “Updatable and Insertable Views”.

12.1.17. DROP DATABASE Syntax

DROP {DATABASE | SCHEMA} [IF EXISTS] db_name

DROP DATABASE drops all tables in the database and deletes the database. Be very careful with this statement! To use DROP DATABASE, you need the DROP privilege on the database. DROP SCHEMA is a synonym for DROP DATABASE.

Important

When a database is dropped, user privileges on the database are not automatically dropped. See Section 12.4.1.3, “GRANT Syntax”.

IF EXISTS is used to prevent an error from occurring if the database does not exist.

If the default database is dropped, the default database is unset (the DATABASE() function returns NULL).

If you use DROP DATABASE on a symbolically linked database, both the link and the original database are deleted.

DROP DATABASE returns the number of tables that were removed. This corresponds to the number of .frm files removed.

The DROP DATABASE statement removes from the given database directory those files and directories that MySQL itself may create during normal operation:

  • All files with the following extensions.

    .BAK.DAT.HSH.MRG
    .MYD.MYI.TRG.TRN
    .db.frm.ibd.ndb
    .par   
  • The db.opt file, if it exists.

If other files or directories remain in the database directory after MySQL removes those just listed, the database directory cannot be removed. In this case, you must remove any remaining files or directories manually and issue the DROP DATABASE statement again.

You can also drop databases with mysqladmin. See Section 4.5.2, “mysqladmin — Client for Administering a MySQL Server”.

12.1.18. DROP EVENT Syntax

DROP EVENT [IF EXISTS] event_name

This statement drops the event named event_name. The event immediately ceases being active, and is deleted completely from the server.

If the event does not exist, the error ERROR 1517 (HY000): Unknown event 'event_name' results. You can override this and cause the statement to generate a warning for nonexistent events instead using IF EXISTS.

This statement requires the EVENT privilege for the schema to which the event to be dropped belongs.

12.1.19. DROP FUNCTION Syntax

The DROP FUNCTION statement is used to drop stored functions and user-defined functions (UDFs):

12.1.20. DROP INDEX Syntax

DROP INDEX index_name ON tbl_name

DROP INDEX drops the index named index_name from the table tbl_name. This statement is mapped to an ALTER TABLE statement to drop the index. See Section 12.1.6, “ALTER TABLE Syntax”.

12.1.21. DROP PROCEDURE and DROP FUNCTION Syntax

DROP {PROCEDURE | FUNCTION} [IF EXISTS] sp_name

This statement is used to drop a stored procedure or function. That is, the specified routine is removed from the server. You must have the ALTER ROUTINE privilege for the routine. (If the automatic_sp_privileges system variable is enabled, that privilege and EXECUTE are granted automatically to the routine creator when the routine is created and dropped from the creator when the routine is dropped. See Section 19.2.2, “Stored Routines and MySQL Privileges”.)

The IF EXISTS clause is a MySQL extension. It prevents an error from occurring if the procedure or function does not exist. A warning is produced that can be viewed with SHOW WARNINGS.

DROP FUNCTION is also used to drop user-defined functions (see Section 12.4.3.3, “DROP FUNCTION Syntax”).

12.1.22. DROP SERVER Syntax

DROP SERVER [ IF EXISTS ] server_name

Drops the server definition for the server named server_name. The corresponding row within the mysql.servers table will be deleted. This statement requires the SUPER privilege.

Dropping a server for a table does not affect any FEDERATED tables that used this connection information when they were created. See Section 12.1.13, “CREATE SERVER Syntax”.

DROP SERVER does not cause an automatic commit.

12.1.23. DROP TABLE Syntax

DROP [TEMPORARY] TABLE [IF EXISTS]
    tbl_name [, tbl_name] ...
    [RESTRICT | CASCADE]

DROP TABLE removes one or more tables. You must have the DROP privilege for each table. All table data and the table definition are removed, so be careful with this statement! If any of the tables named in the argument list do not exist, MySQL returns an error indicating by name which nonexisting tables it was unable to drop, but it also drops all of the tables in the list that do exist.

Important

When a table is dropped, user privileges on the table are not automatically dropped. See Section 12.4.1.3, “GRANT Syntax”.

Note that for a partitioned table, DROP TABLE permanently removes the table definition, all of its partitions, and all of the data which was stored in those partitions. It also removes the partitioning definition (.par) file associated with the dropped table.

Use IF EXISTS to prevent an error from occurring for tables that do not exist. A NOTE is generated for each nonexistent table when using IF EXISTS. See Section 12.4.5.41, “SHOW WARNINGS Syntax”.

RESTRICT and CASCADE are permitted to make porting easier. In MySQL 5.5, they do nothing.

Note

DROP TABLE automatically commits the current active transaction, unless you use the TEMPORARY keyword.

The TEMPORARY keyword has the following effects:

  • The statement drops only TEMPORARY tables.

  • The statement does not end an ongoing transaction.

  • No access rights are checked. (A TEMPORARY table is visible only to the session that created it, so no check is necessary.)

Using TEMPORARY is a good way to ensure that you do not accidentally drop a non-TEMPORARY table.

12.1.24. DROP TRIGGER Syntax

DROP TRIGGER [IF EXISTS] [schema_name.]trigger_name

This statement drops a trigger. The schema (database) name is optional. If the schema is omitted, the trigger is dropped from the default schema. DROP TRIGGER was added in MySQL 5.0.2. Its use requires the TRIGGER privilege for the table associated with the trigger.

Use IF EXISTS to prevent an error from occurring for a trigger that does not exist. A NOTE is generated for a nonexistent trigger when using IF EXISTS. See Section 12.4.5.41, “SHOW WARNINGS Syntax”.

Triggers for a table are also dropped if you drop the table.

Note

When upgrading from a version of MySQL older than MySQL 5.0.10 to 5.0.10 or newer—including all MySQL 5.5 releases—you must drop all triggers before upgrading and re-create them afterward, or else DROP TRIGGER does not work after the upgrade.

12.1.25. DROP VIEW Syntax

DROP VIEW [IF EXISTS]
    view_name [, view_name] ...
    [RESTRICT | CASCADE]

DROP VIEW removes one or more views. You must have the DROP privilege for each view. If any of the views named in the argument list do not exist, MySQL returns an error indicating by name which nonexisting views it was unable to drop, but it also drops all of the views in the list that do exist.

The IF EXISTS clause prevents an error from occurring for views that don't exist. When this clause is given, a NOTE is generated for each nonexistent view. See Section 12.4.5.41, “SHOW WARNINGS Syntax”.

RESTRICT and CASCADE, if given, are parsed and ignored.

12.1.26. RENAME TABLE Syntax

RENAME TABLE tbl_name TO new_tbl_name
    [, tbl_name2 TO new_tbl_name2] ...

This statement renames one or more tables.

The rename operation is done atomically, which means that no other session can access any of the tables while the rename is running. For example, if you have an existing table old_table, you can create another table new_table that has the same structure but is empty, and then replace the existing table with the empty one as follows (assuming that backup_table does not already exist):

CREATE TABLE new_table (...);
RENAME TABLE old_table TO backup_table, new_table TO old_table;

If the statement renames more than one table, renaming operations are done from left to right. If you want to swap two table names, you can do so like this (assuming that tmp_table does not already exist):

RENAME TABLE old_table TO tmp_table,
             new_table TO old_table,
             tmp_table TO new_table;

As long as two databases are on the same file system, you can use RENAME TABLE to move a table from one database to another:

RENAME TABLE current_db.tbl_name TO other_db.tbl_name;

If there are any triggers associated with a table which is moved to a different database using RENAME TABLE, then the statement fails with the error Trigger in wrong schema.

RENAME TABLE also works for views, as long as you do not try to rename a view into a different database.

Any privileges granted specifically for the renamed table or view are not migrated to the new name. They must be changed manually.

When you execute RENAME, you cannot have any locked tables or active transactions. You must also have the ALTER and DROP privileges on the original table, and the CREATE and INSERT privileges on the new table.

If MySQL encounters any errors in a multiple-table rename, it does a reverse rename for all renamed tables to return everything to its original state.

You cannot use RENAME to rename a TEMPORARY table. However, you can use ALTER TABLE instead:

mysql> ALTER TABLE orig_name RENAME new_name;

12.1.27. TRUNCATE TABLE Syntax

TRUNCATE [TABLE] tbl_name

TRUNCATE TABLE empties a table completely. It requires the DROP privilege.

Logically, TRUNCATE TABLE is similar to a DELETE statement that deletes all rows, or a sequence of DROP TABLE and CREATE TABLE statements. To achieve high performance, it bypasses the DML method of deleting data. Thus, it cannot be rolled back, it does not cause ON DELETE triggers to fire, and it cannot be performed for InnoDB tables with parent-child foreign key relationships.

Although TRUNCATE TABLE is similar to DELETE, it is classified as a DDL statement rather than a DML statement. It differs from DELETE in the following ways in MySQL 5.5:

  • Truncate operations drop and re-create the table, which is much faster than deleting rows one by one, particularly for large tables.

  • Truncate operations cause an implicit commit, and so cannot be rolled back.

  • Truncation operations cannot be performed if the session holds an active table lock.

  • TRUNCATE TABLE fails for an InnoDB table if there are any FOREIGN KEY constraints from other tables that reference the table. Foreign key constraints between columns of the same table are allowed.

  • Truncation operations do not return a meaningful value for the number of deleted rows. The usual result is “0 rows affected,” which should be interpreted as “no information.

  • As long as the table format file tbl_name.frm is valid, the table can be re-created as an empty table with TRUNCATE TABLE, even if the data or index files have become corrupted.

  • Any AUTO_INCREMENT value is reset to its start value. This is true even for MyISAM and InnoDB, which normally do not reuse sequence values.

  • When used with partitioned tables, TRUNCATE TABLE preserves the partitioning; that is, the data and index files are dropped and re-created, while the partition definitions (.par) file is unaffected.

  • The TRUNCATE TABLE statement does not invoke ON DELETE triggers.

TRUNCATE TABLE for a table closes all handlers for the table that were opened with HANDLER OPEN.

TRUNCATE TABLE is treated for purposes of binary logging and replication as DROP TABLE followed by CREATE TABLE—that is, as DDL rather than DML. This is due to the fact that, when using InnoDB and other transactional storage engines where the transaction isolation level does not permit statement-based logging (READ COMMITTED or READ UNCOMMITTED), the statement was not logged and replicated when using STATEMENT or MIXED logging mode. (Bug#36763) However, it is still applied on replication slaves using InnoDB in the manner described previously.

TRUNCATE TABLE can be used with Performance Schema summary tables, but the effect is to reset the summary columns to 0 or NULL, not to remove rows. See Section 21.7.4, “Performance Schema Summary Tables”.

12.2. Data Manipulation Statements

12.2.1. CALL Syntax

CALL sp_name([parameter[,...]])
CALL sp_name[()]

The CALL statement invokes a stored procedure that was defined previously with CREATE PROCEDURE.

Stored procedures that take no arguments can be invoked without parentheses. That is, CALL p() and CALL p are equivalent.

CALL can pass back values to its caller using parameters that are declared as OUT or INOUT parameters. When the procedure returns, a client program can also obtain the number of rows affected for the final statement executed within the routine: At the SQL level, call the ROW_COUNT() function; from the C API, call the mysql_affected_rows() function.

To get back a value from a procedure using an OUT or INOUT parameter, pass the parameter by means of a user variable, and then check the value of the variable after the procedure returns. (If you are calling the procedure from within another stored procedure or function, you can also pass a routine parameter or local routine variable as an IN or INOUT parameter.) For an INOUT parameter, initialize its value before passing it to the procedure. The following procedure has an OUT parameter that the procedure sets to the current server version, and an INOUT value that the procedure increments by one from its current value:

CREATE PROCEDURE p (OUT ver_param VARCHAR(25), INOUT incr_param INT)
BEGIN
  # Set value of OUT parameter
  SELECT VERSION() INTO ver_param;
  # Increment value of INOUT parameter
  SET incr_param = incr_param + 1;
END;

Before calling the procedure, initialize the variable to be passed as the INOUT parameter. After calling the procedure, the values of the two variables will have been set or modified:

mysql> SET @increment = 10;
mysql> CALL p(@version, @increment);
mysql> SELECT @version, @increment;
+--------------+------------+
| @version     | @increment |
+--------------+------------+
| 5.5.3-m3-log |         11 |
+--------------+------------+

In prepared CALL statements used with PREPARE and EXECUTE, placeholders can be used for IN parameters. For OUT and INOUT parameters, placeholder support is available as of MySQL 5.5.3. These types of parameters can be used as follows:

mysql> SET @increment = 10;
mysql> PREPARE s FROM 'CALL p(?, ?)';
mysql> EXECUTE s USING @version, @increment;
mysql> SELECT @version, @increment;
+--------------+------------+
| @version     | @increment |
+--------------+------------+
| 5.5.3-m3-log |         11 |
+--------------+------------+

Before MySQL 5.5.3, placeholder support is not available for OUT or INOUT parameters. To work around this limitation for OUT and INOUT parameters, forego the use of placeholders; instead, refer to user variables in the CALL statement itself and do not specify them in the EXECUTE statement:

mysql> SET @increment = 10;
mysql> PREPARE s FROM 'CALL p(@version, @increment)';
mysql> EXECUTE s;
mysql> SELECT @version, @increment;
+--------------+------------+
| @version     | @increment |
+--------------+------------+
| 5.5.0-m2-log |         11 |
+--------------+------------+

To write C programs that use the CALL SQL statement to execute stored procedures that produce result sets, the CLIENT_MULTI_RESULTS flag must be enabled. This is because each CALL returns a result to indicate the call status, in addition to any result sets that might be returned by statements executed within the procedure. CLIENT_MULTI_RESULTS must also be enabled if CALL is used to execute any stored procedure that contains prepared statements. It cannot be determined when such a procedure is loaded whether those statements will produce result sets, so it is necessary to assume that they will.

CLIENT_MULTI_RESULTS can be enabled when you call mysql_real_connect(), either explicitly by passing the CLIENT_MULTI_RESULTS flag itself, or implicitly by passing CLIENT_MULTI_STATEMENTS (which also enables CLIENT_MULTI_RESULTS). As of MySQL 5.5.3, CLIENT_MULTI_RESULTS is enabled by default.

To process the result of a CALL statement executed using mysql_query() or mysql_real_query(), use a loop that calls mysql_next_result() to determine whether there are more results. For an example, see Section 22.9.13, “C API Support for Multiple Statement Execution”.

For programs written in a language that provides a MySQL interface, there is no native method prior to MySQL 5.5.3 for directly retrieving the results of OUT or INOUT parameters from CALL statements. To get the parameter values, pass user-defined variables to the procedure in the CALL statement and then execute a SELECT statement to produce a result set containing the variable values. To handle an INOUT parameter, execute a statement prior to the CALL that sets the corresponding user variable to the value to be passed to the procedure.

The following example illustrates the technique (without error checking) for the stored procedure p described earlier that has an OUT parameter and an INOUT parameter:

mysql_query(mysql, "SET @increment = 10");
mysql_query(mysql, "CALL p(@version, @increment)");
mysql_query(mysql, "SELECT @version, @increment");
result = mysql_store_result(mysql);
row = mysql_fetch_row(result);
mysql_free_result(result);

After the preceding code executes, row[0] and row[1] contain the values of @version and @increment, respectively.

As of MySQL 5.5.3, C programs can use the prepared-statement interface to execute CALL statements and access OUT and INOUT parameters. This is done by processing the result of a CALL statement using a loop that calls mysql_stmt_next_result() to determine whether there are more results. For an example, see Section 22.9.16, “C API Support for Prepared CALL Statements”. Languages that provide a MySQL interface can use prepared CALL statements to directly retrieve OUT and INOUT procedure parameters.

12.2.2. DELETE Syntax

Single-table syntax:

DELETE [LOW_PRIORITY] [QUICK] [IGNORE] FROM tbl_name
    [WHERE where_condition]
    [ORDER BY ...]
    [LIMIT row_count]

Multiple-table syntax:

DELETE [LOW_PRIORITY] [QUICK] [IGNORE]
    tbl_name[.*] [, tbl_name[.*]] ...
    FROM table_references
    [WHERE where_condition]

Or:

DELETE [LOW_PRIORITY] [QUICK] [IGNORE]
    FROM tbl_name[.*] [, tbl_name[.*]] ...
    USING table_references
    [WHERE where_condition]

For the single-table syntax, the DELETE statement deletes rows from tbl_name and returns a count of the number of deleted rows. This count can be obtained by calling the ROW_COUNT() function (see Section 11.14, “Information Functions”). The WHERE clause, if given, specifies the conditions that identify which rows to delete. With no WHERE clause, all rows are deleted. If the ORDER BY clause is specified, the rows are deleted in the order that is specified. The LIMIT clause places a limit on the number of rows that can be deleted.

For the multiple-table syntax, DELETE deletes from each tbl_name the rows that satisfy the conditions. In this case, ORDER BY and LIMIT cannot be used.

where_condition is an expression that evaluates to true for each row to be deleted. It is specified as described in Section 12.2.9, “SELECT Syntax”.

Currently, you cannot delete from a table and select from the same table in a subquery.

You need the DELETE privilege on a table to delete rows from it. You need only the SELECT privilege for any columns that are only read, such as those named in the WHERE clause.

As stated, a DELETE statement with no WHERE clause deletes all rows. A faster way to do this, when you do not need to know the number of deleted rows, is to use TRUNCATE TABLE. However, within a transaction or if you have a lock on the table, TRUNCATE TABLE cannot be used whereas DELETE can. See Section 12.1.27, “TRUNCATE TABLE Syntax”, and Section 12.3.5, “LOCK TABLES and UNLOCK TABLES Syntax”.

If you delete the row containing the maximum value for an AUTO_INCREMENT column, the value is not reused for a MyISAM or InnoDB table. If you delete all rows in the table with DELETE FROM tbl_name (without a WHERE clause) in autocommit mode, the sequence starts over for all storage engines except InnoDB and MyISAM. There are some exceptions to this behavior for InnoDB tables, as discussed in Section 13.6.5.3, “AUTO_INCREMENT Handling in InnoDB.

For MyISAM tables, you can specify an AUTO_INCREMENT secondary column in a multiple-column key. In this case, reuse of values deleted from the top of the sequence occurs even for MyISAM tables. See Section 3.6.9, “Using AUTO_INCREMENT.

The DELETE statement supports the following modifiers:

  • If you specify LOW_PRIORITY, the server delays execution of the DELETE until no other clients are reading from the table. This affects only storage engines that use only table-level locking (such as MyISAM, MEMORY, and MERGE).

  • For MyISAM tables, if you use the QUICK keyword, the storage engine does not merge index leaves during delete, which may speed up some kinds of delete operations.

  • The IGNORE keyword causes MySQL to ignore all errors during the process of deleting rows. (Errors encountered during the parsing stage are processed in the usual manner.) Errors that are ignored due to the use of IGNORE are returned as warnings.

The speed of delete operations may also be affected by factors discussed in Section 7.2.2.3, “Speed of DELETE Statements”.

In MyISAM tables, deleted rows are maintained in a linked list and subsequent INSERT operations reuse old row positions. To reclaim unused space and reduce file sizes, use the OPTIMIZE TABLE statement or the myisamchk utility to reorganize tables. OPTIMIZE TABLE is easier to use, but myisamchk is faster. See Section 12.4.2.4, “OPTIMIZE TABLE Syntax”, and Section 4.6.3, “myisamchk — MyISAM Table-Maintenance Utility”.

The QUICK modifier affects whether index leaves are merged for delete operations. DELETE QUICK is most useful for applications where index values for deleted rows are replaced by similar index values from rows inserted later. In this case, the holes left by deleted values are reused.

DELETE QUICK is not useful when deleted values lead to underfilled index blocks spanning a range of index values for which new inserts occur again. In this case, use of QUICK can lead to wasted space in the index that remains unreclaimed. Here is an example of such a scenario:

  1. Create a table that contains an indexed AUTO_INCREMENT column.

  2. Insert many rows into the table. Each insert results in an index value that is added to the high end of the index.

  3. Delete a block of rows at the low end of the column range using DELETE QUICK.

In this scenario, the index blocks associated with the deleted index values become underfilled but are not merged with other index blocks due to the use of QUICK. They remain underfilled when new inserts occur, because new rows do not have index values in the deleted range. Furthermore, they remain underfilled even if you later use DELETE without QUICK, unless some of the deleted index values happen to lie in index blocks within or adjacent to the underfilled blocks. To reclaim unused index space under these circumstances, use OPTIMIZE TABLE.

If you are going to delete many rows from a table, it might be faster to use DELETE QUICK followed by OPTIMIZE TABLE. This rebuilds the index rather than performing many index block merge operations.

The MySQL-specific LIMIT row_count option to DELETE tells the server the maximum number of rows to be deleted before control is returned to the client. This can be used to ensure that a given DELETE statement does not take too much time. You can simply repeat the DELETE statement until the number of affected rows is less than the LIMIT value.

If the DELETE statement includes an ORDER BY clause, rows are deleted in the order specified by the clause. This is useful primarily in conjunction with LIMIT. For example, the following statement finds rows matching the WHERE clause, sorts them by timestamp_column, and deletes the first (oldest) one:

DELETE FROM somelog WHERE user = 'jcole'
ORDER BY timestamp_column LIMIT 1;

ORDER BY may also be useful in some cases to delete rows in an order required to avoid referential integrity violations.

If you are deleting many rows from a large table, you may exceed the lock table size for an InnoDB table. To avoid this problem, or simply to minimize the time that the table remains locked, the following strategy (which does not use DELETE at all) might be helpful:

  1. Select the rows not to be deleted into an empty table that has the same structure as the original table:

    INSERT INTO t_copy SELECT * FROM t WHERE ... ;
    
  2. Use RENAME TABLE to atomically move the original table out of the way and rename the copy to the original name:

    RENAME TABLE t TO t_old, t_copy TO t;
    
  3. Drop the original table:

    DROP TABLE t_old;
    

No other sessions can access the tables involved while RENAME TABLE executes, so the rename operation is not subject to concurrency problems. See Section 12.1.26, “RENAME TABLE Syntax”.

You can specify multiple tables in a DELETE statement to delete rows from one or more tables depending on the particular condition in the WHERE clause. However, you cannot use ORDER BY or LIMIT in a multiple-table DELETE. The table_references clause lists the tables involved in the join. Its syntax is described in Section 12.2.9.1, “JOIN Syntax”.

For the first multiple-table syntax, only matching rows from the tables listed before the FROM clause are deleted. For the second multiple-table syntax, only matching rows from the tables listed in the FROM clause (before the USING clause) are deleted. The effect is that you can delete rows from many tables at the same time and have additional tables that are used only for searching:

DELETE t1, t2 FROM t1 INNER JOIN t2 INNER JOIN t3
WHERE t1.id=t2.id AND t2.id=t3.id;

Or:

DELETE FROM t1, t2 USING t1 INNER JOIN t2 INNER JOIN t3
WHERE t1.id=t2.id AND t2.id=t3.id;

These statements use all three tables when searching for rows to delete, but delete matching rows only from tables t1 and t2.

The preceding examples use INNER JOIN, but multiple-table DELETE statements can use other types of join permitted in SELECT statements, such as LEFT JOIN. For example, to delete rows that exist in t1 that have no match in t2, use a LEFT JOIN:

DELETE t1 FROM t1 LEFT JOIN t2 ON t1.id=t2.id WHERE t2.id IS NULL;

The syntax permits .* after each tbl_name for compatibility with Access.

If you use a multiple-table DELETE statement involving InnoDB tables for which there are foreign key constraints, the MySQL optimizer might process tables in an order that differs from that of their parent/child relationship. In this case, the statement fails and rolls back. Instead, you should delete from a single table and rely on the ON DELETE capabilities that InnoDB provides to cause the other tables to be modified accordingly.

Note

If you declare an alias for a table, you must use the alias when referring to the table:

DELETE t1 FROM test AS t1, test2 WHERE ...

Table aliases in a multiple-table DELETE should be declared only in the table_references part of the statement.

Correct:

DELETE a1, a2 FROM t1 AS a1 INNER JOIN t2 AS a2
WHERE a1.id=a2.id;

DELETE FROM a1, a2 USING t1 AS a1 INNER JOIN t2 AS a2
WHERE a1.id=a2.id;

Incorrect:

DELETE t1 AS a1, t2 AS a2 FROM t1 INNER JOIN t2
WHERE a1.id=a2.id;

DELETE FROM t1 AS a1, t2 AS a2 USING t1 INNER JOIN t2
WHERE a1.id=a2.id;

Declaration of aliases other than in the table_references part of the statement should be avoided because that can lead to ambiguous statements that have unexpected results such as deleting rows from the wrong table. This is such a statement:

DELETE t1 AS a2 FROM t1 AS a1 INNER JOIN t2 AS a2;

Before MySQL 5.5.3, alias declarations outside the table_references part of the statement are disallowed for the USING variant of multiple-table DELETE syntax. As of MySQL 5.5.3, alias declarations outside table_references are disallowed for all multiple-table DELETE statements.

Before MySQL 5.5.3, for alias references in the list of tables from which to delete rows in a multiple-table delete, the default database is used unless one is specified explicitly. For example, if the default database is db1, the following statement does not work because the unqualified alias reference a2 is interpreted as having a database of db1:

DELETE a1, a2 FROM db1.t1 AS a1 INNER JOIN db2.t2 AS a2
WHERE a1.id=a2.id;

To correctly match an alias that refers to a table outside the default database, you must explicitly qualify the reference with the name of the proper database:

DELETE a1, db2.a2 FROM db1.t1 AS a1 INNER JOIN db2.t2 AS a2
WHERE a1.id=a2.id;

As of MySQL 5.5.3, alias resolution does not require qualification and alias references should not be qualified with the database name. Qualified names are interpreted as referring to tables, not aliases.

12.2.3. DO Syntax

DO expr [, expr] ...

DO executes the expressions but does not return any results. In most respects, DO is shorthand for SELECT expr, ..., but has the advantage that it is slightly faster when you do not care about the result.

DO is useful primarily with functions that have side effects, such as RELEASE_LOCK().

12.2.4. HANDLER Syntax

HANDLER tbl_name OPEN [ [AS] alias]

HANDLER tbl_name READ index_name { = | <= | >= | < | > } (value1,value2,...)
    [ WHERE where_condition ] [LIMIT ... ]
HANDLER tbl_name READ index_name { FIRST | NEXT | PREV | LAST }
    [ WHERE where_condition ] [LIMIT ... ]
HANDLER tbl_name READ { FIRST | NEXT }
    [ WHERE where_condition ] [LIMIT ... ]

HANDLER tbl_name CLOSE

The HANDLER statement provides direct access to table storage engine interfaces. It is available for MyISAM and InnoDB tables.

The HANDLER ... OPEN statement opens a table, making it accessible using subsequent HANDLER ... READ statements. This table object is not shared by other sessions and is not closed until the session calls HANDLER ... CLOSE or the session terminates. If you open the table using an alias, further references to the open table with other HANDLER statements must use the alias rather than the table name.

The first HANDLER ... READ syntax fetches a row where the index specified satisfies the given values and the WHERE condition is met. If you have a multiple-column index, specify the index column values as a comma-separated list. Either specify values for all the columns in the index, or specify values for a leftmost prefix of the index columns. Suppose that an index my_idx includes three columns named col_a, col_b, and col_c, in that order. The HANDLER statement can specify values for all three columns in the index, or for the columns in a leftmost prefix. For example:

HANDLER ... READ my_idx = (col_a_val,col_b_val,col_c_val) ...
HANDLER ... READ my_idx = (col_a_val,col_b_val) ...
HANDLER ... READ my_idx = (col_a_val) ...

To employ the HANDLER interface to refer to a table's PRIMARY KEY, use the quoted identifier `PRIMARY`:

HANDLER tbl_name READ `PRIMARY` ...

The second HANDLER ... READ syntax fetches a row from the table in index order that matches the WHERE condition.

The third HANDLER ... READ syntax fetches a row from the table in natural row order that matches the WHERE condition. It is faster than HANDLER tbl_name READ index_name when a full table scan is desired. Natural row order is the order in which rows are stored in a MyISAM table data file. This statement works for InnoDB tables as well, but there is no such concept because there is no separate data file.

Without a LIMIT clause, all forms of HANDLER ... READ fetch a single row if one is available. To return a specific number of rows, include a LIMIT clause. It has the same syntax as for the SELECT statement. See Section 12.2.9, “SELECT Syntax”.

HANDLER ... CLOSE closes a table that was opened with HANDLER ... OPEN.

There are several reasons to use the HANDLER interface instead of normal SELECT statements:

  • HANDLER is faster than SELECT:

    • A designated storage engine handler object is allocated for the HANDLER ... OPEN. The object is reused for subsequent HANDLER statements for that table; it need not be reinitialized for each one.

    • There is less parsing involved.

    • There is no optimizer or query-checking overhead.

    • The table does not have to be locked between two handler requests.

    • The handler interface does not have to provide a consistent look of the data (for example, dirty reads are permitted), so the storage engine can use optimizations that SELECT does not normally permit.

  • For applications that use a low-level ISAM-like interface, HANDLER makes it much easier to port them to MySQL.

  • HANDLER enables you to traverse a database in a manner that is difficult (or even impossible) to accomplish with SELECT. The HANDLER interface is a more natural way to look at data when working with applications that provide an interactive user interface to the database.

HANDLER is a somewhat low-level statement. For example, it does not provide consistency. That is, HANDLER ... OPEN does not take a snapshot of the table, and does not lock the table. This means that after a HANDLER ... OPEN statement is issued, table data can be modified (by the current session or other sessions) and these modifications might be only partially visible to HANDLER ... NEXT or HANDLER ... PREV scans.

An open handler can be closed and marked for reopen, in which case the handler loses its position in the table. This occurs when both of the following circumstances are true:

  • Any session executes FLUSH TABLES or DDL statements on the handler's table.

  • The session in which the handler is open executes non-HANDLER statements that use tables.

TRUNCATE TABLE for a table closes all handlers for the table that were opened with HANDLER OPEN.

If a table is flushed with FLUSH TABLES tbl_name WITH READ LOCK was opened with HANDLER, the handler is implicitly flushed and loses its position.

12.2.5. INSERT Syntax

INSERT [LOW_PRIORITY | DELAYED | HIGH_PRIORITY] [IGNORE]
    [INTO] tbl_name [(col_name,...)]
    {VALUES | VALUE} ({expr | DEFAULT},...),(...),...
    [ ON DUPLICATE KEY UPDATE
      col_name=expr
        [, col_name=expr] ... ]

Or:

INSERT [LOW_PRIORITY | DELAYED | HIGH_PRIORITY] [IGNORE]
    [INTO] tbl_name
    SET col_name={expr | DEFAULT}, ...
    [ ON DUPLICATE KEY UPDATE
      col_name=expr
        [, col_name=expr] ... ]

Or:

INSERT [LOW_PRIORITY | HIGH_PRIORITY] [IGNORE]
    [INTO] tbl_name [(col_name,...)]
    SELECT ...
    [ ON DUPLICATE KEY UPDATE
      col_name=expr
        [, col_name=expr] ... ]

INSERT inserts new rows into an existing table. The INSERT ... VALUES and INSERT ... SET forms of the statement insert rows based on explicitly specified values. The INSERT ... SELECT form inserts rows selected from another table or tables. INSERT ... SELECT is discussed further in Section 12.2.5.1, “INSERT ... SELECT Syntax”.

You can use REPLACE instead of INSERT to overwrite old rows. REPLACE is the counterpart to INSERT IGNORE in the treatment of new rows that contain unique key values that duplicate old rows: The new rows are used to replace the old rows rather than being discarded. See Section 12.2.8, “REPLACE Syntax”.

tbl_name is the table into which rows should be inserted. The columns for which the statement provides values can be specified as follows:

  • You can provide a comma-separated list of column names following the table name. In this case, a value for each named column must be provided by the VALUES list or the SELECT statement.

  • If you do not specify a list of column names for INSERT ... VALUES or INSERT ... SELECT, values for every column in the table must be provided by the VALUES list or the SELECT statement. If you do not know the order of the columns in the table, use DESCRIBE tbl_name to find out.

  • The SET clause indicates the column names explicitly.

Column values can be given in several ways:

  • If you are not running in strict SQL mode, any column not explicitly given a value is set to its default (explicit or implicit) value. For example, if you specify a column list that does not name all the columns in the table, unnamed columns are set to their default values. Default value assignment is described in Section 10.1.4, “Data Type Default Values”. See also Section 1.8.6.2, “Constraints on Invalid Data”.

    If you want an INSERT statement to generate an error unless you explicitly specify values for all columns that do not have a default value, you should use strict mode. See Section 5.1.7, “Server SQL Modes”.

  • Use the keyword DEFAULT to set a column explicitly to its default value. This makes it easier to write INSERT statements that assign values to all but a few columns, because it enables you to avoid writing an incomplete VALUES list that does not include a value for each column in the table. Otherwise, you would have to write out the list of column names corresponding to each value in the VALUES list.

    You can also use DEFAULT(col_name) as a more general form that can be used in expressions to produce a given column's default value.

  • If both the column list and the VALUES list are empty, INSERT creates a row with each column set to its default value:

    INSERT INTO tbl_name () VALUES();
    

    In strict mode, an error occurs if any column doesn't have a default value. Otherwise, MySQL uses the implicit default value for any column that does not have an explicitly defined default.

  • You can specify an expression expr to provide a column value. This might involve type conversion if the type of the expression does not match the type of the column, and conversion of a given value can result in different inserted values depending on the data type. For example, inserting the string '1999.0e-2' into an INT, FLOAT, DECIMAL(10,6), or YEAR column results in the values 1999, 19.9921, 19.992100, and 1999 being inserted, respectively. The reason the value stored in the INT and YEAR columns is 1999 is that the string-to-integer conversion looks only at as much of the initial part of the string as may be considered a valid integer or year. For the floating-point and fixed-point columns, the string-to-floating-point conversion considers the entire string a valid floating-point value.

    An expression expr can refer to any column that was set earlier in a value list. For example, you can do this because the value for col2 refers to col1, which has previously been assigned:

    INSERT INTO tbl_name (col1,col2) VALUES(15,col1*2);
    

    But the following is not legal, because the value for col1 refers to col2, which is assigned after col1:

    INSERT INTO tbl_name (col1,col2) VALUES(col2*2,15);
    

    One exception involves columns that contain AUTO_INCREMENT values. Because the AUTO_INCREMENT value is generated after other value assignments, any reference to an AUTO_INCREMENT column in the assignment returns a 0.

INSERT statements that use VALUES syntax can insert multiple rows. To do this, include multiple lists of column values, each enclosed within parentheses and separated by commas. Example:

INSERT INTO tbl_name (a,b,c) VALUES(1,2,3),(4,5,6),(7,8,9);

The values list for each row must be enclosed within parentheses. The following statement is illegal because the number of values in the list does not match the number of column names:

INSERT INTO tbl_name (a,b,c) VALUES(1,2,3,4,5,6,7,8,9);

VALUE is a synonym for VALUES in this context. Neither implies anything about the number of values lists, and either may be used whether there is a single values list or multiple lists.

The affected-rows value for an INSERT can be obtained using the ROW_COUNT() function (see Section 11.14, “Information Functions”), or the mysql_affected_rows() C API function (see Section 22.9.3.1, “mysql_affected_rows()).

If you use an INSERT ... VALUES statement with multiple value lists or INSERT ... SELECT, the statement returns an information string in this format:

Records: 100 Duplicates: 0 Warnings: 0

Records indicates the number of rows processed by the statement. (This is not necessarily the number of rows actually inserted because Duplicates can be nonzero.) Duplicates indicates the number of rows that could not be inserted because they would duplicate some existing unique index value. Warnings indicates the number of attempts to insert column values that were problematic in some way. Warnings can occur under any of the following conditions:

  • Inserting NULL into a column that has been declared NOT NULL. For multiple-row INSERT statements or INSERT INTO ... SELECT statements, the column is set to the implicit default value for the column data type. This is 0 for numeric types, the empty string ('') for string types, and the “zero” value for date and time types. INSERT INTO ... SELECT statements are handled the same way as multiple-row inserts because the server does not examine the result set from the SELECT to see whether it returns a single row. (For a single-row INSERT, no warning occurs when NULL is inserted into a NOT NULL column. Instead, the statement fails with an error.)

  • Setting a numeric column to a value that lies outside the column's range. The value is clipped to the closest endpoint of the range.

  • Assigning a value such as '10.34 a' to a numeric column. The trailing nonnumeric text is stripped off and the remaining numeric part is inserted. If the string value has no leading numeric part, the column is set to 0.

  • Inserting a string into a string column (CHAR, VARCHAR, TEXT, or BLOB) that exceeds the column's maximum length. The value is truncated to the column's maximum length.

  • Inserting a value into a date or time column that is illegal for the data type. The column is set to the appropriate zero value for the type.

If you are using the C API, the information string can be obtained by invoking the mysql_info() function. See Section 22.9.3.35, “mysql_info().

If INSERT inserts a row into a table that has an AUTO_INCREMENT column, you can find the value used for that column by using the SQL LAST_INSERT_ID() function. From within the C API, use the mysql_insert_id() function. However, you should note that the two functions do not always behave identically. The behavior of INSERT statements with respect to AUTO_INCREMENT columns is discussed further in Section 11.14, “Information Functions”, and Section 22.9.3.37, “mysql_insert_id().

The INSERT statement supports the following modifiers:

  • If you use the DELAYED keyword, the server puts the row or rows to be inserted into a buffer, and the client issuing the INSERT DELAYED statement can then continue immediately. If the table is in use, the server holds the rows. When the table is free, the server begins inserting rows, checking periodically to see whether there are any new read requests for the table. If there are, the delayed row queue is suspended until the table becomes free again. See Section 12.2.5.2, “INSERT DELAYED Syntax”.

    DELAYED is ignored with INSERT ... SELECT or INSERT ... ON DUPLICATE KEY UPDATE.

    DELAYED is also disregarded for an INSERT that uses functions accessing tables or triggers, or that is called from a function or a trigger.

  • If you use the LOW_PRIORITY keyword, execution of the INSERT is delayed until no other clients are reading from the table. This includes other clients that began reading while existing clients are reading, and while the INSERT LOW_PRIORITY statement is waiting. It is possible, therefore, for a client that issues an INSERT LOW_PRIORITY statement to wait for a very long time (or even forever) in a read-heavy environment. (This is in contrast to INSERT DELAYED, which lets the client continue at once. Note that LOW_PRIORITY should normally not be used with MyISAM tables because doing so disables concurrent inserts. See Section 7.10.3, “Concurrent Inserts”.

    If you specify HIGH_PRIORITY, it overrides the effect of the --low-priority-updates option if the server was started with that option. It also causes concurrent inserts not to be used. See Section 7.10.3, “Concurrent Inserts”.

    LOW_PRIORITY and HIGH_PRIORITY affect only storage engines that use only table-level locking (such as MyISAM, MEMORY, and MERGE).

  • If you use the IGNORE keyword, errors that occur while executing the INSERT statement are treated as warnings instead. For example, without IGNORE, a row that duplicates an existing UNIQUE index or PRIMARY KEY value in the table causes a duplicate-key error and the statement is aborted. With IGNORE, the row still is not inserted, but no error is issued.

    IGNORE has a similar effect on inserts into partitioned tables where no partition matching a given value is found. Without IGNORE, such INSERT statements are aborted with an error; however, when INSERT IGNORE is used, the insert operation fails silently for the row containing the unmatched value, but any rows that are matched are inserted. For an example, see Section 18.2.2, “LIST Partitioning”.

    Data conversions that would trigger errors abort the statement if IGNORE is not specified. With IGNORE, invalid values are adjusted to the closest values and inserted; warnings are produced but the statement does not abort. You can determine with the mysql_info() C API function how many rows were actually inserted into the table.

  • If you specify ON DUPLICATE KEY UPDATE, and a row is inserted that would cause a duplicate value in a UNIQUE index or PRIMARY KEY, an UPDATE of the old row is performed. The affected-rows value per row is 1 if the row is inserted as a new row and 2 if an existing row is updated. See Section 12.2.5.3, “INSERT ... ON DUPLICATE KEY UPDATE Syntax”.

Inserting into a table requires the INSERT privilege for the table. If the ON DUPLICATE KEY UPDATE clause is used and a duplicate key causes an UPDATE to be performed instead, the statement requires the UPDATE privilege for the columns to be updated. For columns that are read but not modified you need only the SELECT privilege (such as for a column referenced only on the right hand side of an col_name=expr assignment in an ON DUPLICATE KEY UPDATE clause).

12.2.5.1. INSERT ... SELECT Syntax

INSERT [LOW_PRIORITY | HIGH_PRIORITY] [IGNORE]
    [INTO] tbl_name [(col_name,...)]
    SELECT ...
    [ ON DUPLICATE KEY UPDATE col_name=expr, ... ]

With INSERT ... SELECT, you can quickly insert many rows into a table from one or many tables. For example:

INSERT INTO tbl_temp2 (fld_id)
  SELECT tbl_temp1.fld_order_id
  FROM tbl_temp1 WHERE tbl_temp1.fld_order_id > 100;

The following conditions hold for a INSERT ... SELECT statements:

  • Specify IGNORE to ignore rows that would cause duplicate-key violations.

  • DELAYED is ignored with INSERT ... SELECT.

  • The target table of the INSERT statement may appear in the FROM clause of the SELECT part of the query. (This was not possible in some older versions of MySQL.) However, you cannot insert into a table and select from the same table in a subquery.

    When selecting from and inserting into a table at the same time, MySQL creates a temporary table to hold the rows from the SELECT and then inserts those rows into the target table. However, it remains true that you cannot use INSERT INTO t ... SELECT ... FROM t when t is a TEMPORARY table, because TEMPORARY tables cannot be referred to twice in the same statement (see Section C.5.7.2, “TEMPORARY Table Problems”).

  • AUTO_INCREMENT columns work as usual.

  • To ensure that the binary log can be used to re-create the original tables, MySQL does not permit concurrent inserts for INSERT ... SELECT statements.

  • To avoid ambiguous column reference problems when the SELECT and the INSERT refer to the same table, provide a unique alias for each table used in the SELECT part, and qualify column names in that part with the appropriate alias.

In the values part of ON DUPLICATE KEY UPDATE, you can refer to columns in other tables, as long as you do not use GROUP BY in the SELECT part. One side effect is that you must qualify nonunique column names in the values part.

The order in which rows are returned by a SELECT statement with no ORDER BY clause is not determined. This means that, when using replication, there is no guarantee that such a SELECT returns rows in the same order on the master and the slave; this can lead to inconsistencies between them. To prevent this from occurring, you should always write INSERT ... SELECT statements that are to be replicated as INSERT ... SELECT ... ORDER BY column. The choice of column does not matter as long as the same order for returning the rows is enforced on both the master and the slave. See also Section 17.4.1.12, “Replication and LIMIT.

12.2.5.2. INSERT DELAYED Syntax

INSERT DELAYED ...

The DELAYED option for the INSERT statement is a MySQL extension to standard SQL that is very useful if you have clients that cannot or need not wait for the INSERT to complete. This is a common situation when you use MySQL for logging and you also periodically run SELECT and UPDATE statements that take a long time to complete.

When a client uses INSERT DELAYED, it gets an okay from the server at once, and the row is queued to be inserted when the table is not in use by any other thread.

Another major benefit of using INSERT DELAYED is that inserts from many clients are bundled together and written in one block. This is much faster than performing many separate inserts.

Note that INSERT DELAYED is slower than a normal INSERT if the table is not otherwise in use. There is also the additional overhead for the server to handle a separate thread for each table for which there are delayed rows. This means that you should use INSERT DELAYED only when you are really sure that you need it.

The queued rows are held only in memory until they are inserted into the table. This means that if you terminate mysqld forcibly (for example, with kill -9) or if mysqld dies unexpectedly, any queued rows that have not been written to disk are lost.

There are some constraints on the use of DELAYED:

  • INSERT DELAYED works only with MyISAM, MEMORY, ARCHIVE, and BLACKHOLE tables. For engines that do not support DELAYED, an error occurs.

  • An error occurs for INSERT DELAYED if used with a table that has been locked with LOCK TABLES because the insert must be handled by a separate thread, not by the session that holds the lock.

  • For MyISAM tables, if there are no free blocks in the middle of the data file, concurrent SELECT and INSERT statements are supported. Under these circumstances, you very seldom need to use INSERT DELAYED with MyISAM.

  • INSERT DELAYED should be used only for INSERT statements that specify value lists. The server ignores DELAYED for INSERT ... SELECT or INSERT ... ON DUPLICATE KEY UPDATE statements.

  • Because the INSERT DELAYED statement returns immediately, before the rows are inserted, you cannot use LAST_INSERT_ID() to get the AUTO_INCREMENT value that the statement might generate.

  • DELAYED rows are not visible to SELECT statements until they actually have been inserted.

  • Prior to MySQL 5.5.7, INSERT DELAYED was treated as a normal INSERT if the statement inserted multiple rows, binary logging was enabled, and the global logging format was statement-based (that is, whenever binlog_format was set to STATEMENT). Beginning with MySQL 5.5.7, INSERT DELAYED is always handled as a simple INSERT (that is, without the DELAYED option) whenever the value of binlog_format is STATEMENT or MIXED. (In the latter case, the statement no longer triggers a switch to row-based logging, and so is logged using the statement-based format.)

    This does not apply when using row-based binary logging mode (binlog_format set to ROW), in which INSERT DELAYED statements are always executed using the DELAYED option as specified, and logged as row-update events.

  • DELAYED is ignored on slave replication servers, so that INSERT DELAYED is treated as a normal INSERT on slaves. This is because DELAYED could cause the slave to have different data than the master.

  • Pending INSERT DELAYED statements are lost if a table is write locked and ALTER TABLE is used to modify the table structure.

  • INSERT DELAYED is not supported for views.

  • INSERT DELAYED is not supported for partitioned tables.

The following describes in detail what happens when you use the DELAYED option to INSERT or REPLACE. In this description, the “thread” is the thread that received an INSERT DELAYED statement and “handler” is the thread that handles all INSERT DELAYED statements for a particular table.

  • When a thread executes a DELAYED statement for a table, a handler thread is created to process all DELAYED statements for the table, if no such handler already exists.

  • The thread checks whether the handler has previously acquired a DELAYED lock; if not, it tells the handler thread to do so. The DELAYED lock can be obtained even if other threads have a READ or WRITE lock on the table. However, the handler waits for all ALTER TABLE locks or FLUSH TABLES statements to finish, to ensure that the table structure is up to date.

  • The thread executes the INSERT statement, but instead of writing the row to the table, it puts a copy of the final row into a queue that is managed by the handler thread. Any syntax errors are noticed by the thread and reported to the client program.

  • The client cannot obtain from the server the number of duplicate rows or the AUTO_INCREMENT value for the resulting row, because the INSERT returns before the insert operation has been completed. (If you use the C API, the mysql_info() function does not return anything meaningful, for the same reason.)

  • The binary log is updated by the handler thread when the row is inserted into the table. In case of multiple-row inserts, the binary log is updated when the first row is inserted.

  • Each time that delayed_insert_limit rows are written, the handler checks whether any SELECT statements are still pending. If so, it permits these to execute before continuing.

  • When the handler has no more rows in its queue, the table is unlocked. If no new INSERT DELAYED statements are received within delayed_insert_timeout seconds, the handler terminates.

  • If more than delayed_queue_size rows are pending in a specific handler queue, the thread requesting INSERT DELAYED waits until there is room in the queue. This is done to ensure that mysqld does not use all memory for the delayed memory queue.

  • The handler thread shows up in the MySQL process list with delayed_insert in the Command column. It is killed if you execute a FLUSH TABLES statement or kill it with KILL thread_id. However, before exiting, it first stores all queued rows into the table. During this time it does not accept any new INSERT statements from other threads. If you execute an INSERT DELAYED statement after this, a new handler thread is created.

    Note that this means that INSERT DELAYED statements have higher priority than normal INSERT statements if there is an INSERT DELAYED handler running. Other update statements have to wait until the INSERT DELAYED queue is empty, someone terminates the handler thread (with KILL thread_id), or someone executes a FLUSH TABLES.

  • The following status variables provide information about INSERT DELAYED statements.

    Status VariableMeaning
    Delayed_insert_threadsNumber of handler threads
    Delayed_writesNumber of rows written with INSERT DELAYED
    Not_flushed_delayed_rowsNumber of rows waiting to be written

    You can view these variables by issuing a SHOW STATUS statement or by executing a mysqladmin extended-status command.

12.2.5.3. INSERT ... ON DUPLICATE KEY UPDATE Syntax

If you specify ON DUPLICATE KEY UPDATE, and a row is inserted that would cause a duplicate value in a UNIQUE index or PRIMARY KEY, an UPDATE of the old row is performed. For example, if column a is declared as UNIQUE and contains the value 1, the following two statements have identical effect:

INSERT INTO table (a,b,c) VALUES (1,2,3)
  ON DUPLICATE KEY UPDATE c=c+1;

UPDATE table SET c=c+1 WHERE a=1;

With ON DUPLICATE KEY UPDATE, the affected-rows value per row is 1 if the row is inserted as a new row and 2 if an existing row is updated.

If column b is also unique, the INSERT is equivalent to this UPDATE statement instead:

UPDATE table SET c=c+1 WHERE a=1 OR b=2 LIMIT 1;

If a=1 OR b=2 matches several rows, only one row is updated. In general, you should try to avoid using an ON DUPLICATE KEY UPDATE clause on tables with multiple unique indexes.

The ON DUPLICATE KEY UPDATE clause can contain multiple column assignments, separated by commas.

You can use the VALUES(col_name) function in the UPDATE clause to refer to column values from the INSERT portion of the INSERT ... ON DUPLICATE KEY UPDATE statement. In other words, VALUES(col_name) in the ON DUPLICATE KEY UPDATE clause refers to the value of col_name that would be inserted, had no duplicate-key conflict occurred. This function is especially useful in multiple-row inserts. The VALUES() function is meaningful only in INSERT ... UPDATE statements and returns NULL otherwise. Example:

INSERT INTO table (a,b,c) VALUES (1,2,3),(4,5,6)
  ON DUPLICATE KEY UPDATE c=VALUES(a)+VALUES(b);

That statement is identical to the following two statements:

INSERT INTO table (a,b,c) VALUES (1,2,3)
  ON DUPLICATE KEY UPDATE c=3;
INSERT INTO table (a,b,c) VALUES (4,5,6)
  ON DUPLICATE KEY UPDATE c=9;

If a table contains an AUTO_INCREMENT column and INSERT ... ON DUPLICATE KEY UPDATE inserts or updates a row, the LAST_INSERT_ID() function returns the AUTO_INCREMENT value.

The DELAYED option is ignored when you use ON DUPLICATE KEY UPDATE.

12.2.6. LOAD DATA INFILE Syntax

LOAD DATA [LOW_PRIORITY | CONCURRENT] [LOCAL] INFILE 'file_name'
    [REPLACE | IGNORE]
    INTO TABLE tbl_name
    [CHARACTER SET charset_name]
    [{FIELDS | COLUMNS}
        [TERMINATED BY 'string']
        [[OPTIONALLY] ENCLOSED BY 'char']
        [ESCAPED BY 'char']
    ]
    [LINES
        [STARTING BY 'string']
        [TERMINATED BY 'string']
    ]
    [IGNORE number LINES]
    [(col_name_or_user_var,...)]
    [SET col_name = expr,...]

The LOAD DATA INFILE statement reads rows from a text file into a table at a very high speed. The file name must be given as a literal string.

LOAD DATA INFILE is the complement of SELECT ... INTO OUTFILE. (See Section 12.2.9, “SELECT Syntax”.) To write data from a table to a file, use SELECT ... INTO OUTFILE. To read the file back into a table, use LOAD DATA INFILE. The syntax of the FIELDS and LINES clauses is the same for both statements. Both clauses are optional, but FIELDS must precede LINES if both are specified.

For more information about the efficiency of INSERT versus LOAD DATA INFILE and speeding up LOAD DATA INFILE, see Section 7.2.2.1, “Speed of INSERT Statements”.

The character set indicated by the character_set_database system variable is used to interpret the information in the file. SET NAMES and the setting of character_set_client do not affect interpretation of input. If the contents of the input file use a character set that differs from the default, it is usually preferable to specify the character set of the file by using the CHARACTER SET clause. A character set of binary specifies “no conversion.

LOAD DATA INFILE interprets all fields in the file as having the same character set, regardless of the data types of the columns into which field values are loaded. For proper interpretation of file contents, you must ensure that it was written with the correct character set. For example, if you write a data file with mysqldump -T or by issuing a SELECT ... INTO OUTFILE statement in mysql, be sure to use a --default-character-set option with mysqldump or mysql so that output is written in the character set to be used when the file is loaded with LOAD DATA INFILE.

Note that it is currently not possible to load data files that use the ucs2, utf16, or utf32 character set.

The character_set_filesystem system variable controls the interpretation of the file name.

You can also load data files by using the mysqlimport utility; it operates by sending a LOAD DATA INFILE statement to the server. The --local option causes mysqlimport to read data files from the client host. You can specify the --compress option to get better performance over slow networks if the client and server support the compressed protocol. See Section 4.5.5, “mysqlimport — A Data Import Program”.

If you use LOW_PRIORITY, execution of the LOAD DATA statement is delayed until no other clients are reading from the table. This affects only storage engines that use only table-level locking (such as MyISAM, MEMORY, and MERGE).

If you specify CONCURRENT with a MyISAM table that satisfies the condition for concurrent inserts (that is, it contains no free blocks in the middle), other threads can retrieve data from the table while LOAD DATA is executing. Using this option affects the performance of LOAD DATA a bit, even if no other thread is using the table at the same time.

Prior to MySQL 5.5.1, CONCURRENT was not replicated when using statement-based replication (see Bug#34628). However, it is replicated when using row-based replication, regardless of the version. See Section 17.4.1.13, “Replication and LOAD DATA INFILE, for more information.

The LOCAL keyword, if specified, is interpreted with respect to the client end of the connection:

  • If LOCAL is specified, the file is read by the client program on the client host and sent to the server. The file can be given as a full path name to specify its exact location. If given as a relative path name, the name is interpreted relative to the directory in which the client program was started.

  • If LOCAL is not specified, the file must be located on the server host and is read directly by the server. The server uses the following rules to locate the file:

    • If the file name is an absolute path name, the server uses it as given.

    • If the file name is a relative path name with one or more leading components, the server searches for the file relative to the server's data directory.

    • If a file name with no leading components is given, the server looks for the file in the database directory of the default database.

Note that, in the non-LOCAL case, these rules mean that a file named as ./myfile.txt is read from the server's data directory, whereas the file named as myfile.txt is read from the database directory of the default database. For example, if db1 is the default database, the following LOAD DATA statement reads the file data.txt from the database directory for db1, even though the statement explicitly loads the file into a table in the db2 database:

LOAD DATA INFILE 'data.txt' INTO TABLE db2.my_table;

Windows path names are specified using forward slashes rather than backslashes. If you do use backslashes, you must double them.

For security reasons, when reading text files located on the server, the files must either reside in the database directory or be readable by all. Also, to use LOAD DATA INFILE on server files, you must have the FILE privilege. See Section 5.4.1, “Privileges Provided by MySQL”. For non-LOCAL load operations, if the secure_file_priv system variable is set to a nonempty directory name, the file to be loaded must be located in that directory.

Using LOCAL is a bit slower than letting the server access the files directly, because the contents of the file must be sent over the connection by the client to the server. On the other hand, you do not need the FILE privilege to load local files.

With LOCAL, the default duplicate-key handling behavior is the same as if IGNORE is specified; this is because the server has no way to stop transmission of the file in the middle of the operation. IGNORE is explained further later in this section.

LOCAL works only if your server and your client both have been configured to permit it. For example, if mysqld was started with --local-infile=0, LOCAL does not work. See Section 5.3.5, “Security Issues with LOAD DATA LOCAL.

On Unix, if you need LOAD DATA to read from a pipe, you can use the following technique (the example loads a listing of the / directory into the table db1.t1):

mkfifo /mysql/data/db1/ls.dat
chmod 666 /mysql/data/db1/ls.dat
find / -ls > /mysql/data/db1/ls.dat &
mysql -e "LOAD DATA INFILE 'ls.dat' INTO TABLE t1" db1

Note that you must run the command that generates the data to be loaded and the mysql commands either on separate terminals, or run the data generation process in the background (as shown in the preceding example). If you do not do this, the pipe will block until data is read by the mysql process.

The REPLACE and IGNORE keywords control handling of input rows that duplicate existing rows on unique key values:

  • If you specify REPLACE, input rows replace existing rows. In other words, rows that have the same value for a primary key or unique index as an existing row. See Section 12.2.8, “REPLACE Syntax”.

  • If you specify IGNORE, input rows that duplicate an existing row on a unique key value are skipped. If you do not specify either option, the behavior depends on whether the LOCAL keyword is specified. Without LOCAL, an error occurs when a duplicate key value is found, and the rest of the text file is ignored. With LOCAL, the default behavior is the same as if IGNORE is specified; this is because the server has no way to stop transmission of the file in the middle of the operation.

If you want to ignore foreign key constraints during the load operation, you can issue a SET foreign_key_checks = 0 statement before executing LOAD DATA.

If you use LOAD DATA INFILE on an empty MyISAM table, all nonunique indexes are created in a separate batch (as for REPAIR TABLE). Normally, this makes LOAD DATA INFILE much faster when you have many indexes. In some extreme cases, you can create the indexes even faster by turning them off with ALTER TABLE ... DISABLE KEYS before loading the file into the table and using ALTER TABLE ... ENABLE KEYS to re-create the indexes after loading the file. See Section 7.2.2.1, “Speed of INSERT Statements”.

For both the LOAD DATA INFILE and SELECT ... INTO OUTFILE statements, the syntax of the FIELDS and LINES clauses is the same. Both clauses are optional, but FIELDS must precede LINES if both are specified.

If you specify a FIELDS clause, each of its subclauses (TERMINATED BY, [OPTIONALLY] ENCLOSED BY, and ESCAPED BY) is also optional, except that you must specify at least one of them.

If you specify no FIELDS or LINES clause, the defaults are the same as if you had written this:

FIELDS TERMINATED BY '\t' ENCLOSED BY '' ESCAPED BY '\\'
LINES TERMINATED BY '\n' STARTING BY ''

(Backslash is the MySQL escape character within strings in SQL statements, so to specify a literal backslash, you must specify two backslashes for the value to be interpreted as a single backslash. The escape sequences '\t' and '\n' specify tab and newline characters, respectively.)

In other words, the defaults cause LOAD DATA INFILE to act as follows when reading input:

  • Look for line boundaries at newlines.

  • Do not skip over any line prefix.

  • Break lines into fields at tabs.

  • Do not expect fields to be enclosed within any quoting characters.

  • Interpret characters preceded by the escape character “\” as escape sequences. For example, “\t”, “\n”, and “\\” signify tab, newline, and backslash, respectively. See the discussion of FIELDS ESCAPED BY later for the full list of escape sequences.

Conversely, the defaults cause SELECT ... INTO OUTFILE to act as follows when writing output:

  • Write tabs between fields.

  • Do not enclose fields within any quoting characters.

  • Use “\” to escape instances of tab, newline, or “\” that occur within field values.

  • Write newlines at the ends of lines.

Note

If you have generated the text file on a Windows system, you might have to use LINES TERMINATED BY '\r\n' to read the file properly, because Windows programs typically use two characters as a line terminator. Some programs, such as WordPad, might use \r as a line terminator when writing files. To read such files, use LINES TERMINATED BY '\r'.

If all the lines you want to read in have a common prefix that you want to ignore, you can use LINES STARTING BY 'prefix_string' to skip over the prefix, and anything before it. If a line does not include the prefix, the entire line is skipped. Suppose that you issue the following statement:

LOAD DATA INFILE '/tmp/test.txt' INTO TABLE test
  FIELDS TERMINATED BY ','  LINES STARTING BY 'xxx';

If the data file looks like this:

xxx"abc",1
something xxx"def",2
"ghi",3

The resulting rows will be ("abc",1) and ("def",2). The third row in the file is skipped because it does not contain the prefix.

The IGNORE number LINES option can be used to ignore lines at the start of the file. For example, you can use IGNORE 1 LINES to skip over an initial header line containing column names:

LOAD DATA INFILE '/tmp/test.txt' INTO TABLE test IGNORE 1 LINES;

When you use SELECT ... INTO OUTFILE in tandem with LOAD DATA INFILE to write data from a database into a file and then read the file back into the database later, the field- and line-handling options for both statements must match. Otherwise, LOAD DATA INFILE will not interpret the contents of the file properly. Suppose that you use SELECT ... INTO OUTFILE to write a file with fields delimited by commas:

SELECT * INTO OUTFILE 'data.txt'
  FIELDS TERMINATED BY ','
  FROM table2;

To read the comma-delimited file back in, the correct statement would be:

LOAD DATA INFILE 'data.txt' INTO TABLE table2
  FIELDS TERMINATED BY ',';

If instead you tried to read in the file with the statement shown following, it wouldn't work because it instructs LOAD DATA INFILE to look for tabs between fields:

LOAD DATA INFILE 'data.txt' INTO TABLE table2
  FIELDS TERMINATED BY '\t';

The likely result is that each input line would be interpreted as a single field.

LOAD DATA INFILE can be used to read files obtained from external sources. For example, many programs can export data in comma-separated values (CSV) format, such that lines have fields separated by commas and enclosed within double quotation marks, with an initial line of column names. If the lines in such a file are terminated by carriage return/newline pairs, the statement shown here illustrates the field- and line-handling options you would use to load the file:

LOAD DATA INFILE 'data.txt' INTO TABLE tbl_name
  FIELDS TERMINATED BY ',' ENCLOSED BY '"'
  LINES TERMINATED BY '\r\n'
  IGNORE 1 LINES;

If the input values are not necessarily enclosed within quotation marks, use OPTIONALLY before the ENCLOSED BY keywords.

Any of the field- or line-handling options can specify an empty string (''). If not empty, the FIELDS [OPTIONALLY] ENCLOSED BY and FIELDS ESCAPED BY values must be a single character. The FIELDS TERMINATED BY, LINES STARTING BY, and LINES TERMINATED BY values can be more than one character. For example, to write lines that are terminated by carriage return/linefeed pairs, or to read a file containing such lines, specify a LINES TERMINATED BY '\r\n' clause.

To read a file containing jokes that are separated by lines consisting of %%, you can do this

CREATE TABLE jokes
  (a INT NOT NULL AUTO_INCREMENT PRIMARY KEY,
  joke TEXT NOT NULL);
LOAD DATA INFILE '/tmp/jokes.txt' INTO TABLE jokes
  FIELDS TERMINATED BY ''
  LINES TERMINATED BY '\n%%\n' (joke);

FIELDS [OPTIONALLY] ENCLOSED BY controls quoting of fields. For output (SELECT ... INTO OUTFILE), if you omit the word OPTIONALLY, all fields are enclosed by the ENCLOSED BY character. An example of such output (using a comma as the field delimiter) is shown here:

"1","a string","100.20"
"2","a string containing a , comma","102.20"
"3","a string containing a \" quote","102.20"
"4","a string containing a \", quote and comma","102.20"

If you specify OPTIONALLY, the ENCLOSED BY character is used only to enclose values from columns that have a string data type (such as CHAR, BINARY, TEXT, or ENUM):

1,"a string",100.20
2,"a string containing a , comma",102.20
3,"a string containing a \" quote",102.20
4,"a string containing a \", quote and comma",102.20

Note that occurrences of the ENCLOSED BY character within a field value are escaped by prefixing them with the ESCAPED BY character. Also note that if you specify an empty ESCAPED BY value, it is possible to inadvertently generate output that cannot be read properly by LOAD DATA INFILE. For example, the preceding output just shown would appear as follows if the escape character is empty. Observe that the second field in the fourth line contains a comma following the quote, which (erroneously) appears to terminate the field:

1,"a string",100.20
2,"a string containing a , comma",102.20
3,"a string containing a " quote",102.20
4,"a string containing a ", quote and comma",102.20

For input, the ENCLOSED BY character, if present, is stripped from the ends of field values. (This is true regardless of whether OPTIONALLY is specified; OPTIONALLY has no effect on input interpretation.) Occurrences of the ENCLOSED BY character preceded by the ESCAPED BY character are interpreted as part of the current field value.

If the field begins with the ENCLOSED BY character, instances of that character are recognized as terminating a field value only if followed by the field or line TERMINATED BY sequence. To avoid ambiguity, occurrences of the ENCLOSED BY character within a field value can be doubled and are interpreted as a single instance of the character. For example, if ENCLOSED BY '"' is specified, quotation marks are handled as shown here:

"The ""BIG"" boss"  -> The "BIG" boss
The "BIG" boss      -> The "BIG" boss
The ""BIG"" boss    -> The ""BIG"" boss

FIELDS ESCAPED BY controls how to read or write special characters:

  • For input, if the FIELDS ESCAPED BY character is not empty, occurrences of that character are stripped and the following character is taken literally as part of a field value. Some two-character sequences that are exceptions, where the first character is the escape character. These sequences are shown in the following table (using “\” for the escape character). The rules for NULL handling are described later in this section.

    CharacterEscape Sequence
    \0 An ASCII NUL (0x00) character
    \b A backspace character
    \n A newline (linefeed) character
    \r A carriage return character
    \t A tab character.
    \Z ASCII 26 (Control-Z)
    \N NULL

    For more information about “\”-escape syntax, see Section 8.1.1, “Strings”.

    If the FIELDS ESCAPED BY character is empty, escape-sequence interpretation does not occur.

  • For output, if the FIELDS ESCAPED BY character is not empty, it is used to prefix the following characters on output:

    • The FIELDS ESCAPED BY character

    • The FIELDS [OPTIONALLY] ENCLOSED BY character

    • The first character of the FIELDS TERMINATED BY and LINES TERMINATED BY values

    • ASCII 0 (what is actually written following the escape character is ASCII “0”, not a zero-valued byte)

    If the FIELDS ESCAPED BY character is empty, no characters are escaped and NULL is output as NULL, not \N. It is probably not a good idea to specify an empty escape character, particularly if field values in your data contain any of the characters in the list just given.

In certain cases, field- and line-handling options interact:

  • If LINES TERMINATED BY is an empty string and FIELDS TERMINATED BY is nonempty, lines are also terminated with FIELDS TERMINATED BY.

  • If the FIELDS TERMINATED BY and FIELDS ENCLOSED BY values are both empty (''), a fixed-row (nondelimited) format is used. With fixed-row format, no delimiters are used between fields (but you can still have a line terminator). Instead, column values are read and written using a field width wide enough to hold all values in the field. For TINYINT, SMALLINT, MEDIUMINT, INT, and BIGINT, the field widths are 4, 6, 8, 11, and 20, respectively, no matter what the declared display width is.

    LINES TERMINATED BY is still used to separate lines. If a line does not contain all fields, the rest of the columns are set to their default values. If you do not have a line terminator, you should set this to ''. In this case, the text file must contain all fields for each row.

    Fixed-row format also affects handling of NULL values, as described later. Note that fixed-size format does not work if you are using a multi-byte character set.

Handling of NULL values varies according to the FIELDS and LINES options in use:

  • For the default FIELDS and LINES values, NULL is written as a field value of \N for output, and a field value of \N is read as NULL for input (assuming that the ESCAPED BY character is “\”).

  • If FIELDS ENCLOSED BY is not empty, a field containing the literal word NULL as its value is read as a NULL value. This differs from the word NULL enclosed within FIELDS ENCLOSED BY characters, which is read as the string 'NULL'.

  • If FIELDS ESCAPED BY is empty, NULL is written as the word NULL.

  • With fixed-row format (which is used when FIELDS TERMINATED BY and FIELDS ENCLOSED BY are both empty), NULL is written as an empty string. Note that this causes both NULL values and empty strings in the table to be indistinguishable when written to the file because both are written as empty strings. If you need to be able to tell the two apart when reading the file back in, you should not use fixed-row format.

An attempt to load NULL into a NOT NULL column causes assignment of the implicit default value for the column's data type and a warning, or an error in strict SQL mode. Implicit default values are discussed in Section 10.1.4, “Data Type Default Values”.

Some cases are not supported by LOAD DATA INFILE:

  • Fixed-size rows (FIELDS TERMINATED BY and FIELDS ENCLOSED BY both empty) and BLOB or TEXT columns.

  • If you specify one separator that is the same as or a prefix of another, LOAD DATA INFILE cannot interpret the input properly. For example, the following FIELDS clause would cause problems:

    FIELDS TERMINATED BY '"' ENCLOSED BY '"'
    
  • If FIELDS ESCAPED BY is empty, a field value that contains an occurrence of FIELDS ENCLOSED BY or LINES TERMINATED BY followed by the FIELDS TERMINATED BY value causes LOAD DATA INFILE to stop reading a field or line too early. This happens because LOAD DATA INFILE cannot properly determine where the field or line value ends.

The following example loads all columns of the persondata table:

LOAD DATA INFILE 'persondata.txt' INTO TABLE persondata;

By default, when no column list is provided at the end of the LOAD DATA INFILE statement, input lines are expected to contain a field for each table column. If you want to load only some of a table's columns, specify a column list:

LOAD DATA INFILE 'persondata.txt' INTO TABLE persondata (col1,col2,...);

You must also specify a column list if the order of the fields in the input file differs from the order of the columns in the table. Otherwise, MySQL cannot tell how to match input fields with table columns.

The column list can contain either column names or user variables. With user variables, the SET clause enables you to perform transformations on their values before assigning the result to columns.

User variables in the SET clause can be used in several ways. The following example uses the first input column directly for the value of t1.column1, and assigns the second input column to a user variable that is subjected to a division operation before being used for the value of t1.column2:

LOAD DATA INFILE 'file.txt'
  INTO TABLE t1
  (column1, @var1)
  SET column2 = @var1/100;

The SET clause can be used to supply values not derived from the input file. The following statement sets column3 to the current date and time:

LOAD DATA INFILE 'file.txt'
  INTO TABLE t1
  (column1, column2)
  SET column3 = CURRENT_TIMESTAMP;

You can also discard an input value by assigning it to a user variable and not assigning the variable to a table column:

LOAD DATA INFILE 'file.txt'
  INTO TABLE t1
  (column1, @dummy, column2, @dummy, column3);

Use of the column/variable list and SET clause is subject to the following restrictions:

  • Assignments in the SET clause should have only column names on the left hand side of assignment operators.

  • You can use subqueries in the right hand side of SET assignments. A subquery that returns a value to be assigned to a column may be a scalar subquery only. Also, you cannot use a subquery to select from the table that is being loaded.

  • Lines ignored by an IGNORE clause are not processed for the column/variable list or SET clause.

  • User variables cannot be used when loading data with fixed-row format because user variables do not have a display width.

When processing an input line, LOAD DATA splits it into fields and uses the values according to the column/variable list and the SET clause, if they are present. Then the resulting row is inserted into the table. If there are BEFORE INSERT or AFTER INSERT triggers for the table, they are activated before or after inserting the row, respectively.

If an input line has too many fields, the extra fields are ignored and the number of warnings is incremented.

If an input line has too few fields, the table columns for which input fields are missing are set to their default values. Default value assignment is described in Section 10.1.4, “Data Type Default Values”.

An empty field value is interpreted differently than if the field value is missing:

  • For string types, the column is set to the empty string.

  • For numeric types, the column is set to 0.

  • For date and time types, the column is set to the appropriate “zero” value for the type. See Section 10.3, “Date and Time Types”.

These are the same values that result if you assign an empty string explicitly to a string, numeric, or date or time type explicitly in an INSERT or UPDATE statement.

TIMESTAMP columns are set to the current date and time only if there is a NULL value for the column (that is, \N) and the column is not declared to permit NULL values, or if the TIMESTAMP column's default value is the current timestamp and it is omitted from the field list when a field list is specified.

LOAD DATA INFILE regards all input as strings, so you cannot use numeric values for ENUM or SET columns the way you can with INSERT statements. All ENUM and SET values must be specified as strings.

BIT values cannot be loaded using binary notation (for example, b'011010'). To work around this, specify the values as regular integers and use the SET clause to convert them so that MySQL performs a numeric type conversion and loads them into the BIT column properly:

shell> cat /tmp/bit_test.txt
2
127
shell> mysql test
mysql> LOAD DATA INFILE '/tmp/bit_test.txt'
    -> INTO TABLE bit_test (@var1) SET b= CAST(@var1 AS UNSIGNED);
Query OK, 2 rows affected (0.00 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

mysql> SELECT BIN(b+0) FROM bit_test;
+----------+
| bin(b+0) |
+----------+
| 10       |
| 1111111  |
+----------+
2 rows in set (0.00 sec)

When the LOAD DATA INFILE statement finishes, it returns an information string in the following format:

Records: 1  Deleted: 0  Skipped: 0  Warnings: 0

If you are using the C API, you can get information about the statement by calling the mysql_info() function. See Section 22.9.3.35, “mysql_info().

Warnings occur under the same circumstances as when values are inserted using the INSERT statement (see Section 12.2.5, “INSERT Syntax”), except that LOAD DATA INFILE also generates warnings when there are too few or too many fields in the input row. The warnings are not stored anywhere; the number of warnings can be used only as an indication of whether everything went well.

You can use SHOW WARNINGS to get a list of the first max_error_count warnings as information about what went wrong. See Section 12.4.5.41, “SHOW WARNINGS Syntax”.

12.2.7. LOAD XML Syntax

LOAD XML [LOW_PRIORITY | CONCURRENT] [LOCAL] INFILE 'file_name'
[REPLACE | IGNORE]
INTO TABLE [db_name.]tbl_name
[CHARACTER SET charset_name]
[ROWS IDENTIFIED BY '<tagname>']
[IGNORE number [LINES | ROWS]]
[(column_or_user_var,...)]
[SET col_name = expr,...]

The LOAD XML statement reads data from an XML file into a table. The file_name must be given as a literal string. The tagname in the optional ROWS IDENTIFIED BY clause must also be given as a literal string, and must be surrounded by angle brackets (< and >).

LOAD XML acts as the complement of running the mysql client in XML output mode (that is, starting the client with the --xml option). To write data from a table to an XML file, use a command such as the following one from the system shell:

shell> mysql --xml -e 'SELECT * FROM mytable' > file.xml

To read the file back into a table, use LOAD XML INFILE. By default, the <row> element is considered to be the equivalent of a database table row; this can be changed using the ROWS IDENTIFIED BY clause.

This statement supports three different XML formats:

  • Column names as attributes and column values as attribute values:

    <row column1="value1" column2="value2" .../>
    
  • Column names as tags and column values as the content of these tags:

    <row>
      <column1>value1</column1>
      <column2>value2</column2>
    </row>
    
  • Column names are the name attributes of <field> tags, and values are the contents of these tags:

    <row>
      <field name='column1'>value1</field>
      <field name='column2'>value2</field>
    </row>
    

    This is the format used by other MySQL tools, such as mysqldump.

All 3 formats can be used in the same XML file; the import routine automatically detects the format for each row and interpets it correctly. Tags are matched based on the tag or attribute name and the column name.

The following clauses work essentially the same way for LOAD XML as they do for LOAD DATA:

  • LOW_PRIORITY or CONCURRENT

  • LOCAL

  • REPLACE or IGNORE

  • CHARACTER SET

  • (column_or_user_var,...)

  • SET

See Section 12.2.6, “LOAD DATA INFILE Syntax”, for more information about these clauses.

The IGNORE number LINES or IGNORE number ROWS clause causes the first number rows in the XML file to be skipped. It is analogous to the LOAD DATA statement's IGNORE ... LINES clause.

To illustrate how this statement is used, suppose that we have a table created as follows:

USE test;

CREATE TABLE person (
    person_id INT NOT NULL PRIMARY KEY,
    fname VARCHAR(40) NULL,
    lname VARCHAR(40) NULL,
    created TIMESTAMP
);

Suppose further that this table is initially empty.

Now suppose that we have a simple XML file person.xml, whose contents are as shown here:

<?xml version="1.0"?>
<list>
  <person person_id="1" fname="Pekka" lname="Nousiainen"/>
  <person person_id="2" fname="Jonas" lname="Oreland"/>
  <person person_id="3"><fname>Mikael</fname><lname>Ronström</lname></person>
  <person person_id="4"><fname>Lars</fname><lname>Thalmann</lname></person>
  <person><field name="person_id">5</field><field name="fname">Tomas</field><field name="lname">Ulin</field></person>
  <person><field name="person_id">6</field><field name="fname">Martin</field><field name="lname">Sköld</field></person>
</list>

Each of the permissible XML formats discussed previously is represented in this example file.

To import the data in person.xml into the person table, you can use this statement:

mysql> LOAD XML LOCAL INFILE 'person.xml'
    ->   INTO TABLE person
    ->   ROWS IDENTIFIED BY '<person>';

Query OK, 6 rows affected (0.00 sec)
Records: 6  Deleted: 0  Skipped: 0  Warnings: 0

Here, we assume that person.xml is located in the MySQL data directory. If the file cannot be found, the following error results:

ERROR 2 (HY000): File '/person.xml' not found (Errcode: 2)

The ROWS IDENTIFIED BY '<person>' clause means that each <person> element in the XML file is considered equivalent to a row in the table into which the data is to be imported. In this case, this is the person table in the test database.

As can be seen by the response from the server, 6 rows were imported into the test.person table. This can be verified by a simple SELECT statement:

mysql> SELECT * FROM person;
+-----------+--------+------------+---------------------+
| person_id | fname  | lname      | created             |
+-----------+--------+------------+---------------------+
|         1 | Pekka  | Nousiainen | 2007-07-13 16:18:47 |
|         2 | Jonas  | Oreland    | 2007-07-13 16:18:47 |
|         3 | Mikael | Ronström   | 2007-07-13 16:18:47 |
|         4 | Lars   | Thalmann   | 2007-07-13 16:18:47 |
|         5 | Tomas  | Ulin       | 2007-07-13 16:18:47 |
|         6 | Martin | Sköld      | 2007-07-13 16:18:47 |
+-----------+--------+------------+---------------------+
6 rows in set (0.00 sec)

This shows, as stated earlier in this section, that any or all of the 3 permitted XML formats may appear in a single file and be read in using LOAD XML.

The inverse of the above operation—that is, dumping MySQL table data into an XML file—can be accomplished using the mysql client from the system shell, as shown here:

Note

The --xml option causes the mysql client to use XML formatting for its output; the -e option causes the client to execute the SQL statement immediately following the option.

shell> mysql --xml -e "SELECT * FROM test.person" > person-dump.xml
shell> cat person-dump.xml
<?xml version="1.0"?>

<resultset statement="SELECT * FROM test.person" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
  <row>
        <field name="person_id">1</field>
        <field name="fname">Pekka</field>
        <field name="lname">Nousiainen</field>
        <field name="created">2007-07-13 16:18:47</field>
  </row>

  <row>
        <field name="person_id">2</field>
        <field name="fname">Jonas</field>
        <field name="lname">Oreland</field>
        <field name="created">2007-07-13 16:18:47</field>
  </row>

  <row>
        <field name="person_id">3</field>
        <field name="fname">Mikael</field>
        <field name="lname">Ronström</field>
        <field name="created">2007-07-13 16:18:47</field>
  </row>

  <row>
        <field name="person_id">4</field>
        <field name="fname">Lars</field>
        <field name="lname">Thalmann</field>
        <field name="created">2007-07-13 16:18:47</field>
  </row>

  <row>
        <field name="person_id">5</field>
        <field name="fname">Tomas</field>
        <field name="lname">Ulin</field>
        <field name="created">2007-07-13 16:18:47</field>
  </row>

  <row>
        <field name="person_id">6</field>
        <field name="fname">Martin</field>
        <field name="lname">Sköld</field>
        <field name="created">2007-07-13 16:18:47</field>
  </row>
</resultset>

You can verify that the dump is valid by creating a copy of the person and then importing the dump file into the new table, like this:

mysql> USE test;
mysql> CREATE TABLE person2 LIKE person;
Query OK, 0 rows affected (0.00 sec)

mysql> LOAD XML LOCAL INFILE 'person-dump.xml'
    ->   INTO TABLE person2;
Query OK, 6 rows affected (0.01 sec)
Records: 6  Deleted: 0  Skipped: 0  Warnings: 0

mysql> SELECT * FROM person2;
+-----------+--------+------------+---------------------+
| person_id | fname  | lname      | created             |
+-----------+--------+------------+---------------------+
|         1 | Pekka  | Nousiainen | 2007-07-13 16:18:47 |
|         2 | Jonas  | Oreland    | 2007-07-13 16:18:47 |
|         3 | Mikael | Ronström   | 2007-07-13 16:18:47 |
|         4 | Lars   | Thalmann   | 2007-07-13 16:18:47 |
|         5 | Tomas  | Ulin       | 2007-07-13 16:18:47 |
|         6 | Martin | Sköld      | 2007-07-13 16:18:47 |
+-----------+--------+------------+---------------------+
6 rows in set (0.00 sec)

Using a ROWS IDENTIFIED BY '<tagname>' clause, it is possible to import data from the same XML file into database tables with different definitions. For this example, suppose that you have a file named address.xml which contains the following XML:

<?xml version="1.0"?>

<list>
  <person person_id="1">
    <fname>Robert</fname>
    <lname>Jones</lname>
    <address address_id="1" street="Mill Creek Road" zip="45365" city="Sidney"/>
    <address address_id="2" street="Main Street" zip="28681" city="Taylorsville"/>
  </person>

  <person person_id="2">
    <fname>Mary</fname>
    <lname>Smith</lname>
    <address address_id="3" street="River Road" zip="80239" city="Denver"/>
    <!-- <address address_id="4" street="North Street" zip="37920" city="Knoxville"/> -->
  </person>

</list>

You can again use the test.person table as defined previously in this section, after clearing all the existing records from the table and then showing its structure as shown here:

mysql< TRUNCATE person;
Query OK, 0 rows affected (0.04 sec)

mysql< SHOW CREATE TABLE person\G
*************************** 1. row ***************************
       Table: person
Create Table: CREATE TABLE `person` (
  `person_id` int(11) NOT NULL,
  `fname` varchar(40) DEFAULT NULL,
  `lname` varchar(40) DEFAULT NULL,
  `created` timestamp NOT NULL DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
  PRIMARY KEY (`person_id`)
) ENGINE=MyISAM DEFAULT CHARSET=latin1
1 row in set (0.00 sec)

Now create an address table in the test database using the following CREATE TABLE statement:

CREATE TABLE address (
    address_id INT NOT NULL PRIMARY KEY,
    person_id INT NULL,
    street VARCHAR(40) NULL,
    zip INT NULL,
    city VARCHAR(40) NULL,
    created TIMESTAMP
);

To import the data from the XML file into the person table, execute the following LOAD XML statement, which specifies that rows are to be specified by the <person> element, as shown here;

mysql> LOAD XML LOCAL INFILE 'address.xml'
    ->   INTO TABLE person
    ->   ROWS IDENTIFIED BY '<person>';
Query OK, 2 rows affected (0.00 sec)
Records: 2  Deleted: 0  Skipped: 0  Warnings: 0

You can verify that the records were imported using a SELECT statement:

mysql> SELECT * FROM person;
+-----------+--------+-------+---------------------+
| person_id | fname  | lname | created             |
+-----------+--------+-------+---------------------+
|         1 | Robert | Jones | 2007-07-24 17:37:06 |
|         2 | Mary   | Smith | 2007-07-24 17:37:06 |
+-----------+--------+-------+---------------------+
2 rows in set (0.00 sec)

Since the <address> elements in the XML file have no corresponding columns in the person table, they are skipped.

To import the data from the <address> elements into the address table, use the LOAD XML statement shown here:

mysql> LOAD XML LOCAL INFILE 'address.xml'
    ->   INTO TABLE address
    ->   ROWS IDENTIFIED BY '<address>';
Query OK, 3 rows affected (0.00 sec)
Records: 3  Deleted: 0  Skipped: 0  Warnings: 0

You can see that the data was imported using a SELECT statement such as this one:

mysql> SELECT * FROM address;
+------------+-----------+-----------------+-------+--------------+---------------------+
| address_id | person_id | street          | zip   | city         | created             |
+------------+-----------+-----------------+-------+--------------+---------------------+
|          1 |         1 | Mill Creek Road | 45365 | Sidney       | 2007-07-24 17:37:37 |
|          2 |         1 | Main Street     | 28681 | Taylorsville | 2007-07-24 17:37:37 |
|          3 |         2 | River Road      | 80239 | Denver       | 2007-07-24 17:37:37 |
+------------+-----------+-----------------+-------+--------------+---------------------+
3 rows in set (0.00 sec)

The data from the <address> element that is enclosed in XML comments is not imported. However, since there is a person_id column in the address table, the value of the person_id attribute from the parent <person> element for each <address> is imported into the address table.

Security Considerations.  As with the LOAD DATA statement, the transfer of the XML file from the client host to the server host is initiated by the MySQL server. In theory, a patched server could be built that would tell the client program to transfer a file of the server's choosing rather than the file named by the client in the LOAD XML statement. Such a server could access any file on the client host to which the client user has read access.

In a Web environment, clients usually connect to MySQL from a Web server. A user that can run any command against the MySQL server can use LOAD XML LOCAL to read any files to which the Web server process has read access. In this environment, the client with respect to the MySQL server is actually the Web server, not the remote program being run by the user who connects to the Web server.

You can disable loading of XML files from clients by starting the server with --local-infile=0 or --local-infile=OFF. This option can also be used when starting the mysql client to disable LOAD XML for the duration of the client session.

To prevent a client from loading XML files from the server, do not grant the FILE privilege to the corresponding MySQL user account, or revoke this privilege if the client user account already has it.

Important

Revoking the FILE privilege (or not granting it in the first place) keeps the user only from executing the LOAD XML INFILE statement (as well as the LOAD_FILE() function; it does not prevent the user from executing LOAD XML LOCAL INFILE. To disallow this statement, you must start the server or the client with --local-infile=OFF.

In other words, the FILE privilege affects only whether the client can read files on the server; it has no bearing on whether the client can read files on the local file system.

12.2.8. REPLACE Syntax

REPLACE [LOW_PRIORITY | DELAYED]
    [INTO] tbl_name [(col_name,...)]
    {VALUES | VALUE} ({expr | DEFAULT},...),(...),...

Or:

REPLACE [LOW_PRIORITY | DELAYED]
    [INTO] tbl_name
    SET col_name={expr | DEFAULT}, ...

Or:

REPLACE [LOW_PRIORITY | DELAYED]
    [INTO] tbl_name [(col_name,...)]
    SELECT ...

REPLACE works exactly like INSERT, except that if an old row in the table has the same value as a new row for a PRIMARY KEY or a UNIQUE index, the old row is deleted before the new row is inserted. See Section 12.2.5, “INSERT Syntax”.

REPLACE is a MySQL extension to the SQL standard. It either inserts, or deletes and inserts. For another MySQL extension to standard SQL—that either inserts or updates—see Section 12.2.5.3, “INSERT ... ON DUPLICATE KEY UPDATE Syntax”.

Note that unless the table has a PRIMARY KEY or UNIQUE index, using a REPLACE statement makes no sense. It becomes equivalent to INSERT, because there is no index to be used to determine whether a new row duplicates another.

Values for all columns are taken from the values specified in the REPLACE statement. Any missing columns are set to their default values, just as happens for INSERT. You cannot refer to values from the current row and use them in the new row. If you use an assignment such as SET col_name = col_name + 1, the reference to the column name on the right hand side is treated as DEFAULT(col_name), so the assignment is equivalent to SET col_name = DEFAULT(col_name) + 1.

To use REPLACE, you must have both the INSERT and DELETE privileges for the table.

The REPLACE statement returns a count to indicate the number of rows affected. This is the sum of the rows deleted and inserted. If the count is 1 for a single-row REPLACE, a row was inserted and no rows were deleted. If the count is greater than 1, one or more old rows were deleted before the new row was inserted. It is possible for a single row to replace more than one old row if the table contains multiple unique indexes and the new row duplicates values for different old rows in different unique indexes.

The affected-rows count makes it easy to determine whether REPLACE only added a row or whether it also replaced any rows: Check whether the count is 1 (added) or greater (replaced).

If you are using the C API, the affected-rows count can be obtained using the mysql_affected_rows() function.

Currently, you cannot replace into a table and select from the same table in a subquery.

MySQL uses the following algorithm for REPLACE (and LOAD DATA ... REPLACE):

  1. Try to insert the new row into the table

  2. While the insertion fails because a duplicate-key error occurs for a primary key or unique index:

    1. Delete from the table the conflicting row that has the duplicate key value

    2. Try again to insert the new row into the table

It is possible that in the case of a duplicate-key error, a storage engine may perform the REPLACE as an update rather than a delete plus insert, but the semantics are the same. There are no user-visible effects other than a possible difference in how the storage engine increments Handler_xxx status variables.

12.2.9. SELECT Syntax

SELECT
    [ALL | DISTINCT | DISTINCTROW ]
      [HIGH_PRIORITY]
      [STRAIGHT_JOIN]
      [SQL_SMALL_RESULT] [SQL_BIG_RESULT] [SQL_BUFFER_RESULT]
      [SQL_CACHE | SQL_NO_CACHE] [SQL_CALC_FOUND_ROWS]
    select_expr [, select_expr ...]
    [FROM table_references
    [WHERE where_condition]
    [GROUP BY {col_name | expr | position}
      [ASC | DESC], ... [WITH ROLLUP]]
    [HAVING where_condition]
    [ORDER BY {col_name | expr | position}
      [ASC | DESC], ...]
    [LIMIT {[offset,] row_count | row_count OFFSET offset}]
    [PROCEDURE procedure_name(argument_list)]
    [INTO OUTFILE 'file_name'
        [CHARACTER SET charset_name]
        export_options
      | INTO DUMPFILE 'file_name'
      | INTO var_name [, var_name]]
    [FOR UPDATE | LOCK IN SHARE MODE]]

SELECT is used to retrieve rows selected from one or more tables, and can include UNION statements and subqueries. See Section 12.2.9.3, “UNION Syntax”, and Section 12.2.10, “Subquery Syntax”.

The most commonly used clauses of SELECT statements are these:

  • Each select_expr indicates a column that you want to retrieve. There must be at least one select_expr.

  • table_references indicates the table or tables from which to retrieve rows. Its syntax is described in Section 12.2.9.1, “JOIN Syntax”.

  • The WHERE clause, if given, indicates the condition or conditions that rows must satisfy to be selected. where_condition is an expression that evaluates to true for each row to be selected. The statement selects all rows if there is no WHERE clause.

    In the WHERE expression, you can use any of the functions and operators that MySQL supports, except for aggregate (summary) functions. See Section 8.5, “Expression Syntax”, and Chapter 11, Functions and Operators.

SELECT can also be used to retrieve rows computed without reference to any table.

For example:

mysql> SELECT 1 + 1;
        -> 2

You are permitted to specify DUAL as a dummy table name in situations where no tables are referenced:

mysql> SELECT 1 + 1 FROM DUAL;
        -> 2

DUAL is purely for the convenience of people who require that all SELECT statements should have FROM and possibly other clauses. MySQL may ignore the clauses. MySQL does not require FROM DUAL if no tables are referenced.

In general, clauses used must be given in exactly the order shown in the syntax description. For example, a HAVING clause must come after any GROUP BY clause and before any ORDER BY clause. The exception is that the INTO clause can appear either as shown in the syntax description or immediately following the select_expr list.

The list of select_expr terms comprises the select list that indicates which columns to retrieve. Terms specify a column or expression or can use *-shorthand:

  • A select list consisting only of a single unqualified * can be used as shorthand to select all columns from all tables:

    SELECT * FROM t1 INNER JOIN t2 ...
    
  • tbl_name.* can be used as a qualified shorthand to select all columns from the named table:

    SELECT t1.*, t2.* FROM t1 INNER JOIN t2 ...
    
  • Use of an unqualified * with other items in the select list may produce a parse error. To avoid this problem, use a qualified tbl_name.* reference

    SELECT AVG(score), t1.* FROM t1 ...
    

The following list provides additional information about other SELECT clauses:

  • A select_expr can be given an alias using AS alias_name. The alias is used as the expression's column name and can be used in GROUP BY, ORDER BY, or HAVING clauses. For example:

    SELECT CONCAT(last_name,', ',first_name) AS full_name
      FROM mytable ORDER BY full_name;
    

    The AS keyword is optional when aliasing a select_expr with an identifier. The preceding example could have been written like this:

    SELECT CONCAT(last_name,', ',first_name) full_name
      FROM mytable ORDER BY full_name;
    

    However, because the AS is optional, a subtle problem can occur if you forget the comma between two select_expr expressions: MySQL interprets the second as an alias name. For example, in the following statement, columnb is treated as an alias name:

    SELECT columna columnb FROM mytable;
    

    For this reason, it is good practice to be in the habit of using AS explicitly when specifying column aliases.

    It is not permissible to refer to a column alias in a WHERE clause, because the column value might not yet be determined when the WHERE clause is executed. See Section C.5.5.4, “Problems with Column Aliases”.

  • The FROM table_references clause indicates the table or tables from which to retrieve rows. If you name more than one table, you are performing a join. For information on join syntax, see Section 12.2.9.1, “JOIN Syntax”. For each table specified, you can optionally specify an alias.

    tbl_name [[AS] alias] [index_hint]
    

    The use of index hints provides the optimizer with information about how to choose indexes during query processing. For a description of the syntax for specifying these hints, see Section 12.2.9.2, “Index Hint Syntax”.

    You can use SET max_seeks_for_key=value as an alternative way to force MySQL to prefer key scans instead of table scans. See Section 5.1.4, “Server System Variables”.

  • You can refer to a table within the default database as tbl_name, or as db_name.tbl_name to specify a database explicitly. You can refer to a column as col_name, tbl_name.col_name, or db_name.tbl_name.col_name. You need not specify a tbl_name or db_name.tbl_name prefix for a column reference unless the reference would be ambiguous. See Section 8.2.1, “Identifier Qualifiers”, for examples of ambiguity that require the more explicit column reference forms.

  • A table reference can be aliased using tbl_name AS alias_name or tbl_name alias_name:

    SELECT t1.name, t2.salary FROM employee AS t1, info AS t2
      WHERE t1.name = t2.name;
    
    SELECT t1.name, t2.salary FROM employee t1, info t2
      WHERE t1.name = t2.name;
    
  • Columns selected for output can be referred to in ORDER BY and GROUP BY clauses using column names, column aliases, or column positions. Column positions are integers and begin with 1:

    SELECT college, region, seed FROM tournament
      ORDER BY region, seed;
    
    SELECT college, region AS r, seed AS s FROM tournament
      ORDER BY r, s;
    
    SELECT college, region, seed FROM tournament
      ORDER BY 2, 3;
    

    To sort in reverse order, add the DESC (descending) keyword to the name of the column in the ORDER BY clause that you are sorting by. The default is ascending order; this can be specified explicitly using the ASC keyword.

    If ORDER BY occurs within a subquery and also is applied in the outer query, the outermost ORDER BY takes precedence. For example, results for the following statement are sorted in descending order, not ascending order:

    (SELECT ... ORDER BY a) ORDER BY a DESC;
    

    Use of column positions is deprecated because the syntax has been removed from the SQL standard.

  • If you use GROUP BY, output rows are sorted according to the GROUP BY columns as if you had an ORDER BY for the same columns. To avoid the overhead of sorting that GROUP BY produces, add ORDER BY NULL:

    SELECT a, COUNT(b) FROM test_table GROUP BY a ORDER BY NULL;
    
  • MySQL extends the GROUP BY clause so that you can also specify ASC and DESC after columns named in the clause:

    SELECT a, COUNT(b) FROM test_table GROUP BY a DESC;
    
  • MySQL extends the use of GROUP BY to permit selecting fields that are not mentioned in the GROUP BY clause. If you are not getting the results that you expect from your query, please read the description of GROUP BY found in Section 11.16, “Functions and Modifiers for Use with GROUP BY Clauses”.

  • GROUP BY permits a WITH ROLLUP modifier. See Section 11.16.2, “GROUP BY Modifiers”.

  • The HAVING clause is applied nearly last, just before items are sent to the client, with no optimization. (LIMIT is applied after HAVING.)

    The SQL standard requires that HAVING must reference only columns in the GROUP BY clause or columns used in aggregate functions. However, MySQL supports an extension to this behavior, and permits HAVING to refer to columns in the SELECT list and columns in outer subqueries as well.

    If the HAVING clause refers to a column that is ambiguous, a warning occurs. In the following statement, col2 is ambiguous because it is used as both an alias and a column name:

    SELECT COUNT(col1) AS col2 FROM t GROUP BY col2 HAVING col2 = 2;
    

    Preference is given to standard SQL behavior, so if a HAVING column name is used both in GROUP BY and as an aliased column in the output column list, preference is given to the column in the GROUP BY column.

  • Do not use HAVING for items that should be in the WHERE clause. For example, do not write the following:

    SELECT col_name FROM tbl_name HAVING col_name > 0;
    

    Write this instead:

    SELECT col_name FROM tbl_name WHERE col_name > 0;
    
  • The HAVING clause can refer to aggregate functions, which the WHERE clause cannot:

    SELECT user, MAX(salary) FROM users
      GROUP BY user HAVING MAX(salary) > 10;
    

    (This did not work in some older versions of MySQL.)

  • MySQL permits duplicate column names. That is, there can be more than one select_expr with the same name. This is an extension to standard SQL. Because MySQL also permits GROUP BY and HAVING to refer to select_expr values, this can result in an ambiguity:

    SELECT 12 AS a, a FROM t GROUP BY a;
    

    In that statement, both columns have the name a. To ensure that the correct column is used for grouping, use different names for each select_expr.

  • MySQL resolves unqualified column or alias references in ORDER BY clauses by searching in the select_expr values, then in the columns of the tables in the FROM clause. For GROUP BY or HAVING clauses, it searches the FROM clause before searching in the select_expr values. (For GROUP BY and HAVING, this differs from the pre-MySQL 5.0 behavior that used the same rules as for ORDER BY.)

  • The LIMIT clause can be used to constrain the number of rows returned by the SELECT statement. LIMIT takes one or two numeric arguments, which must both be nonnegative integer constants, with these exceptions:

    • Within prepared statements, LIMIT parameters can be specified using ? placeholder markers.

    • Within stored programs, LIMIT parameters can be specified using integer-valued routine parameters or local variables as of MySQL 5.5.6.

    With two arguments, the first argument specifies the offset of the first row to return, and the second specifies the maximum number of rows to return. The offset of the initial row is 0 (not 1):

    SELECT * FROM tbl LIMIT 5,10;  # Retrieve rows 6-15
    

    To retrieve all rows from a certain offset up to the end of the result set, you can use some large number for the second parameter. This statement retrieves all rows from the 96th row to the last:

    SELECT * FROM tbl LIMIT 95,18446744073709551615;
    

    With one argument, the value specifies the number of rows to return from the beginning of the result set:

    SELECT * FROM tbl LIMIT 5;     # Retrieve first 5 rows
    

    In other words, LIMIT row_count is equivalent to LIMIT 0, row_count.

    For prepared statements, you can use placeholders. The following statements will return one row from the tbl table:

    SET @a=1;
    PREPARE STMT FROM 'SELECT * FROM tbl LIMIT ?';
    EXECUTE STMT USING @a;
    

    The following statements will return the second to sixth row from the tbl table:

    SET @skip=1; SET @numrows=5;
    PREPARE STMT FROM 'SELECT * FROM tbl LIMIT ?, ?';
    EXECUTE STMT USING @skip, @numrows;
    

    For compatibility with PostgreSQL, MySQL also supports the LIMIT row_count OFFSET offset syntax.

    If LIMIT occurs within a subquery and also is applied in the outer query, the outermost LIMIT takes precedence. For example, the following statement produces two rows, not one:

    (SELECT ... LIMIT 1) LIMIT 2;
    
  • A PROCEDURE clause names a procedure that should process the data in the result set. For an example, see Section 23.4.1, “PROCEDURE ANALYSE, which describes ANALYSE, a procedure that can be used to obtain suggestions for optimal column data types that may help reduce table sizes.

  • The SELECT ... INTO OUTFILE 'file_name' form of SELECT writes the selected rows to a file. The file is created on the server host, so you must have the FILE privilege to use this syntax. file_name cannot be an existing file, which among other things prevents files such as /etc/passwd and database tables from being destroyed. The character_set_filesystem system variable controls the interpretation of the file name.

    The SELECT ... INTO OUTFILE statement is intended primarily to let you very quickly dump a table to a text file on the server machine. If you want to create the resulting file on some other host than the server host, you normally cannot use SELECT ... INTO OUTFILE since there is no way to write a path to the file relative to the server host's file system.

    However, if the MySQL client software is installed on the remote machine, you can instead use a client command such as mysql -e "SELECT ..." > file_name to generate the file on the client host.

    It is also possible to create the resulting file on a different host other than the server host, if the location of the file on the remote host can be accessed using a network-mapped path on the server's file system. In this case, the presence of mysql (or some other MySQL client program) is not required on the target host.

    SELECT ... INTO OUTFILE is the complement of LOAD DATA INFILE. Column values are written converted to the character set specified in the CHARACTER SET clause. If no such clause is present, values are dumped using the binary character set. In effect, there is no character set conversion. If a table contains columns in several character sets, the output data file will as well and you may not be able to reload the file correctly.

    The syntax for the export_options part of the statement consists of the same FIELDS and LINES clauses that are used with the LOAD DATA INFILE statement. See Section 12.2.6, “LOAD DATA INFILE Syntax”, for information about the FIELDS and LINES clauses, including their default values and permissible values.

    FIELDS ESCAPED BY controls how to write special characters. If the FIELDS ESCAPED BY character is not empty, it is used as a prefix that precedes following characters on output:

    • The FIELDS ESCAPED BY character

    • The FIELDS [OPTIONALLY] ENCLOSED BY character

    • The first character of the FIELDS TERMINATED BY and LINES TERMINATED BY values

    • ASCII NUL (the zero-valued byte; what is actually written following the escape character is ASCII “0”, not a zero-valued byte)

    The FIELDS TERMINATED BY, ENCLOSED BY, ESCAPED BY, or LINES TERMINATED BY characters must be escaped so that you can read the file back in reliably. ASCII NUL is escaped to make it easier to view with some pagers.

    The resulting file does not have to conform to SQL syntax, so nothing else need be escaped.

    If the FIELDS ESCAPED BY character is empty, no characters are escaped and NULL is output as NULL, not \N. It is probably not a good idea to specify an empty escape character, particularly if field values in your data contain any of the characters in the list just given.

    Here is an example that produces a file in the comma-separated values (CSV) format used by many programs:

    SELECT a,b,a+b INTO OUTFILE '/tmp/result.txt'
      FIELDS TERMINATED BY ',' OPTIONALLY ENCLOSED BY '"'
      LINES TERMINATED BY '\n'
      FROM test_table;
    
  • If you use INTO DUMPFILE instead of INTO OUTFILE, MySQL writes only one row into the file, without any column or line termination and without performing any escape processing. This is useful if you want to store a BLOB value in a file.

  • Note

    Any file created by INTO OUTFILE or INTO DUMPFILE is writable by all users on the server host. The reason for this is that the MySQL server cannot create a file that is owned by anyone other than the user under whose account it is running. (You should never run mysqld as root for this and other reasons.) The file thus must be world-writable so that you can manipulate its contents.

    If the secure_file_priv system variable is set to a nonempty directory name, the file to be written must be located in that directory.

  • The INTO clause can name a list of one or more variables, which can be user-defined variables, or parameters or local variables within a stored function or procedure body (see Section 12.7.3.3, “SELECT ... INTO Statement”). The selected values are assigned to the variables. The number of variables must match the number of columns. The query should return a single row. If the query returns no rows, a warning with error code 1329 occurs (No data), and the variable values remain unchanged. If the query returns multiple rows, error 1172 occurs (Result consisted of more than one row). If it is possible that the statement may retrieve multiple rows, you can use LIMIT 1 to limit the result set to a single row.

    In the context of such statements that occur as part of events executed by the Event Scheduler, diagnostics messages (not only errors, but also warnings) are written to the error log, and, on Windows, to the application event log. For additional information, see Section 19.4.5, “Event Scheduler Status”.

  • The SELECT syntax description at the beginning this section shows the INTO clause near the end of the statement. It is also possible to use INTO immediately following the select_expr list.

  • An INTO clause should not be used in a nested SELECT because such a SELECT must return its result to the outer context.

  • If you use FOR UPDATE with a storage engine that uses page or row locks, rows examined by the query are write-locked until the end of the current transaction. Using LOCK IN SHARE MODE sets a shared lock that permits other transactions to read the examined rows but not to update or delete them. See Section 13.6.9.3, “SELECT ... FOR UPDATE and SELECT ... LOCK IN SHARE MODE Locking Reads”.

Following the SELECT keyword, you can use a number of options that affect the operation of the statement. HIGH_PRIORITY, STRAIGHT_JOIN, and options beginning with SQL_ are MySQL extensions to standard SQL.

  • The ALL and DISTINCT options specify whether duplicate rows should be returned. ALL (the default) specifies that all matching rows should be returned, including duplicates. DISTINCT specifies removal of duplicate rows from the result set. It is an error to specify both options. DISTINCTROW is a synonym for DISTINCT.

  • HIGH_PRIORITY gives the SELECT higher priority than a statement that updates a table. You should use this only for queries that are very fast and must be done at once. A SELECT HIGH_PRIORITY query that is issued while the table is locked for reading runs even if there is an update statement waiting for the table to be free. This affects only storage engines that use only table-level locking (such as MyISAM, MEMORY, and MERGE).

    HIGH_PRIORITY cannot be used with SELECT statements that are part of a UNION.

  • STRAIGHT_JOIN forces the optimizer to join the tables in the order in which they are listed in the FROM clause. You can use this to speed up a query if the optimizer joins the tables in nonoptimal order. STRAIGHT_JOIN also can be used in the table_references list. See Section 12.2.9.1, “JOIN Syntax”.

    STRAIGHT_JOIN does not apply to any table that the optimizer treats as a const or system table. Such a table produces a single row, is read during the optimization phase of query execution, and references to its columns are replaced with the appropriate column values before query execution proceeds. These tables will appear first in the query plan displayed by EXPLAIN. See Section 7.8.1, “Optimizing Queries with EXPLAIN. This exception may not apply to const or system tables that are used on the NULL-complemented side of an outer join (that is, the right-side table of a LEFT JOIN or the left-side table of a RIGHT JOIN.

  • SQL_BIG_RESULT or SQL_SMALL_RESULT can be used with GROUP BY or DISTINCT to tell the optimizer that the result set has many rows or is small, respectively. For SQL_BIG_RESULT, MySQL directly uses disk-based temporary tables if needed, and prefers sorting to using a temporary table with a key on the GROUP BY elements. For SQL_SMALL_RESULT, MySQL uses fast temporary tables to store the resulting table instead of using sorting. This should not normally be needed.

  • SQL_BUFFER_RESULT forces the result to be put into a temporary table. This helps MySQL free the table locks early and helps in cases where it takes a long time to send the result set to the client. This option can be used only for top-level SELECT statements, not for subqueries or following UNION.

  • SQL_CALC_FOUND_ROWS tells MySQL to calculate how many rows there would be in the result set, disregarding any LIMIT clause. The number of rows can then be retrieved with SELECT FOUND_ROWS(). See Section 11.14, “Information Functions”.

  • The SQL_CACHE and SQL_NO_CACHE options affect caching of query results in the query cache (see Section 7.9.3, “The MySQL Query Cache”). SQL_CACHE tells MySQL to store the result in the query cache if it is cacheable and the value of the query_cache_type system variable is 2 or DEMAND. SQL_NO_CACHE tells MySQL not to store the result in the query cache.

    For views, SQL_NO_CACHE applies if it appears in any SELECT in the query. For a cacheable query, SQL_CACHE applies if it appears in the first SELECT of a view referred to by the query.

    As of MySQL 5.5.3, these two options are mutually exclusive and an error occurs if they are both specified. Also, these options are not permitted in subqueries (including subqueries in the FROM clause), and SELECT statements in unions other than the first SELECT.

    Before MySQL 5.5.3, for a query that uses UNION or subqueries, the following rules apply:

    • SQL_NO_CACHE applies if it appears in any SELECT in the query.

    • For a cacheable query, SQL_CACHE applies if it appears in the first SELECT of the query.

12.2.9.1. JOIN Syntax

MySQL supports the following JOIN syntaxes for the table_references part of SELECT statements and multiple-table DELETE and UPDATE statements:

table_references:
    table_reference [, table_reference] ...

table_reference:
    table_factor
  | join_table

table_factor:
    tbl_name [[AS] alias] [index_hint_list]
  | table_subquery [AS] alias
  | ( table_references )
  | { OJ table_reference LEFT OUTER JOIN table_reference
        ON conditional_expr }

join_table:
    table_reference [INNER | CROSS] JOIN table_factor [join_condition]
  | table_reference STRAIGHT_JOIN table_factor
  | table_reference STRAIGHT_JOIN table_factor ON conditional_expr
  | table_reference {LEFT|RIGHT} [OUTER] JOIN table_reference join_condition
  | table_reference NATURAL [{LEFT|RIGHT} [OUTER]] JOIN table_factor

join_condition:
    ON conditional_expr
  | USING (column_list)

index_hint_list:
    index_hint [, index_hint] ...

index_hint:
    USE {INDEX|KEY}
      [{FOR {JOIN|ORDER BY|GROUP BY}] ([index_list])
  | IGNORE {INDEX|KEY}
      [{FOR {JOIN|ORDER BY|GROUP BY}] (index_list)
  | FORCE {INDEX|KEY}
      [{FOR {JOIN|ORDER BY|GROUP BY}] (index_list)

index_list:
    index_name [, index_name] ...

A table reference is also known as a join expression.

The syntax of table_factor is extended in comparison with the SQL Standard. The latter accepts only table_reference, not a list of them inside a pair of parentheses.

This is a conservative extension if we consider each comma in a list of table_reference items as equivalent to an inner join. For example:

SELECT * FROM t1 LEFT JOIN (t2, t3, t4)
                 ON (t2.a=t1.a AND t3.b=t1.b AND t4.c=t1.c)

is equivalent to:

SELECT * FROM t1 LEFT JOIN (t2 CROSS JOIN t3 CROSS JOIN t4)
                 ON (t2.a=t1.a AND t3.b=t1.b AND t4.c=t1.c)

In MySQL, CROSS JOIN is a syntactic equivalent to INNER JOIN (they can replace each other). In standard SQL, they are not equivalent. INNER JOIN is used with an ON clause, CROSS JOIN is used otherwise.

In general, parentheses can be ignored in join expressions containing only inner join operations. MySQL also supports nested joins (see Section 7.13.7, “Nested Join Optimization”).

Index hints can be specified to affect how the MySQL optimizer makes use of indexes. For more information, see Section 12.2.9.2, “Index Hint Syntax”.

The following list describes general factors to take into account when writing joins.

  • A table reference can be aliased using tbl_name AS alias_name or tbl_name alias_name:

    SELECT t1.name, t2.salary
      FROM employee AS t1 INNER JOIN info AS t2 ON t1.name = t2.name;
    
    SELECT t1.name, t2.salary
      FROM employee t1 INNER JOIN info t2 ON t1.name = t2.name;
    
  • A table_subquery is also known as a subquery in the FROM clause. Such subqueries must include an alias to give the subquery result a table name. A trivial example follows; see also Section 12.2.10.8, “Subqueries in the FROM Clause”.

    SELECT * FROM (SELECT 1, 2, 3) AS t1;
    
  • INNER JOIN and , (comma) are semantically equivalent in the absence of a join condition: both produce a Cartesian product between the specified tables (that is, each and every row in the first table is joined to each and every row in the second table).

    However, the precedence of the comma operator is less than of INNER JOIN, CROSS JOIN, LEFT JOIN, and so on. If you mix comma joins with the other join types when there is a join condition, an error of the form Unknown column 'col_name' in 'on clause' may occur. Information about dealing with this problem is given later in this section.

  • The conditional_expr used with ON is any conditional expression of the form that can be used in a WHERE clause. Generally, you should use the ON clause for conditions that specify how to join tables, and the WHERE clause to restrict which rows you want in the result set.

  • If there is no matching row for the right table in the ON or USING part in a LEFT JOIN, a row with all columns set to NULL is used for the right table. You can use this fact to find rows in a table that have no counterpart in another table:

    SELECT left_tbl.*
      FROM left_tbl LEFT JOIN right_tbl ON left_tbl.id = right_tbl.id
      WHERE right_tbl.id IS NULL;
    

    This example finds all rows in left_tbl with an id value that is not present in right_tbl (that is, all rows in left_tbl with no corresponding row in right_tbl). This assumes that right_tbl.id is declared NOT NULL. See Section 7.13.5, “LEFT JOIN and RIGHT JOIN Optimization”.

  • The USING(column_list) clause names a list of columns that must exist in both tables. If tables a and b both contain columns c1, c2, and c3, the following join compares corresponding columns from the two tables:

    a LEFT JOIN b USING (c1,c2,c3)
    
  • The NATURAL [LEFT] JOIN of two tables is defined to be semantically equivalent to an INNER JOIN or a LEFT JOIN with a USING clause that names all columns that exist in both tables.

  • RIGHT JOIN works analogously to LEFT JOIN. To keep code portable across databases, it is recommended that you use LEFT JOIN instead of RIGHT JOIN.

  • The { OJ ... LEFT OUTER JOIN ...} syntax shown in the join syntax description exists only for compatibility with ODBC. The curly braces in the syntax should be written literally; they are not metasyntax as used elsewhere in syntax descriptions.

    SELECT left_tbl.*
        FROM { OJ left_tbl LEFT OUTER JOIN right_tbl ON left_tbl.id = right_tbl.id }
        WHERE right_tbl.id IS NULL;
    

    You can use other types of joins within { OJ ... }, such as INNER JOIN or RIGHT OUTER JOIN. This helps with compatibility with some third-party applications, but is not official ODBC syntax.

  • STRAIGHT_JOIN is similar to JOIN, except that the left table is always read before the right table. This can be used for those (few) cases for which the join optimizer puts the tables in the wrong order.

Some join examples:

SELECT * FROM table1, table2;

SELECT * FROM table1 INNER JOIN table2 ON table1.id=table2.id;

SELECT * FROM table1 LEFT JOIN table2 ON table1.id=table2.id;

SELECT * FROM table1 LEFT JOIN table2 USING (id);

SELECT * FROM table1 LEFT JOIN table2 ON table1.id=table2.id
  LEFT JOIN table3 ON table2.id=table3.id;

Join Processing Changes in MySQL 5.0.12

Note

Natural joins and joins with USING, including outer join variants, are processed according to the SQL:2003 standard. The goal was to align the syntax and semantics of MySQL with respect to NATURAL JOIN and JOIN ... USING according to SQL:2003. However, these changes in join processing can result in different output columns for some joins. Also, some queries that appeared to work correctly in older versions (prior to 5.0.12) must be rewritten to comply with the standard.

These changes have five main aspects:

  • The way that MySQL determines the result columns of NATURAL or USING join operations (and thus the result of the entire FROM clause).

  • Expansion of SELECT * and SELECT tbl_name.* into a list of selected columns.

  • Resolution of column names in NATURAL or USING joins.

  • Transformation of NATURAL or USING joins into JOIN ... ON.

  • Resolution of column names in the ON condition of a JOIN ... ON.

The following list provides more detail about several effects of current join processing versus join processing in older versions. The term “previously” means “prior to MySQL 5.0.12.

  • The columns of a NATURAL join or a USING join may be different from previously. Specifically, redundant output columns no longer appear, and the order of columns for SELECT * expansion may be different from before.

    Consider this set of statements:

    CREATE TABLE t1 (i INT, j INT);
    CREATE TABLE t2 (k INT, j INT);
    INSERT INTO t1 VALUES(1,1);
    INSERT INTO t2 VALUES(1,1);
    SELECT * FROM t1 NATURAL JOIN t2;
    SELECT * FROM t1 JOIN t2 USING (j);
    

    Previously, the statements produced this output:

    +------+------+------+------+
    | i    | j    | k    | j    |
    +------+------+------+------+
    |    1 |    1 |    1 |    1 |
    +------+------+------+------+
    +------+------+------+------+
    | i    | j    | k    | j    |
    +------+------+------+------+
    |    1 |    1 |    1 |    1 |
    +------+------+------+------+
    

    In the first SELECT statement, column j appears in both tables and thus becomes a join column, so, according to standard SQL, it should appear only once in the output, not twice. Similarly, in the second SELECT statement, column j is named in the USING clause and should appear only once in the output, not twice. But in both cases, the redundant column is not eliminated. Also, the order of the columns is not correct according to standard SQL.

    Now the statements produce this output:

    +------+------+------+
    | j    | i    | k    |
    +------+------+------+
    |    1 |    1 |    1 |
    +------+------+------+
    +------+------+------+
    | j    | i    | k    |
    +------+------+------+
    |    1 |    1 |    1 |
    +------+------+------+
    

    The redundant column is eliminated and the column order is correct according to standard SQL:

    • First, coalesced common columns of the two joined tables, in the order in which they occur in the first table

    • Second, columns unique to the first table, in order in which they occur in that table

    • Third, columns unique to the second table, in order in which they occur in that table

    The single result column that replaces two common columns is defined using the coalesce operation. That is, for two t1.a and t2.a the resulting single join column a is defined as a = COALESCE(t1.a, t2.a), where:

    COALESCE(x, y) = (CASE WHEN V1 IS NOT NULL THEN V1 ELSE V2 END)
    

    If the join operation is any other join, the result columns of the join consists of the concatenation of all columns of the joined tables. This is the same as previously.

    A consequence of the definition of coalesced columns is that, for outer joins, the coalesced column contains the value of the non-NULL column if one of the two columns is always NULL. If neither or both columns are NULL, both common columns have the same value, so it doesn't matter which one is chosen as the value of the coalesced column. A simple way to interpret this is to consider that a coalesced column of an outer join is represented by the common column of the inner table of a JOIN. Suppose that the tables t1(a,b) and t2(a,c) have the following contents:

    t1    t2
    ----  ----
    1 x   2 z
    2 y   3 w
    

    Then:

    mysql> SELECT * FROM t1 NATURAL LEFT JOIN t2;
    +------+------+------+
    | a    | b    | c    |
    +------+------+------+
    |    1 | x    | NULL |
    |    2 | y    | z    |
    +------+------+------+
    

    Here column a contains the values of t1.a.

    mysql> SELECT * FROM t1 NATURAL RIGHT JOIN t2;
    +------+------+------+
    | a    | c    | b    |
    +------+------+------+
    |    2 | z    | y    |
    |    3 | w    | NULL |
    +------+------+------+
    

    Here column a contains the values of t2.a.

    Compare these results to the otherwise equivalent queries with JOIN ... ON:

    mysql> SELECT * FROM t1 LEFT JOIN t2 ON (t1.a = t2.a);
    +------+------+------+------+
    | a    | b    | a    | c    |
    +------+------+------+------+
    |    1 | x    | NULL | NULL |
    |    2 | y    |    2 | z    |
    +------+------+------+------+
    
    mysql> SELECT * FROM t1 RIGHT JOIN t2 ON (t1.a = t2.a);
    +------+------+------+------+
    | a    | b    | a    | c    |
    +------+------+------+------+
    |    2 | y    |    2 | z    |
    | NULL | NULL |    3 | w    |
    +------+------+------+------+
    
  • Previously, a USING clause could be rewritten as an ON clause that compares corresponding columns. For example, the following two clauses were semantically identical:

    a LEFT JOIN b USING (c1,c2,c3)
    a LEFT JOIN b ON a.c1=b.c1 AND a.c2=b.c2 AND a.c3=b.c3
    

    Now the two clauses no longer are quite the same:

    • With respect to determining which rows satisfy the join condition, both joins remain semantically identical.

    • With respect to determining which columns to display for SELECT * expansion, the two joins are not semantically identical. The USING join selects the coalesced value of corresponding columns, whereas the ON join selects all columns from all tables. For the preceding USING join, SELECT * selects these values:

      COALESCE(a.c1,b.c1), COALESCE(a.c2,b.c2), COALESCE(a.c3,b.c3)
      

      For the ON join, SELECT * selects these values:

      a.c1, a.c2, a.c3, b.c1, b.c2, b.c3
      

      With an inner join, COALESCE(a.c1,b.c1) is the same as either a.c1 or b.c1 because both columns will have the same value. With an outer join (such as LEFT JOIN), one of the two columns can be NULL. That column will be omitted from the result.

  • The evaluation of multi-way natural joins differs in a very important way that affects the result of NATURAL or USING joins and that can require query rewriting. Suppose that you have three tables t1(a,b), t2(c,b), and t3(a,c) that each have one row: t1(1,2), t2(10,2), and t3(7,10). Suppose also that you have this NATURAL JOIN on the three tables:

    SELECT ... FROM t1 NATURAL JOIN t2 NATURAL JOIN t3;
    

    Previously, the left operand of the second join was considered to be t2, whereas it should be the nested join (t1 NATURAL JOIN t2). As a result, the columns of t3 are checked for common columns only in t2, and, if t3 has common columns with t1, these columns are not used as equi-join columns. Thus, previously, the preceding query was transformed to the following equi-join:

    SELECT ... FROM t1, t2, t3
      WHERE t1.b = t2.b AND t2.c = t3.c;
    

    That join is missing one more equi-join predicate (t1.a = t3.a). As a result, it produces one row, not the empty result that it should. The correct equivalent query is this:

    SELECT ... FROM t1, t2, t3
      WHERE t1.b = t2.b AND t2.c = t3.c AND t1.a = t3.a;
    

    If you require the same query result in current versions of MySQL as in older versions, rewrite the natural join as the first equi-join.

  • Previously, the comma operator (,) and JOIN both had the same precedence, so the join expression t1, t2 JOIN t3 was interpreted as ((t1, t2) JOIN t3). Now JOIN has higher precedence, so the expression is interpreted as (t1, (t2 JOIN t3)). This change affects statements that use an ON clause, because that clause can refer only to columns in the operands of the join, and the change in precedence changes interpretation of what those operands are.

    Example:

    CREATE TABLE t1 (i1 INT, j1 INT);
    CREATE TABLE t2 (i2 INT, j2 INT);
    CREATE TABLE t3 (i3 INT, j3 INT);
    INSERT INTO t1 VALUES(1,1);
    INSERT INTO t2 VALUES(1,1);
    INSERT INTO t3 VALUES(1,1);
    SELECT * FROM t1, t2 JOIN t3 ON (t1.i1 = t3.i3);
    

    Previously, the SELECT was legal due to the implicit grouping of t1,t2 as (t1,t2). Now the JOIN takes precedence, so the operands for the ON clause are t2 and t3. Because t1.i1 is not a column in either of the operands, the result is an Unknown column 't1.i1' in 'on clause' error. To allow the join to be processed, group the first two tables explicitly with parentheses so that the operands for the ON clause are (t1,t2) and t3:

    SELECT * FROM (t1, t2) JOIN t3 ON (t1.i1 = t3.i3);
    

    Alternatively, avoid the use of the comma operator and use JOIN instead:

    SELECT * FROM t1 JOIN t2 JOIN t3 ON (t1.i1 = t3.i3);
    

    This change also applies to statements that mix the comma operator with INNER JOIN, CROSS JOIN, LEFT JOIN, and RIGHT JOIN, all of which now have higher precedence than the comma operator.

  • Previously, the ON clause could refer to columns in tables named to its right. Now an ON clause can refer only to its operands.

    Example:

    CREATE TABLE t1 (i1 INT);
    CREATE TABLE t2 (i2 INT);
    CREATE TABLE t3 (i3 INT);
    SELECT * FROM t1 JOIN t2 ON (i1 = i3) JOIN t3;
    

    Previously, the SELECT statement was legal. Now the statement fails with an Unknown column 'i3' in 'on clause' error because i3 is a column in t3, which is not an operand of the ON clause. The statement should be rewritten as follows:

    SELECT * FROM t1 JOIN t2 JOIN t3 ON (i1 = i3);
    
  • Resolution of column names in NATURAL or USING joins is different than previously. For column names that are outside the FROM clause, MySQL now handles a superset of the queries compared to previously. That is, in cases when MySQL formerly issued an error that some column is ambiguous, the query now is handled correctly. This is due to the fact that MySQL now treats the common columns of NATURAL or USING joins as a single column, so when a query refers to such columns, the query compiler does not consider them as ambiguous.

    Example:

    SELECT * FROM t1 NATURAL JOIN t2 WHERE b > 1;
    

    Previously, this query would produce an error ERROR 1052 (23000): Column 'b' in where clause is ambiguous. Now the query produces the correct result:

    +------+------+------+
    | b    | c    | y    |
    +------+------+------+
    |    4 |    2 |    3 |
    +------+------+------+
    

    One extension of MySQL compared to the SQL:2003 standard is that MySQL enables you to qualify the common (coalesced) columns of NATURAL or USING joins (just as previously), while the standard disallows that.

12.2.9.2. Index Hint Syntax

You can provide hints to give the optimizer information about how to choose indexes during query processing. Section 12.2.9.1, “JOIN Syntax”, describes the general syntax for specifying tables in a SELECT statement. The syntax for an individual table, including that for index hints, looks like this:

tbl_name [[AS] alias] [index_hint_list]

index_hint_list:
    index_hint [, index_hint] ...

index_hint:
    USE {INDEX|KEY}
      [{FOR {JOIN|ORDER BY|GROUP BY}] ([index_list])
  | IGNORE {INDEX|KEY}
      [{FOR {JOIN|ORDER BY|GROUP BY}] (index_list)
  | FORCE {INDEX|KEY}
      [{FOR {JOIN|ORDER BY|GROUP BY}] (index_list)

index_list:
    index_name [, index_name] ...

By specifying USE INDEX (index_list), you can tell MySQL to use only one of the named indexes to find rows in the table. The alternative syntax IGNORE INDEX (index_list) can be used to tell MySQL to not use some particular index or indexes. These hints are useful if EXPLAIN shows that MySQL is using the wrong index from the list of possible indexes.

You can also use FORCE INDEX, which acts like USE INDEX (index_list) but with the addition that a table scan is assumed to be very expensive. In other words, a table scan is used only if there is no way to use one of the given indexes to find rows in the table.

Each hint requires the names of indexes, not the names of columns. The name of a PRIMARY KEY is PRIMARY. To see the index names for a table, use SHOW INDEX.

An index_name value need not be a full index name. It can be an unambiguous prefix of an index name. If a prefix is ambiguous, an error occurs.

Examples:

SELECT * FROM table1 USE INDEX (col1_index,col2_index)
  WHERE col1=1 AND col2=2 AND col3=3;

SELECT * FROM table1 IGNORE INDEX (col3_index)
  WHERE col1=1 AND col2=2 AND col3=3;

The syntax for index hints has the following characteristics:

  • It is syntactically valid to specify an empty index_list for USE INDEX, which means “use no indexes.” Specifying an empty index_list for FORCE INDEX or IGNORE INDEX is a syntax error.

  • You can specify the scope of a index hint by adding a FOR clause to the hint. This provides more fine-grained control over the optimizer's selection of an execution plan for various phases of query processing. To affect only the indexes used when MySQL decides how to find rows in the table and how to process joins, use FOR JOIN. To influence index usage for sorting or grouping rows, use FOR ORDER BY or FOR GROUP BY. (However, if there is a covering index for the table and it is used to access the table, the optimizer will ignore IGNORE INDEX FOR {ORDER BY|GROUP BY} hints that disable that index.)

  • You can specify multiple index hints:

    SELECT * FROM t1 USE INDEX (i1) IGNORE INDEX FOR ORDER BY (i2) ORDER BY a;
    

    It is not a error to name the same index in several hints (even within the same hint):

    SELECT * FROM t1 USE INDEX (i1) USE INDEX (i1,i1);
    

    However, it is an error to mix USE INDEX and FORCE INDEX for the same table:

    SELECT * FROM t1 USE INDEX FOR JOIN (i1) FORCE INDEX FOR JOIN (i2);
    

if you specify no FOR clause for an index hint, the hint by default applies to all parts of the statement. For example, this hint:

IGNORE INDEX (i1)

is equivalent to this combination of hints:

IGNORE INDEX FOR JOIN (i1)
IGNORE INDEX FOR ORDER BY (i1)
IGNORE INDEX FOR GROUP BY (i1)

To cause the server to use the older behavior for hint scope when no FOR clause is present (so that hints apply only to row retrieval), enable the old system variable at server startup. Take care about enabling this variable in a replication setup. With statement-based binary logging, having different modes for the master and slaves might lead to replication errors.

When index hints are processed, they are collected in a single list by type (USE, FORCE, IGNORE) and by scope (FOR JOIN, FOR ORDER BY, FOR GROUP BY). For example:

SELECT * FROM t1
  USE INDEX () IGNORE INDEX (i2) USE INDEX (i1) USE INDEX (i2);

is equivalent to:

SELECT * FROM t1
   USE INDEX (i1,i2) IGNORE INDEX (i2);

The index hints then are applied for each scope in the following order:

  1. {USE|FORCE} INDEX is applied if present. (If not, the optimizer-determined set of indexes is used.)

  2. IGNORE INDEX is applied over the result of the previous step. For example, the following two queries are equivalent:

    SELECT * FROM t1 USE INDEX (i1) IGNORE INDEX (i2) USE INDEX (i2);
    
    SELECT * FROM t1 USE INDEX (i1);
    

For FULLTEXT searches, index hints work as follows:

  • For natural language mode searches, index hints are silently ignored. For example, IGNORE INDEX(i) is ignored with no warning and the index is still used.

    For boolean mode searches, index hints with FOR ORDER BY or FOR GROUP BY are silently ignored. Index hints with FOR JOIN or no FOR modifier are honored. In contrast to how hints apply for non-FULLTEXT searches, the hint is used for all phases of query execution (finding rows and retrieval, grouping, and ordering). This is true even if the hint is given for a non-FULLTEXT index.

For example, the following two queries are equivalent:

SELECT * FROM t
  USE INDEX (index1)
  IGNORE INDEX (index1) FOR ORDER BY
  IGNORE INDEX (index1) FOR GROUP BY
  WHERE ... IN BOOLEAN MODE ... ;

SELECT * FROM t
  USE INDEX (index1)
  WHERE ... IN BOOLEAN MODE ... ;

Index hints are accepted but ignored for UPDATE statements.

12.2.9.3. UNION Syntax

SELECT ...
UNION [ALL | DISTINCT] SELECT ...
[UNION [ALL | DISTINCT] SELECT ...]

UNION is used to combine the result from multiple SELECT statements into a single result set.

The column names from the first SELECT statement are used as the column names for the results returned. Selected columns listed in corresponding positions of each SELECT statement should have the same data type. (For example, the first column selected by the first statement should have the same type as the first column selected by the other statements.)

If the data types of corresponding SELECT columns do not match, the types and lengths of the columns in the UNION result take into account the values retrieved by all of the SELECT statements. For example, consider the following:

mysql> SELECT REPEAT('a',1) UNION SELECT REPEAT('b',10);
+---------------+
| REPEAT('a',1) |
+---------------+
| a             |
| bbbbbbbbbb    |
+---------------+

(In some earlier versions of MySQL, only the type and length from the first SELECT would have been used and the second row would have been truncated to a length of 1.)

The SELECT statements are normal select statements, but with the following restrictions:

  • Only the last SELECT statement can use INTO OUTFILE. (However, the entire UNION result is written to the file.)

  • HIGH_PRIORITY cannot be used with SELECT statements that are part of a UNION. If you specify it for the first SELECT, it has no effect. If you specify it for any subsequent SELECT statements, a syntax error results.

The default behavior for UNION is that duplicate rows are removed from the result. The optional DISTINCT keyword has no effect other than the default because it also specifies duplicate-row removal. With the optional ALL keyword, duplicate-row removal does not occur and the result includes all matching rows from all the SELECT statements.

You can mix UNION ALL and UNION DISTINCT in the same query. Mixed UNION types are treated such that a DISTINCT union overrides any ALL union to its left. A DISTINCT union can be produced explicitly by using UNION DISTINCT or implicitly by using UNION with no following DISTINCT or ALL keyword.

To use an ORDER BY or LIMIT clause to sort or limit the entire UNION result, parenthesize the individual SELECT statements and place the ORDER BY or LIMIT after the last one. The following example uses both clauses:

(SELECT a FROM t1 WHERE a=10 AND B=1)
UNION
(SELECT a FROM t2 WHERE a=11 AND B=2)
ORDER BY a LIMIT 10;

This kind of ORDER BY cannot use column references that include a table name (that is, names in tbl_name.col_name format). Instead, provide a column alias in the first SELECT statement and refer to the alias in the ORDER BY. (Alternatively, refer to the column in the ORDER BY using its column position. However, use of column positions is deprecated.)

Also, if a column to be sorted is aliased, the ORDER BY clause must refer to the alias, not the column name. The first of the following statements will work, but the second will fail with an Unknown column 'a' in 'order clause' error:

(SELECT a AS b FROM t) UNION (SELECT ...) ORDER BY b;
(SELECT a AS b FROM t) UNION (SELECT ...) ORDER BY a;

To apply ORDER BY or LIMIT to an individual SELECT, place the clause inside the parentheses that enclose the SELECT:

(SELECT a FROM t1 WHERE a=10 AND B=1 ORDER BY a LIMIT 10)
UNION
(SELECT a FROM t2 WHERE a=11 AND B=2 ORDER BY a LIMIT 10);

However, use of ORDER BY for individual SELECT statements implies nothing about the order in which the rows appear in the final result because UNION by default produces an unordered set of rows. Therefore, the use of ORDER BY in this context is typically in conjunction with LIMIT, so that it is used to determine the subset of the selected rows to retrieve for the SELECT, even though it does not necessarily affect the order of those rows in the final UNION result. If ORDER BY appears without LIMIT in a SELECT, it is optimized away because it will have no effect anyway.

To cause rows in a UNION result to consist of the sets of rows retrieved by each SELECT one after the other, select an additional column in each SELECT to use as a sort column and add an ORDER BY following the last SELECT:

(SELECT 1 AS sort_col, col1a, col1b, ... FROM t1)
UNION
(SELECT 2, col2a, col2b, ... FROM t2) ORDER BY sort_col;

To additionally maintain sort order within individual SELECT results, add a secondary column to the ORDER BY clause:

(SELECT 1 AS sort_col, col1a, col1b, ... FROM t1)
UNION
(SELECT 2, col2a, col2b, ... FROM t2) ORDER BY sort_col, col1a;

Use of an additional column also enables you to determine which SELECT each row comes from. Extra columns can provide other identifying information as well, such as a string that indicates a table name.

12.2.10. Subquery Syntax

A subquery is a SELECT statement within another statement.

Starting with MySQL 4.1, all subquery forms and operations that the SQL standard requires are supported, as well as a few features that are MySQL-specific.

Here is an example of a subquery:

SELECT * FROM t1 WHERE column1 = (SELECT column1 FROM t2);

In this example, SELECT * FROM t1 ... is the outer query (or outer statement), and (SELECT column1 FROM t2) is the subquery. We say that the subquery is nested within the outer query, and in fact it is possible to nest subqueries within other subqueries, to a considerable depth. A subquery must always appear within parentheses.

The main advantages of subqueries are:

  • They allow queries that are structured so that it is possible to isolate each part of a statement.

  • They provide alternative ways to perform operations that would otherwise require complex joins and unions.

  • Many people find subqueries more readable than complex joins or unions. Indeed, it was the innovation of subqueries that gave people the original idea of calling the early SQL “Structured Query Language.

Here is an example statement that shows the major points about subquery syntax as specified by the SQL standard and supported in MySQL:

DELETE FROM t1
WHERE s11 > ANY
 (SELECT COUNT(*) /* no hint */ FROM t2
  WHERE NOT EXISTS
   (SELECT * FROM t3
    WHERE ROW(5*t2.s1,77)=
     (SELECT 50,11*s1 FROM t4 UNION SELECT 50,77 FROM
      (SELECT * FROM t5) AS t5)));

A subquery can return a scalar (a single value), a single row, a single column, or a table (one or more rows of one or more columns). These are called scalar, column, row, and table subqueries. Subqueries that return a particular kind of result often can be used only in certain contexts, as described in the following sections.

There are few restrictions on the type of statements in which subqueries can be used. A subquery can contain many of the keywords or clauses that an ordinary SELECT can contain: DISTINCT, GROUP BY, ORDER BY, LIMIT, joins, index hints, UNION constructs, comments, functions, and so on.

One restriction is that a subquery's outer statement must be one of: SELECT, INSERT, UPDATE, DELETE, SET, or DO. Another restriction is that currently you cannot modify a table and select from the same table in a subquery. This applies to statements such as DELETE, INSERT, REPLACE, UPDATE, and (because subqueries can be used in the SET clause) LOAD DATA INFILE.

A more comprehensive discussion of restrictions on subquery use, including performance issues for certain forms of subquery syntax, is given in Section E.4, “Restrictions on Subqueries”.

12.2.10.1. The Subquery as Scalar Operand

In its simplest form, a subquery is a scalar subquery that returns a single value. A scalar subquery is a simple operand, and you can use it almost anywhere a single column value or literal is legal, and you can expect it to have those characteristics that all operands have: a data type, a length, an indication that it can be NULL, and so on. For example:

CREATE TABLE t1 (s1 INT, s2 CHAR(5) NOT NULL);
INSERT INTO t1 VALUES(100, 'abcde');
SELECT (SELECT s2 FROM t1);

The subquery in this SELECT returns a single value ('abcde') that has a data type of CHAR, a length of 5, a character set and collation equal to the defaults in effect at CREATE TABLE time, and an indication that the value in the column can be NULL. Nullability of the value selected by a scalar subquery is not copied because if the subquery result is empty, the result is NULL. For the subquery just shown, if t1 were empty, the result would be NULL even though s2 is NOT NULL.

There are a few contexts in which a scalar subquery cannot be used. If a statement permits only a literal value, you cannot use a subquery. For example, LIMIT requires literal integer arguments, and LOAD DATA INFILE requires a literal string file name. You cannot use subqueries to supply these values.

When you see examples in the following sections that contain the rather spartan construct (SELECT column1 FROM t1), imagine that your own code contains much more diverse and complex constructions.

Suppose that we make two tables:

CREATE TABLE t1 (s1 INT);
INSERT INTO t1 VALUES (1);
CREATE TABLE t2 (s1 INT);
INSERT INTO t2 VALUES (2);

Then perform a SELECT:

SELECT (SELECT s1 FROM t2) FROM t1;

The result is 2 because there is a row in t2 containing a column s1 that has a value of 2.

A scalar subquery can be part of an expression, but remember the parentheses, even if the subquery is an operand that provides an argument for a function. For example:

SELECT UPPER((SELECT s1 FROM t1)) FROM t2;

12.2.10.2. Comparisons Using Subqueries

The most common use of a subquery is in the form:

non_subquery_operand comparison_operator (subquery)

Where comparison_operator is one of these operators:

=  >  <  >=  <=  <>  !=  <=>

For example:

... WHERE 'a' = (SELECT column1 FROM t1)

MySQL also permits this construct:

non_subquery_operand LIKE (subquery)

At one time the only legal place for a subquery was on the right side of a comparison, and you might still find some old DBMSs that insist on this.

Here is an example of a common-form subquery comparison that you cannot do with a join. It finds all the rows in table t1 for which the column1 value is equal to a maximum value in table t2:

SELECT * FROM t1
  WHERE column1 = (SELECT MAX(column2) FROM t2);

Here is another example, which again is impossible with a join because it involves aggregating for one of the tables. It finds all rows in table t1 containing a value that occurs twice in a given column:

SELECT * FROM t1 AS t
  WHERE 2 = (SELECT COUNT(*) FROM t1 WHERE t1.id = t.id);

For a comparison of the subquery to a scalar, the subquery must return a scalar. For a comparison of the subquery to a row constructor, the subquery must be a row subquery that returns a row with the same number of values as the row constructor. See Section 12.2.10.5, “Row Subqueries”.

12.2.10.3. Subqueries with ANY, IN, or SOME

Syntax:

operand comparison_operator ANY (subquery)
operand IN (subquery)
operand comparison_operator SOME (subquery)

Where comparison_operator is one of these operators:

=  >  <  >=  <=  <>  !=

The ANY keyword, which must follow a comparison operator, means “return TRUE if the comparison is TRUE for ANY of the values in the column that the subquery returns.” For example:

SELECT s1 FROM t1 WHERE s1 > ANY (SELECT s1 FROM t2);

Suppose that there is a row in table t1 containing (10). The expression is TRUE if table t2 contains (21,14,7) because there is a value 7 in t2 that is less than 10. The expression is FALSE if table t2 contains (20,10), or if table t2 is empty. The expression is unknown (that is, NULL) if table t2 contains (NULL,NULL,NULL).

When used with a subquery, the word IN is an alias for = ANY. Thus, these two statements are the same:

SELECT s1 FROM t1 WHERE s1 = ANY (SELECT s1 FROM t2);
SELECT s1 FROM t1 WHERE s1 IN    (SELECT s1 FROM t2);

IN and = ANY are not synonyms when used with an expression list. IN can take an expression list, but = ANY cannot. See Section 11.3.2, “Comparison Functions and Operators”.

NOT IN is not an alias for <> ANY, but for <> ALL. See Section 12.2.10.4, “Subqueries with ALL.

The word SOME is an alias for ANY. Thus, these two statements are the same:

SELECT s1 FROM t1 WHERE s1 <> ANY  (SELECT s1 FROM t2);
SELECT s1 FROM t1 WHERE s1 <> SOME (SELECT s1 FROM t2);

Use of the word SOME is rare, but this example shows why it might be useful. To most people, the English phrase “a is not equal to any b” means “there is no b which is equal to a,” but that is not what is meant by the SQL syntax. The syntax means “there is some b to which a is not equal.” Using <> SOME instead helps ensure that everyone understands the true meaning of the query.

12.2.10.4. Subqueries with ALL

Syntax:

operand comparison_operator ALL (subquery)

The word ALL, which must follow a comparison operator, means “return TRUE if the comparison is TRUE for ALL of the values in the column that the subquery returns.” For example:

SELECT s1 FROM t1 WHERE s1 > ALL (SELECT s1 FROM t2);

Suppose that there is a row in table t1 containing (10). The expression is TRUE if table t2 contains (-5,0,+5) because 10 is greater than all three values in t2. The expression is FALSE if table t2 contains (12,6,NULL,-100) because there is a single value 12 in table t2 that is greater than 10. The expression is unknown (that is, NULL) if table t2 contains (0,NULL,1).

Finally, the expression is TRUE if table t2 is empty. So, the following expression is TRUE when table t2 is empty:

SELECT * FROM t1 WHERE 1 > ALL (SELECT s1 FROM t2);

But this expression is NULL when table t2 is empty:

SELECT * FROM t1 WHERE 1 > (SELECT s1 FROM t2);

In addition, the following expression is NULL when table t2 is empty:

SELECT * FROM t1 WHERE 1 > ALL (SELECT MAX(s1) FROM t2);

In general, tables containing NULL values and empty tables are “edge cases.” When writing subqueries, always consider whether you have taken those two possibilities into account.

NOT IN is an alias for <> ALL. Thus, these two statements are the same:

SELECT s1 FROM t1 WHERE s1 <> ALL (SELECT s1 FROM t2);
SELECT s1 FROM t1 WHERE s1 NOT IN (SELECT s1 FROM t2);

12.2.10.5. Row Subqueries

The discussion to this point has been of scalar or column subqueries; that is, subqueries that return a single value or a column of values. A row subquery is a subquery variant that returns a single row and can thus return more than one column value. Legal operators for row subquery comparisons are:

=  >  <  >=  <=  <>  !=  <=>

Here are two examples:

SELECT * FROM t1
  WHERE (col1,col2) = (SELECT col3, col4 FROM t2 WHERE id = 10);
SELECT * FROM t1
  WHERE ROW(col1,col2) = (SELECT col3, col4 FROM t2 WHERE id = 10);

For both queries, if the table t2 contains a single row with id = 10, the subquery returns a single row. If this row has col3 and col4 values equal to the col1 and col2 values of any rows in t1, the WHERE expression is TRUE and each query returns those t1 rows. If the t2 row col3 and col4 values are not equal the col1 and col2 values of any t1 row, the expression is FALSE and the query returns an empty result set. The expression is unknown (that is, NULL) if the subquery produces no rows. An error occurs if the subquery produces multiple rows because a row subquery can return at most one row.

The expressions (1,2) and ROW(1,2) are sometimes called row constructors. The two are equivalent. The row constructor and the row returned by the subquery must contain the same number of values.

A row constructor is used for comparisons with subqueries that return two or more columns. When a subquery returns a single column, this is regarded as a scalar value and not as a row, so a row constructor cannot be used with a subquery that does not return at least two columns. Thus, the following query fails with a syntax error:

SELECT * FROM t1 WHERE ROW(1) = (SELECT column1 FROM t2)

Row constructors are legal in other contexts. For example, the following two statements are semantically equivalent (and are handled in the same way by the optimizer):

SELECT * FROM t1 WHERE (column1,column2) = (1,1);
SELECT * FROM t1 WHERE column1 = 1 AND column2 = 1;

The following query answers the request, “find all rows in table t1 that also exist in table t2”:

SELECT column1,column2,column3
  FROM t1
  WHERE (column1,column2,column3) IN
         (SELECT column1,column2,column3 FROM t2);

12.2.10.6. Subqueries with EXISTS or NOT EXISTS

If a subquery returns any rows at all, EXISTS subquery is TRUE, and NOT EXISTS subquery is FALSE. For example:

SELECT column1 FROM t1 WHERE EXISTS (SELECT * FROM t2);

Traditionally, an EXISTS subquery starts with SELECT *, but it could begin with SELECT 5 or SELECT column1 or anything at all. MySQL ignores the SELECT list in such a subquery, so it makes no difference.

For the preceding example, if t2 contains any rows, even rows with nothing but NULL values, the EXISTS condition is TRUE. This is actually an unlikely example because a [NOT] EXISTS subquery almost always contains correlations. Here are some more realistic examples:

  • What kind of store is present in one or more cities?

    SELECT DISTINCT store_type FROM stores
      WHERE EXISTS (SELECT * FROM cities_stores
                    WHERE cities_stores.store_type = stores.store_type);
    
  • What kind of store is present in no cities?

    SELECT DISTINCT store_type FROM stores
      WHERE NOT EXISTS (SELECT * FROM cities_stores
                        WHERE cities_stores.store_type = stores.store_type);
    
  • What kind of store is present in all cities?

    SELECT DISTINCT store_type FROM stores s1
      WHERE NOT EXISTS (
        SELECT * FROM cities WHERE NOT EXISTS (
          SELECT * FROM cities_stores
           WHERE cities_stores.city = cities.city
           AND cities_stores.store_type = stores.store_type));
    

The last example is a double-nested NOT EXISTS query. That is, it has a NOT EXISTS clause within a NOT EXISTS clause. Formally, it answers the question “does a city exist with a store that is not in Stores”? But it is easier to say that a nested NOT EXISTS answers the question “is x TRUE for all y?

12.2.10.7. Correlated Subqueries

A correlated subquery is a subquery that contains a reference to a table that also appears in the outer query. For example:

SELECT * FROM t1
  WHERE column1 = ANY (SELECT column1 FROM t2
                       WHERE t2.column2 = t1.column2);

Notice that the subquery contains a reference to a column of t1, even though the subquery's FROM clause does not mention a table t1. So, MySQL looks outside the subquery, and finds t1 in the outer query.

Suppose that table t1 contains a row where column1 = 5 and column2 = 6; meanwhile, table t2 contains a row where column1 = 5 and column2 = 7. The simple expression ... WHERE column1 = ANY (SELECT column1 FROM t2) would be TRUE, but in this example, the WHERE clause within the subquery is FALSE (because (5,6) is not equal to (5,7)), so the expression as a whole is FALSE.

Scoping rule: MySQL evaluates from inside to outside. For example:

SELECT column1 FROM t1 AS x
  WHERE x.column1 = (SELECT column1 FROM t2 AS x
    WHERE x.column1 = (SELECT column1 FROM t3
      WHERE x.column2 = t3.column1));

In this statement, x.column2 must be a column in table t2 because SELECT column1 FROM t2 AS x ... renames t2. It is not a column in table t1 because SELECT column1 FROM t1 ... is an outer query that is farther out.

For subqueries in HAVING or ORDER BY clauses, MySQL also looks for column names in the outer select list.

For certain cases, a correlated subquery is optimized. For example:

val IN (SELECT key_val FROM tbl_name WHERE correlated_condition)

Otherwise, they are inefficient and likely to be slow. Rewriting the query as a join might improve performance.

Aggregate functions in correlated subqueries may contain outer references, provided the function contains nothing but outer references, and provided the function is not contained in another function or expression.

12.2.10.8. Subqueries in the FROM Clause

Subqueries are legal in a SELECT statement's FROM clause. The actual syntax is:

SELECT ... FROM (subquery) [AS] name ...

The [AS] name clause is mandatory, because every table in a FROM clause must have a name. Any columns in the subquery select list must have unique names.

For the sake of illustration, assume that you have this table:

CREATE TABLE t1 (s1 INT, s2 CHAR(5), s3 FLOAT);

Here is how to use a subquery in the FROM clause, using the example table:

INSERT INTO t1 VALUES (1,'1',1.0);
INSERT INTO t1 VALUES (2,'2',2.0);
SELECT sb1,sb2,sb3
  FROM (SELECT s1 AS sb1, s2 AS sb2, s3*2 AS sb3 FROM t1) AS sb
  WHERE sb1 > 1;

Result: 2, '2', 4.0.

Here is another example: Suppose that you want to know the average of a set of sums for a grouped table. This does not work:

SELECT AVG(SUM(column1)) FROM t1 GROUP BY column1;

However, this query provides the desired information:

SELECT AVG(sum_column1)
  FROM (SELECT SUM(column1) AS sum_column1
        FROM t1 GROUP BY column1) AS t1;

Notice that the column name used within the subquery (sum_column1) is recognized in the outer query.

Subqueries in the FROM clause can return a scalar, column, row, or table. Subqueries in the FROM clause cannot be correlated subqueries, unless used within the ON clause of a JOIN operation.

Subqueries in the FROM clause are executed even for the EXPLAIN statement (that is, derived temporary tables are built). This occurs because upper-level queries need information about all tables during the optimization phase, and the table represented by a subquery in the FROM clause is unavailable unless the subquery is executed.

It is possible under certain circumstances to modify table data using EXPLAIN SELECT. This can occur if the outer query accesses any tables and an inner query invokes a stored function that changes one or more rows of a table. Suppose that there are two tables t1 and t2 in database d1, created as shown here:

mysql> CREATE DATABASE d1;
Query OK, 1 row affected (0.00 sec)

mysql> USE d1;
Database changed

mysql> CREATE TABLE t1 (c1 INT);
Query OK, 0 rows affected (0.15 sec)

mysql> CREATE TABLE t2 (c1 INT);
Query OK, 0 rows affected (0.08 sec)

Now we create a stored function f1 which modifies t2:

mysql> DELIMITER //
mysql> CREATE FUNCTION f1(p1 INT) RETURNS INT
mysql>   BEGIN
mysql>     INSERT INTO t2 VALUES (p1);
mysql>     RETURN p1;
mysql>   END //
Query OK, 0 rows affected (0.01 sec)

mysql> DELIMITER ;

Referencing the function directly in an EXPLAIN SELECT does not have any effect on t2, as shown here:

mysql> SELECT * FROM t2;
Empty set (0.00 sec)

mysql> EXPLAIN SELECT f1(5);
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows | Extra          |
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
|  1 | SIMPLE      | NULL  | NULL | NULL          | NULL | NULL    | NULL | NULL | No tables used |
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
1 row in set (0.00 sec)

mysql> SELECT * FROM t2;
Empty set (0.00 sec)

This is because the SELECT statement did not reference any tables, as can be seen in the table and Extra columns of the output. This is also true of the following nested SELECT:

mysql> EXPLAIN SELECT NOW() AS a1, (SELECT f1(5)) AS a2;
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
| id | select_type | table | type | possible_keys | key  | key_len | ref  | rows | Extra          |
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
|  1 | PRIMARY     | NULL  | NULL | NULL          | NULL | NULL    | NULL | NULL | No tables used |
+----+-------------+-------+------+---------------+------+---------+------+------+----------------+
1 row in set, 1 warning (0.00 sec)

mysql> SHOW WARNINGS;
+-------+------+------------------------------------------+
| Level | Code | Message                                  |
+-------+------+------------------------------------------+
| Note  | 1249 | Select 2 was reduced during optimization |
+-------+------+------------------------------------------+
1 row in set (0.00 sec)

mysql> SELECT * FROM t2;
Empty set (0.00 sec)

However, if the outer SELECT references any tables, the optimizer executes the statement in the subquery as well:

mysql> EXPLAIN SELECT * FROM t1 AS a1, (SELECT f1(5)) AS a2;
+----+-------------+------------+--------+---------------+------+---------+------+------+---------------------+
| id | select_type | table      | type   | possible_keys | key  | key_len | ref  | rows | Extra               |
+----+-------------+------------+--------+---------------+------+---------+------+------+---------------------+
|  1 | PRIMARY     | a1         | system | NULL          | NULL | NULL    | NULL |    0 | const row not found |
|  1 | PRIMARY     | <derived2> | system | NULL          | NULL | NULL    | NULL |    1 |                     |
|  2 | DERIVED     | NULL       | NULL   | NULL          | NULL | NULL    | NULL | NULL | No tables used      |
+----+-------------+------------+--------+---------------+------+---------+------+------+---------------------+
3 rows in set (0.00 sec)

mysql> SELECT * FROM t2;
+------+
| c1   |
+------+
|    5 |
+------+
1 row in set (0.00 sec)

This also means that an EXPLAIN SELECT statement such as the one shown here may take a long time to execute because the BENCHMARK() function is executed once for each row in t1:

EXPLAIN SELECT * FROM t1 AS a1, (SELECT BENCHMARK(1000000, MD5(NOW())));

12.2.10.9. Subquery Errors

There are some errors that apply only to subqueries. This section describes them.

  • Unsupported subquery syntax:

    ERROR 1235 (ER_NOT_SUPPORTED_YET)
    SQLSTATE = 42000
    Message = "This version of MySQL doesn't yet support
    'LIMIT & IN/ALL/ANY/SOME subquery'"
    

    This means that MySQL does not support statements of the following form:

    SELECT * FROM t1 WHERE s1 IN (SELECT s2 FROM t2 ORDER BY s1 LIMIT 1)
    
  • Incorrect number of columns from subquery:

    ERROR 1241 (ER_OPERAND_COL)
    SQLSTATE = 21000
    Message = "Operand should contain 1 column(s)"
    

    This error occurs in cases like this:

    SELECT (SELECT column1, column2 FROM t2) FROM t1;
    

    You may use a subquery that returns multiple columns, if the purpose is row comparison. In other contexts, the subquery must be a scalar operand. See Section 12.2.10.5, “Row Subqueries”.

  • Incorrect number of rows from subquery:

    ERROR 1242 (ER_SUBSELECT_NO_1_ROW)
    SQLSTATE = 21000
    Message = "Subquery returns more than 1 row"
    

    This error occurs for statements where the subquery must return at most one row but returns multiple rows. Consider the following example:

    SELECT * FROM t1 WHERE column1 = (SELECT column1 FROM t2);
    

    If SELECT column1 FROM t2 returns just one row, the previous query will work. If the subquery returns more than one row, error 1242 will occur. In that case, the query should be rewritten as:

    SELECT * FROM t1 WHERE column1 = ANY (SELECT column1 FROM t2);
    
  • Incorrectly used table in subquery:

    Error 1093 (ER_UPDATE_TABLE_USED)
    SQLSTATE = HY000
    Message = "You can't specify target table 'x'
    for update in FROM clause"
    

    This error occurs in cases such as the following, which attempts to modify a table and select from the same table in the subquery:

    UPDATE t1 SET column2 = (SELECT MAX(column1) FROM t1);
    

    You can use a subquery for assignment within an UPDATE statement because subqueries are legal in UPDATE and DELETE statements as well as in SELECT statements. However, you cannot use the same table (in this case, table t1) for both the subquery FROM clause and the update target.

For transactional storage engines, the failure of a subquery causes the entire statement to fail. For nontransactional storage engines, data modifications made before the error was encountered are preserved.

12.2.10.10. Optimizing Subqueries

Development is ongoing, so no optimization tip is reliable for the long term. The following list provides some interesting tricks that you might want to play with:

  • Use subquery clauses that affect the number or order of the rows in the subquery. For example:

    SELECT * FROM t1 WHERE t1.column1 IN
      (SELECT column1 FROM t2 ORDER BY column1);
    SELECT * FROM t1 WHERE t1.column1 IN
      (SELECT DISTINCT column1 FROM t2);
    SELECT * FROM t1 WHERE EXISTS
      (SELECT * FROM t2 LIMIT 1);
    
  • Replace a join with a subquery. For example, try this:

    SELECT DISTINCT column1 FROM t1 WHERE t1.column1 IN (
      SELECT column1 FROM t2);
    

    Instead of this:

    SELECT DISTINCT t1.column1 FROM t1, t2
      WHERE t1.column1 = t2.column1;
    
  • Some subqueries can be transformed to joins for compatibility with older versions of MySQL that do not support subqueries. However, in some cases, converting a subquery to a join may improve performance. See Section 12.2.10.11, “Rewriting Subqueries as Joins”.

  • Move clauses from outside to inside the subquery. For example, use this query:

    SELECT * FROM t1
      WHERE s1 IN (SELECT s1 FROM t1 UNION ALL SELECT s1 FROM t2);
    

    Instead of this query:

    SELECT * FROM t1
      WHERE s1 IN (SELECT s1 FROM t1) OR s1 IN (SELECT s1 FROM t2);
    

    For another example, use this query:

    SELECT (SELECT column1 + 5 FROM t1) FROM t2;
    

    Instead of this query:

    SELECT (SELECT column1 FROM t1) + 5 FROM t2;
    
  • Use a row subquery instead of a correlated subquery. For example, use this query:

    SELECT * FROM t1
      WHERE (column1,column2) IN (SELECT column1,column2 FROM t2);
    

    Instead of this query:

    SELECT * FROM t1
      WHERE EXISTS (SELECT * FROM t2 WHERE t2.column1=t1.column1
                    AND t2.column2=t1.column2);
    
  • Use NOT (a = ANY (...)) rather than a <> ALL (...).

  • Use x = ANY (table containing (1,2)) rather than x=1 OR x=2.

  • Use = ANY rather than EXISTS.

  • For uncorrelated subqueries that always return one row, IN is always slower than =. For example, use this query:

    SELECT * FROM t1
      WHERE t1.col_name = (SELECT a FROM t2 WHERE b = some_const);
    

    Instead of this query:

    SELECT * FROM t1
      WHERE t1.col_name IN (SELECT a FROM t2 WHERE b = some_const);
    

These tricks might cause programs to go faster or slower. Using MySQL facilities like the BENCHMARK() function, you can get an idea about what helps in your own situation. See Section 11.14, “Information Functions”.

Some optimizations that MySQL itself makes are:

  • MySQL executes uncorrelated subqueries only once. Use EXPLAIN to make sure that a given subquery really is uncorrelated.

  • MySQL rewrites IN, ALL, ANY, and SOME subqueries in an attempt to take advantage of the possibility that the select-list columns in the subquery are indexed.

  • MySQL replaces subqueries of the following form with an index-lookup function, which EXPLAIN describes as a special join type (unique_subquery or index_subquery):

    ... IN (SELECT indexed_column FROM single_table ...)
    
  • MySQL enhances expressions of the following form with an expression involving MIN() or MAX(), unless NULL values or empty sets are involved:

    value {ALL|ANY|SOME} {> | < | >= | <=} (uncorrelated subquery)
    

    For example, this WHERE clause:

    WHERE 5 > ALL (SELECT x FROM t)
    

    might be treated by the optimizer like this:

    WHERE 5 > (SELECT MAX(x) FROM t)
    

See also the MySQL Internals Manual chapter How MySQL Transforms Subqueries.

12.2.10.11. Rewriting Subqueries as Joins

Sometimes there are other ways to test membership in a set of values than by using a subquery. Also, on some occasions, it is not only possible to rewrite a query without a subquery, but it can be more efficient to make use of some of these techniques rather than to use subqueries. One of these is the IN() construct:

For example, this query:

SELECT * FROM t1 WHERE id IN (SELECT id FROM t2);

Can be rewritten as:

SELECT DISTINCT t1.* FROM t1, t2 WHERE t1.id=t2.id;

The queries:

SELECT * FROM t1 WHERE id NOT IN (SELECT id FROM t2);
SELECT * FROM t1 WHERE NOT EXISTS (SELECT id FROM t2 WHERE t1.id=t2.id);

Can be rewritten as:

SELECT table1.*
  FROM table1 LEFT JOIN table2 ON table1.id=table2.id
  WHERE table2.id IS NULL;

A LEFT [OUTER] JOIN can be faster than an equivalent subquery because the server might be able to optimize it better—a fact that is not specific to MySQL Server alone. Prior to SQL-92, outer joins did not exist, so subqueries were the only way to do certain things. Today, MySQL Server and many other modern database systems offer a wide range of outer join types.

MySQL Server supports multiple-table DELETE statements that can be used to efficiently delete rows based on information from one table or even from many tables at the same time. Multiple-table UPDATE statements are also supported. See Section 12.2.2, “DELETE Syntax”, and Section 12.2.11, “UPDATE Syntax”.

12.2.11. UPDATE Syntax

Single-table syntax:

UPDATE [LOW_PRIORITY] [IGNORE] table_reference
    SET col_name1={expr1|DEFAULT} [, col_name2={expr2|DEFAULT}] ...
    [WHERE where_condition]
    [ORDER BY ...]
    [LIMIT row_count]

Multiple-table syntax:

UPDATE [LOW_PRIORITY] [IGNORE] table_references
    SET col_name1={expr1|DEFAULT} [, col_name2={expr2|DEFAULT}] ...
    [WHERE where_condition]

For the single-table syntax, the UPDATE statement updates columns of existing rows in the named table with new values. The SET clause indicates which columns to modify and the values they should be given. Each value can be given as an expression, or the keyword DEFAULT to set a column explicitly to its default value. The WHERE clause, if given, specifies the conditions that identify which rows to update. With no WHERE clause, all rows are updated. If the ORDER BY clause is specified, the rows are updated in the order that is specified. The LIMIT clause places a limit on the number of rows that can be updated.

For the multiple-table syntax, UPDATE updates rows in each table named in table_references that satisfy the conditions. In this case, ORDER BY and LIMIT cannot be used.

where_condition is an expression that evaluates to true for each row to be updated. For expression syntax, see Section 8.5, “Expression Syntax”.

table_references and where_condition are is specified as described in Section 12.2.9, “SELECT Syntax”.

You need the UPDATE privilege only for columns referenced in an UPDATE that are actually updated. You need only the SELECT privilege for any columns that are read but not modified.

The UPDATE statement supports the following modifiers:

  • With the LOW_PRIORITY keyword, execution of the UPDATE is delayed until no other clients are reading from the table. This affects only storage engines that use only table-level locking (such as MyISAM, MEMORY, and MERGE).

  • With the IGNORE keyword, the update statement does not abort even if errors occur during the update. Rows for which duplicate-key conflicts occur are not updated. Rows for which columns are updated to values that would cause data conversion errors are updated to the closest valid values instead.

If you access a column from the table to be updated in an expression, UPDATE uses the current value of the column. For example, the following statement sets col1 to one more than its current value:

UPDATE t1 SET col1 = col1 + 1;

The second assignment in the following statement sets col2 to the current (updated) col1 value, not the original col1 value. The result is that col1 and col2 have the same value. This behavior differs from standard SQL.

UPDATE t1 SET col1 = col1 + 1, col2 = col1;

Single-table UPDATE assignments are generally evaluated from left to right. For multiple-table updates, there is no guarantee that assignments are carried out in any particular order.

If you set a column to the value it currently has, MySQL notices this and does not update it.

If you update a column that has been declared NOT NULL by setting to NULL, an error occurs if strict SQL mode is enabled; otherwise, the column is set to the implicit default value for the column data type and the warning count is incremented. The implicit default value is 0 for numeric types, the empty string ('') for string types, and the “zero” value for date and time types. See Section 10.1.4, “Data Type Default Values”.

UPDATE returns the number of rows that were actually changed. The mysql_info() C API function returns the number of rows that were matched and updated and the number of warnings that occurred during the UPDATE.

You can use LIMIT row_count to restrict the scope of the UPDATE. A LIMIT clause is a rows-matched restriction. The statement stops as soon as it has found row_count rows that satisfy the WHERE clause, whether or not they actually were changed.

If an UPDATE statement includes an ORDER BY clause, the rows are updated in the order specified by the clause. This can be useful in certain situations that might otherwise result in an error. Suppose that a table t contains a column id that has a unique index. The following statement could fail with a duplicate-key error, depending on the order in which rows are updated:

UPDATE t SET id = id + 1;

For example, if the table contains 1 and 2 in the id column and 1 is updated to 2 before 2 is updated to 3, an error occurs. To avoid this problem, add an ORDER BY clause to cause the rows with larger id values to be updated before those with smaller values:

UPDATE t SET id = id + 1 ORDER BY id DESC;

You can also perform UPDATE operations covering multiple tables. However, you cannot use ORDER BY or LIMIT with a multiple-table UPDATE. The table_references clause lists the tables involved in the join. Its syntax is described in Section 12.2.9.1, “JOIN Syntax”. Here is an example:

UPDATE items,month SET items.price=month.price
WHERE items.id=month.id;

The preceding example shows an inner join that uses the comma operator, but multiple-table UPDATE statements can use any type of join permitted in SELECT statements, such as LEFT JOIN.

If you use a multiple-table UPDATE statement involving InnoDB tables for which there are foreign key constraints, the MySQL optimizer might process tables in an order that differs from that of their parent/child relationship. In this case, the statement fails and rolls back. Instead, update a single table and rely on the ON UPDATE capabilities that InnoDB provides to cause the other tables to be modified accordingly. See Section 13.6.5.4, “FOREIGN KEY Constraints”.

Currently, you cannot update a table and select from the same table in a subquery.

Index hints (see Section 12.2.9.2, “Index Hint Syntax”) are accepted but ignored for UPDATE statements.

12.3. MySQL Transactional and Locking Statements

MySQL supports local transactions (within a given client session) through statements such as SET autocommit, START TRANSACTION, COMMIT, and ROLLBACK. See Section 12.3.1, “START TRANSACTION, COMMIT, and ROLLBACK Syntax”. XA transaction support enables MySQL to participate in distributed transactions as well. See Section 12.3.7, “XA Transactions”.

12.3.1. START TRANSACTION, COMMIT, and ROLLBACK Syntax

START TRANSACTION [WITH CONSISTENT SNAPSHOT] | BEGIN [WORK]
COMMIT [WORK] [AND [NO] CHAIN | [NO] RELEASE]
ROLLBACK [WORK] [AND [NO] CHAIN | [NO] RELEASE]
SET autocommit = {0 | 1}

The START TRANSACTION or BEGIN statement begins a new transaction. COMMIT commits the current transaction, making its changes permanent. ROLLBACK rolls back the current transaction, canceling its changes. The SET autocommit statement disables or enables the default autocommit mode for the current session.

The optional WORK keyword is supported for COMMIT and ROLLBACK, as are the CHAIN and RELEASE clauses. CHAIN and RELEASE can be used for additional control over transaction completion. The value of the completion_type system variable determines the default completion behavior. See Section 5.1.4, “Server System Variables”.

Note

Within all stored programs (stored procedures and functions, triggers, and events), the parser treats BEGIN [WORK] as the beginning of a BEGIN ... END block. Begin a transaction in this context with START TRANSACTION instead.

The AND CHAIN clause causes a new transaction to begin as soon as the current one ends, and the new transaction has the same isolation level as the just-terminated transaction. The RELEASE clause causes the server to disconnect the current client session after terminating the current transaction. Including the NO keyword suppresses CHAIN or RELEASE completion, which can be useful if the completion_type system variable is set to cause chaining or release completion by default.

By default, MySQL runs with autocommit mode enabled. This means that as soon as you execute a statement that updates (modifies) a table, MySQL stores the update on disk to make it permanent. To disable autocommit mode, use the following statement:

SET autocommit=0;

After disabling autocommit mode by setting the autocommit variable to zero, changes to transaction-safe tables (such as those for InnoDB or NDBCLUSTER) are not made permanent immediately. You must use COMMIT to store your changes to disk or ROLLBACK to ignore the changes.

autocommit is a session variable and must be set for each session. For information about how to do this for each new connection, see the description of the init_connect system variable.

To disable autocommit mode for a single series of statements, use the START TRANSACTION statement:

START TRANSACTION;
SELECT @A:=SUM(salary) FROM table1 WHERE type=1;
UPDATE table2 SET summary=@A WHERE type=1;
COMMIT;

With START TRANSACTION, autocommit remains disabled until you end the transaction with COMMIT or ROLLBACK. The autocommit mode then reverts to its previous state.

BEGIN and BEGIN WORK are supported as aliases of START TRANSACTION for initiating a transaction. START TRANSACTION is standard SQL syntax and is the recommended way to start an ad-hoc transaction.

Important

Many APIs used for writing MySQL client applications (such as JDBC) provide their own methods for starting transactions that can (and sometimes should) be used instead of sending a START TRANSACTION statement from the client. See Chapter 22, Connectors and APIs, or the documentation for your API, for more information.

The BEGIN statement differs from the use of the BEGIN keyword that starts a BEGIN ... END compound statement. The latter does not begin a transaction. See Section 12.7.1, “BEGIN ... END Compound Statement Syntax”.

You can also begin a transaction like this:

START TRANSACTION WITH CONSISTENT SNAPSHOT;

The WITH CONSISTENT SNAPSHOT clause starts a consistent read for storage engines that are capable of it. This applies only to InnoDB. The effect is the same as issuing a START TRANSACTION followed by a SELECT from any InnoDB table. See Section 13.6.9.2, “Consistent Nonlocking Reads”. The WITH CONSISTENT SNAPSHOT clause does not change the current transaction isolation level, so it provides a consistent snapshot only if the current isolation level is one that permits consistent read (REPEATABLE READ or SERIALIZABLE).

Beginning a transaction causes any pending transaction to be committed. See Section 12.3.3, “Statements That Cause an Implicit Commit”, for more information.

Beginning a transaction also causes table locks acquired with LOCK TABLES to be released, as though you had executed UNLOCK TABLES. Beginning a transaction does not release a global read lock acquired with FLUSH TABLES WITH READ LOCK.

For best results, transactions should be performed using only tables managed by a single transaction-safe storage engine. Otherwise, the following problems can occur:

  • If you use tables from more than one transaction-safe storage engine (such as InnoDB), and the transaction isolation level is not SERIALIZABLE, it is possible that when one transaction commits, another ongoing transaction that uses the same tables will see only some of the changes made by the first transaction. That is, the atomicity of transactions is not guaranteed with mixed engines and inconsistencies can result. (If mixed-engine transactions are infrequent, you can use SET TRANSACTION ISOLATION LEVEL to set the isolation level to SERIALIZABLE on a per-transaction basis as necessary.)

  • If you use tables that are not transaction-safe within a transaction, changes to those tables are stored at once, regardless of the status of autocommit mode.

  • If you issue a ROLLBACK statement after updating a nontransactional table within a transaction, an ER_WARNING_NOT_COMPLETE_ROLLBACK warning occurs. Changes to transaction-safe tables are rolled back, but not changes to nontransaction-safe tables.

Each transaction is stored in the binary log in one chunk, upon COMMIT. Transactions that are rolled back are not logged. (Exception: Modifications to nontransactional tables cannot be rolled back. If a transaction that is rolled back includes modifications to nontransactional tables, the entire transaction is logged with a ROLLBACK statement at the end to ensure that modifications to the nontransactional tables are replicated.) See Section 5.2.4, “The Binary Log”.

You can change the isolation level for transactions with SET TRANSACTION ISOLATION LEVEL. See Section 12.3.6, “SET TRANSACTION Syntax”.

Rolling back can be a slow operation that may occur implicitly without the user having explicitly asked for it (for example, when an error occurs). Because of this, SHOW PROCESSLIST displays Rolling back in the State column for the session, not only for explicit rollbacks performed with the ROLLBACK statement but also for implicit rollbacks.

Note

In MySQL 5.5, BEGIN, COMMIT, and ROLLBACK are not affected by --replicate-do-db or --replicate-ignore-db rules.

12.3.2. Statements That Cannot Be Rolled Back

Some statements cannot be rolled back. In general, these include data definition language (DDL) statements, such as those that create or drop databases, those that create, drop, or alter tables or stored routines.

You should design your transactions not to include such statements. If you issue a statement early in a transaction that cannot be rolled back, and then another statement later fails, the full effect of the transaction cannot be rolled back in such cases by issuing a ROLLBACK statement.

12.3.3. Statements That Cause an Implicit Commit

The statements listed in this section (and any synonyms for them) implicitly end a transaction, as if you had done a COMMIT before executing the statement. As of MySQL 5.5.3, most of these statements also cause an implicit commit after executing; for additional details, see the end of this section.

As of MySQL 5.5.3, most statements that previously caused an implicit commit before executing also do so after executing. The intent is to handle each such statement in its own special transaction because it cannot be rolled back anyway. The following list provides additional details pertaining to this change:

12.3.4. SAVEPOINT and ROLLBACK TO SAVEPOINT Syntax

SAVEPOINT identifier
ROLLBACK [WORK] TO [SAVEPOINT] identifier
RELEASE SAVEPOINT identifier

InnoDB supports the SQL statements SAVEPOINT, ROLLBACK TO SAVEPOINT, RELEASE SAVEPOINT and the optional WORK keyword for ROLLBACK.

The SAVEPOINT statement sets a named transaction savepoint with a name of identifier. If the current transaction has a savepoint with the same name, the old savepoint is deleted and a new one is set.

The ROLLBACK TO SAVEPOINT statement rolls back a transaction to the named savepoint without terminating the transaction. Modifications that the current transaction made to rows after the savepoint was set are undone in the rollback, but InnoDB does not release the row locks that were stored in memory after the savepoint. (For a new inserted row, the lock information is carried by the transaction ID stored in the row; the lock is not separately stored in memory. In this case, the row lock is released in the undo.) Savepoints that were set at a later time than the named savepoint are deleted.

If the ROLLBACK TO SAVEPOINT statement returns the following error, it means that no savepoint with the specified name exists:

ERROR 1305 (42000): SAVEPOINT identifier does not exist

The RELEASE SAVEPOINT statement removes the named savepoint from the set of savepoints of the current transaction. No commit or rollback occurs. It is an error if the savepoint does not exist.

All savepoints of the current transaction are deleted if you execute a COMMIT, or a ROLLBACK that does not name a savepoint.

A new savepoint level is created when a stored function is invoked or a trigger is activated. The savepoints on previous levels become unavailable and thus do not conflict with savepoints on the new level. When the function or trigger terminates, any savepoints it created are released and the previous savepoint level is restored.

12.3.5. LOCK TABLES and UNLOCK TABLES Syntax

LOCK TABLES
    tbl_name [[AS] alias] lock_type
    [, tbl_name [[AS] alias] lock_type] ...

lock_type:
    READ [LOCAL]
  | [LOW_PRIORITY] WRITE

UNLOCK TABLES

MySQL enables client sessions to acquire table locks explicitly for the purpose of cooperating with other sessions for access to tables, or to prevent other sessions from modifying tables during periods when a session requires exclusive access to them. A session can acquire or release locks only for itself. One session cannot acquire locks for another session or release locks held by another session.

Locks may be used to emulate transactions or to get more speed when updating tables. This is explained in more detail later in this section.

LOCK TABLES explicitly acquires table locks for the current client session. Table locks can be acquired for base tables or views. You must have the LOCK TABLES privilege, and the SELECT privilege for each object to be locked.

For view locking, LOCK TABLES adds all base tables used in the view to the set of tables to be locked and locks them automatically. If you lock a table explicitly with LOCK TABLES, any tables used in triggers are also locked implicitly, as described in Section 12.3.5.2, “LOCK TABLES and Triggers”.

UNLOCK TABLES explicitly releases any table locks held by the current session.

Another use for UNLOCK TABLES is to release the global read lock acquired with the FLUSH TABLES WITH READ LOCK statement, which enables you to lock all tables in all databases. See Section 12.4.6.3, “FLUSH Syntax”. (This is a very convenient way to get backups if you have a file system such as Veritas that can take snapshots in time.)

A table lock protects only against inappropriate reads or writes by other sessions. The session holding the lock, even a read lock, can perform table-level operations such as DROP TABLE. Truncate operations are not transaction-safe, so an error occurs if the session attempts one during an active transaction or while holding a table lock.

The following discussion applies only to non-TEMPORARY tables. LOCK TABLES is permitted (but ignored) for a TEMPORARY table. The table can be accessed freely by the session within which it was created, regardless of what other locking may be in effect. No lock is necessary because no other session can see the table.

LOCK TABLES and UNLOCK TABLES cannot be used within stored programs.

Tables in the performance_schema database cannot be locked with LOCK TABLES, except the SETUP_xxx tables.

Rules for Lock Acquisition

To acquire table locks within the current session, use the LOCK TABLES statement. The following lock types are available:

READ [LOCAL] lock:

  • The session that holds the lock can read the table (but not write it).

  • Multiple sessions can acquire a READ lock for the table at the same time.

  • Other sessions can read the table without explicitly acquiring a READ lock.

  • The LOCAL modifier enables nonconflicting INSERT statements (concurrent inserts) by other sessions to execute while the lock is held. (See Section 7.10.3, “Concurrent Inserts”.) However, READ LOCAL cannot be used if you are going to manipulate the database using processes external to the server while you hold the lock. For InnoDB tables, READ LOCAL is the same as READ.

[LOW_PRIORITY] WRITE lock:

  • The session that holds the lock can read and write the table.

  • Only the session that holds the lock can access the table. No other session can access it until the lock is released.

  • Lock requests for the table by other sessions block while the WRITE lock is held.

  • The LOW_PRIORITY modifier affects lock scheduling if the WRITE lock request must wait, as described later.

If the LOCK TABLES statement must wait due to locks held by other sessions on any of the tables, it blocks until all locks can be acquired.

A session that requires locks must acquire all the locks that it needs in a single LOCK TABLES statement. While the locks thus obtained are held, the session can access only the locked tables. For example, in the following sequence of statements, an error occurs for the attempt to access t2 because it was not locked in the LOCK TABLES statement:

mysql> LOCK TABLES t1 READ;
mysql> SELECT COUNT(*) FROM t1;
+----------+
| COUNT(*) |
+----------+
|        3 |
+----------+
mysql> SELECT COUNT(*) FROM t2;
ERROR 1100 (HY000): Table 't2' was not locked with LOCK TABLES

Tables in the INFORMATION_SCHEMA database are an exception. They can be accessed without being locked explicitly even while a session holds table locks obtained with LOCK TABLES.

You cannot refer to a locked table multiple times in a single query using the same name. Use aliases instead, and obtain a separate lock for the table and each alias:

mysql> LOCK TABLE t WRITE, t AS t1 READ;
mysql> INSERT INTO t SELECT * FROM t;
ERROR 1100: Table 't' was not locked with LOCK TABLES
mysql> INSERT INTO t SELECT * FROM t AS t1;

The error occurs for the first INSERT because there are two references to the same name for a locked table. The second INSERT succeeds because the references to the table use different names.

If your statements refer to a table by means of an alias, you must lock the table using that same alias. It does not work to lock the table without specifying the alias:

mysql> LOCK TABLE t READ;
mysql> SELECT * FROM t AS myalias;
ERROR 1100: Table 'myalias' was not locked with LOCK TABLES

Conversely, if you lock a table using an alias, you must refer to it in your statements using that alias:

mysql> LOCK TABLE t AS myalias READ;
mysql> SELECT * FROM t;
ERROR 1100: Table 't' was not locked with LOCK TABLES
mysql> SELECT * FROM t AS myalias;

WRITE locks normally have higher priority than READ locks to ensure that updates are processed as soon as possible. This means that if one session obtains a READ lock and then another session requests a WRITE lock, subsequent READ lock requests wait until the session that requested the WRITE lock has obtained the lock and released it. A request for a LOW_PRIORITY WRITE lock, by contrast, permits subsequent READ lock requests by other sessions to be satisfied first if they occur while the LOW_PRIORITY WRITE request is waiting. You should use LOW_PRIORITY WRITE locks only if you are sure that eventually there will be a time when no sessions have a READ lock. For InnoDB tables in transactional mode (autocommit = 0), a waiting LOW_PRIORITY WRITE lock acts like a regular WRITE lock and causes subsequent READ lock requests to wait.

LOCK TABLES acquires locks as follows:

  1. Sort all tables to be locked in an internally defined order. From the user standpoint, this order is undefined.

  2. If a table is to be locked with a read and a write lock, put the write lock request before the read lock request.

  3. Lock one table at a time until the session gets all locks.

This policy ensures that table locking is deadlock free. There are, however, other things you need to be aware of about this policy: If you are using a LOW_PRIORITY WRITE lock for a table, it means only that MySQL waits for this particular lock until there are no other sessions that want a READ lock. When the session has gotten the WRITE lock and is waiting to get the lock for the next table in the lock table list, all other sessions wait for the WRITE lock to be released. If this becomes a serious problem with your application, you should consider converting some of your tables to transaction-safe tables.

Rules for Lock Release

When the table locks held by a session are released, they are all released at the same time. A session can release its locks explicitly, or locks may be released implicitly under certain conditions.

If the connection for a client session terminates, whether normally or abnormally, the server implicitly releases all table locks held by the session (transactional and nontransactional). If the client reconnects, the locks will no longer be in effect. In addition, if the client had an active transaction, the server rolls back the transaction upon disconnect, and if reconnect occurs, the new session begins with autocommit enabled. For this reason, clients may wish to disable auto-reconnect. With auto-reconnect in effect, the client is not notified if reconnect occurs but any table locks or current transaction will have been lost. With auto-reconnect disabled, if the connection drops, an error occurs for the next statement issued. The client can detect the error and take appropriate action such as reacquiring the locks or redoing the transaction. See Section 22.9.12, “Controlling Automatic Reconnection Behavior”.

Note

If you use ALTER TABLE on a locked table, it may become unlocked. For example, if you attempt a second ALTER TABLE operation, the result may be an error Table 'tbl_name' was not locked with LOCK TABLES. To handle this, lock the table again prior to the second alteration. See also Section C.5.7.1, “Problems with ALTER TABLE.

12.3.5.1. Interaction of Table Locking and Transactions

LOCK TABLES and UNLOCK TABLES interact with the use of transactions as follows:

  • LOCK TABLES is not transaction-safe and implicitly commits any active transaction before attempting to lock the tables.

  • UNLOCK TABLES implicitly commits any active transaction, but only if LOCK TABLES has been used to acquire table locks. For example, in the following set of statements, UNLOCK TABLES releases the global read lock but does not commit the transaction because no table locks are in effect:

    FLUSH TABLES WITH READ LOCK;
    START TRANSACTION;
    SELECT ... ;
    UNLOCK TABLES;
    
  • Beginning a transaction (for example, with START TRANSACTION) implicitly commits any current transaction and releases existing locks.

  • Other statements that implicitly cause transactions to be committed do not release existing locks. For a list of such statements, see Section 12.3.3, “Statements That Cause an Implicit Commit”.

  • The correct way to use LOCK TABLES and UNLOCK TABLES with transactional tables, such as InnoDB tables, is to begin a transaction with SET autocommit = 0 (not START TRANSACTION) followed by LOCK TABLES, and to not call UNLOCK TABLES until you commit the transaction explicitly. For example, if you need to write to table t1 and read from table t2, you can do this:

    SET autocommit=0;
    LOCK TABLES t1 WRITE, t2 READ, ...;
    ... do something with tables t1 and t2 here ...
    COMMIT;
    UNLOCK TABLES;
    

    When you call LOCK TABLES, InnoDB internally takes its own table lock, and MySQL takes its own table lock. InnoDB releases its internal table lock at the next commit, but for MySQL to release its table lock, you have to call UNLOCK TABLES. You should not have autocommit = 1, because then InnoDB releases its internal table lock immediately after the call of LOCK TABLES, and deadlocks can very easily happen. InnoDB does not acquire the internal table lock at all if autocommit = 1, to help old applications avoid unnecessary deadlocks.

  • ROLLBACK does not release table locks.

  • FLUSH TABLES WITH READ LOCK acquires a global read lock and not table locks, so it is not subject to the same behavior as LOCK TABLES and UNLOCK TABLES with respect to table locking and implicit commits. See Section 12.4.6.3, “FLUSH Syntax”.

12.3.5.2. LOCK TABLES and Triggers

If you lock a table explicitly with LOCK TABLES, any tables used in triggers are also locked implicitly:

  • The locks are taken as the same time as those acquired explicitly with the LOCK TABLES statement.

  • The lock on a table used in a trigger depends on whether the table is used only for reading. If so, a read lock suffices. Otherwise, a write lock is used.

  • If a table is locked explicitly for reading with LOCK TABLES, but needs to be locked for writing because it might be modified within a trigger, a write lock is taken rather than a read lock. (That is, an implicit write lock needed due to the table's appearance within a trigger causes an explicit read lock request for the table to be converted to a write lock request.)

Suppose that you lock two tables, t1 and t2, using this statement:

LOCK TABLES t1 WRITE, t2 READ;

If t1 or t2 have any triggers, tables used within the triggers will also be locked. Suppose that t1 has a trigger defined like this:

CREATE TRIGGER t1_a_ins AFTER INSERT ON t1 FOR EACH ROW
BEGIN
  UPDATE t4 SET count = count+1
      WHERE id = NEW.id AND EXISTS (SELECT a FROM t3);
  INSERT INTO t2 VALUES(1, 2);
END;

The result of the LOCK TABLES statement is that t1 and t2 are locked because they appear in the statement, and t3 and t4 are locked because they are used within the trigger:

  • t1 is locked for writing per the WRITE lock request.

  • t2 is locked for writing, even though the request is for a READ lock. This occurs because t2 is inserted into within the trigger, so the READ request is converted to a WRITE request.

  • t3 is locked for reading because it is only read from within the trigger.

  • t4 is locked for writing because it might be updated within the trigger.

12.3.5.3. Other Table-Locking Notes

You can safely use KILL to terminate a session that is waiting for a table lock. See Section 12.4.6.4, “KILL Syntax”.

You should not lock any tables that you are using with INSERT DELAYED. An INSERT DELAYED in this case results in an error because the insert must be handled by a separate thread, not by the session which holds the lock.

As of MySQL 5.5.3, you cannot execute CREATE TABLE or CREATE TABLE ... LIKE while a LOCK TABLES statement is in effect.

As of MySQL 5.5.8, DDL statements on events are prohibited while a LOCK TABLES statement is in effect.

For some operations, system tables in the mysql database must be accessed. For example, the HELP statement requires the contents of the server-side help tables, and CONVERT_TZ() might need to read the time zone tables. The server implicitly locks the system tables for reading as necessary so that you need not lock them explicitly. These tables are treated as just described:

mysql.help_category
mysql.help_keyword
mysql.help_relation
mysql.help_topic
mysql.proc
mysql.time_zone
mysql.time_zone_leap_second
mysql.time_zone_name
mysql.time_zone_transition
mysql.time_zone_transition_type

If you want to explicitly place a WRITE lock on any of those tables with a LOCK TABLES statement, the table must be the only one locked; no other table can be locked with the same statement.

Normally, you do not need to lock tables, because all single UPDATE statements are atomic; no other session can interfere with any other currently executing SQL statement. However, there are a few cases when locking tables may provide an advantage:

  • If you are going to run many operations on a set of MyISAM tables, it is much faster to lock the tables you are going to use. Locking MyISAM tables speeds up inserting, updating, or deleting on them because MySQL does not flush the key cache for the locked tables until UNLOCK TABLES is called. Normally, the key cache is flushed after each SQL statement.

    The downside to locking the tables is that no session can update a READ-locked table (including the one holding the lock) and no session can access a WRITE-locked table other than the one holding the lock.

  • If you are using tables for a nontransactional storage engine, you must use LOCK TABLES if you want to ensure that no other session modifies the tables between a SELECT and an UPDATE. The example shown here requires LOCK TABLES to execute safely:

    LOCK TABLES trans READ, customer WRITE;
    SELECT SUM(value) FROM trans WHERE customer_id=some_id;
    UPDATE customer
      SET total_value=sum_from_previous_statement
      WHERE customer_id=some_id;
    UNLOCK TABLES;
    

    Without LOCK TABLES, it is possible that another session might insert a new row in the trans table between execution of the SELECT and UPDATE statements.

You can avoid using LOCK TABLES in many cases by using relative updates (UPDATE customer SET value=value+new_value) or the LAST_INSERT_ID() function. See Section 1.8.5.3, “Transactions and Atomic Operations”.

You can also avoid locking tables in some cases by using the user-level advisory lock functions GET_LOCK() and RELEASE_LOCK(). These locks are saved in a hash table in the server and implemented with pthread_mutex_lock() and pthread_mutex_unlock() for high speed. See Section 11.15, “Miscellaneous Functions”.

See Section 7.10.1, “Internal Locking Methods”, for more information on locking policy.

12.3.6. SET TRANSACTION Syntax

SET [GLOBAL | SESSION] TRANSACTION ISOLATION LEVEL
  {
       READ UNCOMMITTED
     | READ COMMITTED
     | REPEATABLE READ
     | SERIALIZABLE
   }

This statement sets the transaction isolation level globally, for the current session, or for the next transaction:

  • With the GLOBAL keyword, the statement sets the default transaction level globally for all subsequent sessions. Existing sessions are unaffected.

  • With the SESSION keyword, the statement sets the default transaction level for all subsequent transactions performed within the current session.

  • Without any SESSION or GLOBAL keyword, the statement sets the isolation level for the next (not started) transaction performed within the current session.

A change to the global default isolation level requires the SUPER privilege. Any session is free to change its session isolation level (even in the middle of a transaction), or the isolation level for its next transaction.

SET TRANSACTION ISOLATION LEVEL without GLOBAL or SESSION is not permitted while there is an active transaction:

mysql> START TRANSACTION;
Query OK, 0 rows affected (0.02 sec)

mysql> SET TRANSACTION ISOLATION LEVEL SERIALIZABLE;
ERROR 1568 (25001): Transaction isolation level can't be changed
while a transaction is in progress

To set the global default isolation level at server startup, use the --transaction-isolation=level option to mysqld on the command line or in an option file. Values of level for this option use dashes rather than spaces, so the permissible values are READ-UNCOMMITTED, READ-COMMITTED, REPEATABLE-READ, or SERIALIZABLE. For example, to set the default isolation level to REPEATABLE READ, use these lines in the [mysqld] section of an option file:

[mysqld]
transaction-isolation = REPEATABLE-READ

To determine the global and session transaction isolation levels at runtime, check the value of the tx_isolation system variable:

SELECT @@GLOBAL.tx_isolation, @@tx_isolation;

InnoDB supports each of the transaction isolation levels described here using different locking strategies. The default level is REPEATABLE READ. For additional information about InnoDB record-level locks and how it uses them to execute various types of statements, see Section 13.6.9.4, “InnoDB Record, Gap, and Next-Key Locks”, and Section 13.6.9.6, “Locks Set by Different SQL Statements in InnoDB.

The following list describes how MySQL supports the different transaction levels:

  • READ UNCOMMITTED

    SELECT statements are performed in a nonlocking fashion, but a possible earlier version of a row might be used. Thus, using this isolation level, such reads are not consistent. This is also called a “dirty read.” Otherwise, this isolation level works like READ COMMITTED.

  • READ COMMITTED

    A somewhat Oracle-like isolation level with respect to consistent (nonlocking) reads: Each consistent read, even within the same transaction, sets and reads its own fresh snapshot. See Section 13.6.9.2, “Consistent Nonlocking Reads”.

    For locking reads (SELECT with FOR UPDATE or LOCK IN SHARE MODE), InnoDB locks only index records, not the gaps before them, and thus permits the free insertion of new records next to locked records. For UPDATE and DELETE statements, locking depends on whether the statement uses a unique index with a unique search condition (such as WHERE id = 100), or a range-type search condition (such as WHERE id > 100). For a unique index with a unique search condition, InnoDB locks only the index record found, not the gap before it. For range-type searches, InnoDB locks the index range scanned, using gap locks or next-key (gap plus index-record) locks to block insertions by other sessions into the gaps covered by the range. This is necessary because “phantom rows” must be blocked for MySQL replication and recovery to work.

    Note

    In MySQL 5.5, if the READ COMMITTED isolation level is used or the innodb_locks_unsafe_for_binlog system variable is enabled, there is no InnoDB gap locking except for foreign-key constraint checking and duplicate-key checking. Also, record locks for nonmatching rows are released after MySQL has evaluated the WHERE condition.

    If you use READ COMMITTED or enable innodb_locks_unsafe_for_binlog, you must use row-based binary logging.

  • REPEATABLE READ

    This is the default isolation level for InnoDB. For consistent reads, there is an important difference from the READ COMMITTED isolation level: All consistent reads within the same transaction read the snapshot established by the first read. This convention means that if you issue several plain (nonlocking) SELECT statements within the same transaction, these SELECT statements are consistent also with respect to each other. See Section 13.6.9.2, “Consistent Nonlocking Reads”.

    For locking reads (SELECT with FOR UPDATE or LOCK IN SHARE MODE), UPDATE, and DELETE statements, locking depends on whether the statement uses a unique index with a unique search condition, or a range-type search condition. For a unique index with a unique search condition, InnoDB locks only the index record found, not the gap before it. For other search conditions, InnoDB locks the index range scanned, using gap locks or next-key (gap plus index-record) locks to block insertions by other sessions into the gaps covered by the range.

  • SERIALIZABLE

    This level is like REPEATABLE READ, but InnoDB implicitly converts all plain SELECT statements to SELECT ... LOCK IN SHARE MODE if autocommit is disabled. If autocommit is enabled, the SELECT is its own transaction. It therefore is known to be read only and can be serialized if performed as a consistent (nonlocking) read and need not block for other transactions. (This means that to force a plain SELECT to block if other transactions have modified the selected rows, you should disable autocommit.)

12.3.7. XA Transactions

Support for XA transactions is available for the InnoDB storage engine. The MySQL XA implementation is based on the X/Open CAE document Distributed Transaction Processing: The XA Specification. This document is published by The Open Group and available at http://www.opengroup.org/public/pubs/catalog/c193.htm. Limitations of the current XA implementation are described in Section E.6, “Restrictions on XA Transactions”.

On the client side, there are no special requirements. The XA interface to a MySQL server consists of SQL statements that begin with the XA keyword. MySQL client programs must be able to send SQL statements and to understand the semantics of the XA statement interface. They do not need be linked against a recent client library. Older client libraries also will work.

Currently, among the MySQL Connectors, MySQL Connector/J 5.0.0 supports XA directly (by means of a class interface that handles the Xan SQL statement interface for you).

XA supports distributed transactions; that is, the ability to permit multiple separate transactional resources to participate in a global transaction. Transactional resources often are RDBMSs but may be other kinds of resources.

A global transaction involves several actions that are transactional in themselves, but that all must either complete successfully as a group, or all be rolled back as a group. In essence, this extends ACID properties “up a level” so that multiple ACID transactions can be executed in concert as components of a global operation that also has ACID properties. (However, for a distributed transaction, you must use the SERIALIZABLE isolation level to achieve ACID properties. It is enough to use REPEATABLE READ for a nondistributed transaction, but not for a distributed transaction.)

Some examples of distributed transactions:

  • An application may act as an integration tool that combines a messaging service with an RDBMS. The application makes sure that transactions dealing with message sending, retrieval, and processing that also involve a transactional database all happen in a global transaction. You can think of this as “transactional email.

  • An application performs actions that involve different database servers, such as a MySQL server and an Oracle server (or multiple MySQL servers), where actions that involve multiple servers must happen as part of a global transaction, rather than as separate transactions local to each server.

  • A bank keeps account information in an RDBMS and distributes and receives money through automated teller machines (ATMs). It is necessary to ensure that ATM actions are correctly reflected in the accounts, but this cannot be done with the RDBMS alone. A global transaction manager integrates the ATM and database resources to ensure overall consistency of financial transactions.

Applications that use global transactions involve one or more Resource Managers and a Transaction Manager:

  • A Resource Manager (RM) provides access to transactional resources. A database server is one kind of resource manager. It must be possible to either commit or roll back transactions managed by the RM.

  • A Transaction Manager (TM) coordinates the transactions that are part of a global transaction. It communicates with the RMs that handle each of these transactions. The individual transactions within a global transaction are “branches” of the global transaction. Global transactions and their branches are identified by a naming scheme described later.

The MySQL implementation of XA MySQL enables a MySQL server to act as a Resource Manager that handles XA transactions within a global transaction. A client program that connects to the MySQL server acts as the Transaction Manager.

To carry out a global transaction, it is necessary to know which components are involved, and bring each component to a point when it can be committed or rolled back. Depending on what each component reports about its ability to succeed, they must all commit or roll back as an atomic group. That is, either all components must commit, or all components musts roll back. To manage a global transaction, it is necessary to take into account that any component or the connecting network might fail.

The process for executing a global transaction uses two-phase commit (2PC). This takes place after the actions performed by the branches of the global transaction have been executed.

  1. In the first phase, all branches are prepared. That is, they are told by the TM to get ready to commit. Typically, this means each RM that manages a branch records the actions for the branch in stable storage. The branches indicate whether they are able to do this, and these results are used for the second phase.

  2. In the second phase, the TM tells the RMs whether to commit or roll back. If all branches indicated when they were prepared that they will be able to commit, all branches are told to commit. If any branch indicated when it was prepared that it will not be able to commit, all branches are told to roll back.

In some cases, a global transaction might use one-phase commit (1PC). For example, when a Transaction Manager finds that a global transaction consists of only one transactional resource (that is, a single branch), that resource can be told to prepare and commit at the same time.

12.3.7.1. XA Transaction SQL Syntax

To perform XA transactions in MySQL, use the following statements:

XA {START|BEGIN} xid [JOIN|RESUME]

XA END xid [SUSPEND [FOR MIGRATE]]

XA PREPARE xid

XA COMMIT xid [ONE PHASE]

XA ROLLBACK xid

XA RECOVER

For XA START, the JOIN and RESUME clauses are not supported.

For XA END the SUSPEND [FOR MIGRATE] clause is not supported.

Each XA statement begins with the XA keyword, and most of them require an xid value. An xid is an XA transaction identifier. It indicates which transaction the statement applies to. xid values are supplied by the client, or generated by the MySQL server. An xid value has from one to three parts:

xid: gtrid [, bqual [, formatID ]]

gtrid is a global transaction identifier, bqual is a branch qualifier, and formatID is a number that identifies the format used by the gtrid and bqual values. As indicated by the syntax, bqual and formatID are optional. The default bqual value is '' if not given. The default formatID value is 1 if not given.

gtrid and bqual must be string literals, each up to 64 bytes (not characters) long. gtrid and bqual can be specified in several ways. You can use a quoted string ('ab'), hex string (0x6162, X'ab'), or bit value (b'nnnn').

formatID is an unsigned integer.

The gtrid and bqual values are interpreted in bytes by the MySQL server's underlying XA support routines. However, while an SQL statement containing an XA statement is being parsed, the server works with some specific character set. To be safe, write gtrid and bqual as hex strings.

xid values typically are generated by the Transaction Manager. Values generated by one TM must be different from values generated by other TMs. A given TM must be able to recognize its own xid values in a list of values returned by the XA RECOVER statement.

XA START xid starts an XA transaction with the given xid value. Each XA transaction must have a unique xid value, so the value must not currently be used by another XA transaction. Uniqueness is assessed using the gtrid and bqual values. All following XA statements for the XA transaction must be specified using the same xid value as that given in the XA START statement. If you use any of those statements but specify an xid value that does not correspond to some existing XA transaction, an error occurs.

One or more XA transactions can be part of the same global transaction. All XA transactions within a given global transaction must use the same gtrid value in the xid value. For this reason, gtrid values must be globally unique so that there is no ambiguity about which global transaction a given XA transaction is part of. The bqual part of the xid value must be different for each XA transaction within a global transaction. (The requirement that bqual values be different is a limitation of the current MySQL XA implementation. It is not part of the XA specification.)

The XA RECOVER statement returns information for those XA transactions on the MySQL server that are in the PREPARED state. (See Section 12.3.7.2, “XA Transaction States”.) The output includes a row for each such XA transaction on the server, regardless of which client started it.

XA RECOVER output rows look like this (for an example xid value consisting of the parts 'abc', 'def', and 7):

mysql> XA RECOVER;
+----------+--------------+--------------+--------+
| formatID | gtrid_length | bqual_length | data   |
+----------+--------------+--------------+--------+
|        7 |            3 |            3 | abcdef |
+----------+--------------+--------------+--------+

The output columns have the following meanings:

  • formatID is the formatID part of the transaction xid

  • gtrid_length is the length in bytes of the gtrid part of the xid

  • bqual_length is the length in bytes of the bqual part of the xid

  • data is the concatenation of the gtrid and bqual parts of the xid

12.3.7.2. XA Transaction States

An XA transaction progresses through the following states:

  1. Use XA START to start an XA transaction and put it in the ACTIVE state.

  2. For an ACTIVE XA transaction, issue the SQL statements that make up the transaction, and then issue an XA END statement. XA END puts the transaction in the IDLE state.

  3. For an IDLE XA transaction, you can issue either an XA PREPARE statement or an XA COMMIT ... ONE PHASE statement:

    • XA PREPARE puts the transaction in the PREPARED state. An XA RECOVER statement at this point will include the transaction's xid value in its output, because XA RECOVER lists all XA transactions that are in the PREPARED state.

    • XA COMMIT ... ONE PHASE prepares and commits the transaction. The xid value will not be listed by XA RECOVER because the transaction terminates.

  4. For a PREPARED XA transaction, you can issue an XA COMMIT statement to commit and terminate the transaction, or XA ROLLBACK to roll back and terminate the transaction.

Here is a simple XA transaction that inserts a row into a table as part of a global transaction:

mysql> XA START 'xatest';
Query OK, 0 rows affected (0.00 sec)

mysql> INSERT INTO mytable (i) VALUES(10);
Query OK, 1 row affected (0.04 sec)

mysql> XA END 'xatest';
Query OK, 0 rows affected (0.00 sec)

mysql> XA PREPARE 'xatest';
Query OK, 0 rows affected (0.00 sec)

mysql> XA COMMIT 'xatest';
Query OK, 0 rows affected (0.00 sec)

Within the context of a given client connection, XA transactions and local (non-XA) transactions are mutually exclusive. For example, if XA START has been issued to begin an XA transaction, a local transaction cannot be started until the XA transaction has been committed or rolled back. Conversely, if a local transaction has been started with START TRANSACTION, no XA statements can be used until the transaction has been committed or rolled back.

Note that if an XA transaction is in the ACTIVE state, you cannot issue any statements that cause an implicit commit. That would violate the XA contract because you could not roll back the XA transaction. You will receive the following error if you try to execute such a statement:

ERROR 1399 (XAE07): XAER_RMFAIL: The command cannot be executed
when global transaction is in the ACTIVE state

Statements to which the preceding remark applies are listed at Section 12.3.3, “Statements That Cause an Implicit Commit”.

12.4. Database Administration Statements

12.4.1. Account Management Statements

MySQL account information is stored in the tables of the mysql database. This database and the access control system are discussed extensively in Chapter 5, MySQL Server Administration, which you should consult for additional details.

Important

Some releases of MySQL introduce changes to the structure of the grant tables to add new privileges or features. Whenever you update to a new version of MySQL, you should update your grant tables to make sure that they have the current structure so that you can take advantage of any new capabilities. See Section 4.4.7, “mysql_upgrade — Check Tables for MySQL Upgrade”.

12.4.1.1. CREATE USER Syntax

CREATE USER user_specification
    [, user_specification] ...

user_specification:
    user
    [
        IDENTIFIED BY [PASSWORD] 'password'
      | IDENTIFIED WITH auth_plugin [AS 'auth_string']
    ]

The CREATE USER statement creates new MySQL accounts. To use it, you must have the global CREATE USER privilege or the INSERT privilege for the mysql database. For each account, CREATE USER creates a new row in the mysql.user table and assigns the account no privileges. An error occurs if the account already exists.

Each account name uses the format described in Section 5.4.3, “Specifying Account Names”. For example:

CREATE USER 'jeffrey'@'localhost' IDENTIFIED BY 'mypass';

If you specify only the user name part of the account name, a host name part of '%' is used.

The user specification may indicate how the user should authenticate when connecting to the server:

  • To enable the user to connect with no password (which is insecure), include no IDENTIFIED BY clause:

    CREATE USER 'jeffrey'@'localhost';
    

    In this case, the server uses built-in authentication and clients must provide no password.

  • To assign a password, use IDENTIFIED BY with the literal plaintext password value:

    CREATE USER 'jeffrey'@'localhost' IDENTIFIED BY 'mypass';
    

    The server uses built-in authentication and clients must match the given password.

  • To avoid specifying the plaintext password if you know its hash value (the value that PASSWORD() would return for the password), specify the hash value preceded by the keyword PASSWORD:

    CREATE USER 'jeffrey'@'localhost'
    IDENTIFIED BY PASSWORD '*90E462C37378CED12064BB3388827D2BA3A9B689';
    

    The server uses built-in authentication and clients must match the given password.

  • If the account should authenticate using a specific authentication plugin, use IDENTIFIED WITH. auth_plugin is an authentication plugin name. It can be an unquoted name or a quoted string literal. 'auth_string' is an optional quoted string literal to pass to the plugin. The plugin interprets the meaning of the string.

    CREATE USER 'jeffrey'@'localhost'
    IDENTIFIED WITH my_auth_plugin;
    

    The server uses the named plugin and clients must provide credentials as required for the authentication method that the plugin implements. If the server cannot find the plugin, an error occurs. IDENTIFIED WITH can be given as of MySQL 5.5.7.

The IDENTIFIED BY and IDENTIFIED WITH clauses are mutually exclusive, so at most one of them can be specified for a given user.

For additional information about setting passwords, see Section 5.5.5, “Assigning Account Passwords”.

Important

CREATE USER may be recorded in server logs or in a history file such as ~/.mysql_history, which means that plaintext passwords may be read by anyone having read access to that information. See Section 5.3.2, “Password Security in MySQL”.

Important

Some releases of MySQL introduce changes to the structure of the grant tables to add new privileges or features. Whenever you update to a new version of MySQL, you should update your grant tables to make sure that they have the current structure so that you can take advantage of any new capabilities. See Section 4.4.7, “mysql_upgrade — Check Tables for MySQL Upgrade”.

12.4.1.2. DROP USER Syntax

DROP USER user [, user] ...

The DROP USER statement removes one or more MySQL accounts and their privileges. It removes privilege rows for the account from all grant tables. To use this statement, you must have the global CREATE USER privilege or the DELETE privilege for the mysql database. Each account name uses the format described in Section 5.4.3, “Specifying Account Names”. For example:

DROP USER 'jeffrey'@'localhost';

If you specify only the user name part of the account name, a host name part of '%' is used.

Important

DROP USER does not automatically close any open user sessions. Rather, in the event that a user with an open session is dropped, the statement does not take effect until that user's session is closed. Once the session is closed, the user is dropped, and that user's next attempt to log in will fail. This is by design.

DROP USER does not automatically drop or invalidate databases or objects within them that the old user created. This includes stored programs or views for which the DEFINER attribute names the dropped user. Attempts to access such objects may produce an error if they execute in definer security context. (For information about security context, see Section 19.6, “Access Control for Stored Programs and Views”.)

12.4.1.3. GRANT Syntax

GRANT
    priv_type [(column_list)]
      [, priv_type [(column_list)]] ...
    ON [object_type] priv_level
    TO user_specification [, user_specification] ...
    [REQUIRE {NONE | ssl_option [[AND] ssl_option] ...}]
    [WITH with_option ...]

GRANT PROXY ON user_specification
    TO user_specification [, user_specification] ...
    [WITH GRANT OPTION]

object_type:
    TABLE
  | FUNCTION
  | PROCEDURE

priv_level:
    *
  | *.*
  | db_name.*
  | db_name.tbl_name
  | tbl_name
  | db_name.routine_name

user_specification:
    user
    [
        IDENTIFIED BY [PASSWORD] 'password'
      | IDENTIFIED WITH auth_plugin [AS 'auth_string']
    ]

ssl_option:
    SSL
  | X509
  | CIPHER 'cipher'
  | ISSUER 'issuer'
  | SUBJECT 'subject'

with_option:
    GRANT OPTION
  | MAX_QUERIES_PER_HOUR count
  | MAX_UPDATES_PER_HOUR count
  | MAX_CONNECTIONS_PER_HOUR count
  | MAX_USER_CONNECTIONS count

The GRANT statement grants privileges to MySQL user accounts. GRANT also serves to specify other account characteristics such as use of secure connections and limits on access to server resources. To use GRANT, you must have the GRANT OPTION privilege, and you must have the privileges that you are granting.

Normally, a database administrator first uses CREATE USER to create an account, then GRANT to define its privileges and characteristics. For example:

CREATE USER 'jeffrey'@'localhost' IDENTIFIED BY 'mypass';
GRANT ALL ON db1.* TO 'jeffrey'@'localhost';
GRANT SELECT ON db2.invoice TO 'jeffrey'@'localhost';
GRANT USAGE ON *.* TO 'jeffrey'@'localhost' WITH MAX_QUERIES_PER_HOUR 90;

However, if an account named in a GRANT statement does not already exist, GRANT may create it under the conditions described later in the discussion of the NO_AUTO_CREATE_USER SQL mode.

The REVOKE statement is related to GRANT and enables administrators to remove account privileges. See Section 12.4.1.5, “REVOKE Syntax”.

To determine what privileges an account has, use SHOW GRANTS. See Section 12.4.5.22, “SHOW GRANTS Syntax”.

There are several aspects to the GRANT statement, described under the folllowing topics in this section:

Important

Some releases of MySQL introduce changes to the structure of the grant tables to add new privileges or features. Whenever you update to a new version of MySQL, you should update your grant tables to make sure that they have the current structure so that you can take advantage of any new capabilities. See Section 4.4.7, “mysql_upgrade — Check Tables for MySQL Upgrade”.

Privileges Supported by MySQL

The following table summarizes the permissible priv_type privilege types that can be specified for the GRANT and REVOKE statements. For additional information about these privileges, see Section 5.4.1, “Privileges Provided by MySQL”.

Table 12.1. Permissible Privileges for GRANT and REVOKE

PrivilegeMeaning
ALL [PRIVILEGES]Grant all privileges at specified access level except GRANT OPTION
ALTEREnable use of ALTER TABLE
ALTER ROUTINEEnable stored routines to be altered or dropped
CREATEEnable database and table creation
CREATE ROUTINEEnable stored routine creation
CREATE TABLESPACEEnable tablespaces and log file groups to be created, altered, or dropped
CREATE TEMPORARY TABLESEnable use of CREATE TEMPORARY TABLE
CREATE USEREnable use of CREATE USER, DROP USER, RENAME USER, and REVOKE ALL PRIVILEGES
CREATE VIEWEnable views to be created or altered
DELETEEnable use of DELETE
DROPEnable databases, tables, and views to be dropped
EVENTEnable use of events for the Event Scheduler
EXECUTEEnable the user to execute stored routines
FILEEnable the user to cause the server to read or write files
GRANT OPTIONEnable privileges to be granted to or removed from other accounts
INDEXEnable indexes to be created or dropped
INSERTEnable use of INSERT
LOCK TABLESEnable use of LOCK TABLES on tables for which you have the SELECT privilege
PROCESSEnable the user to see all processes with SHOW PROCESSLIST
PROXYEnable user proxying
REFERENCESNot implemented
RELOADEnable use of FLUSH operations
REPLICATION CLIENTEnable the user to ask where master or slave servers are
REPLICATION SLAVEEnable replication slaves to read binary log events from the master
SELECTEnable use of SELECT
SHOW DATABASESEnable SHOW DATABASES to show all databases
SHOW VIEWEnable use of SHOW CREATE VIEW
SHUTDOWNEnable use of mysqladmin shutdown
SUPEREnable use of other adminstrative operations such as CHANGE MASTER TO, KILL, PURGE BINARY LOGS, SET GLOBAL, and mysqladmin debug command
TRIGGEREnable trigger operations
UPDATEEnable use of UPDATE
USAGESynonym for “no privileges

The PROXY privilege was added in MySQL 5.5.7.

A trigger is associated with a table, so to create or drop a trigger, you must have the TRIGGER privilege for the table, not the trigger.

In GRANT statements, the ALL [PRIVILEGES] or PROXY privilege must be named by itself and cannot be specified along with other privileges. ALL [PRIVILEGES] stands for all privileges available for the level at which privileges are to be granted except for the GRANT OPTION and PROXY privileges.

USAGE can be specified to create a user that has no privileges, or to specify the REQUIRE or WITH clauses for an account without changing its existing privileges.

MySQL account information is stored in the tables of the mysql database. This database and the access control system are discussed extensively in Section 5.4, “The MySQL Access Privilege System”, which you should consult for additional details.

If the grant tables hold privilege rows that contain mixed-case database or table names and the lower_case_table_names system variable is set to a nonzero value, REVOKE cannot be used to revoke these privileges. It will be necessary to manipulate the grant tables directly. (GRANT will not create such rows when lower_case_table_names is set, but such rows might have been created prior to setting that variable.)

Privileges can be granted at several levels, depending on the syntax used for the ON clause. For REVOKE, the same ON syntax specifies which privileges to take away. The examples shown here include no IDENTIFIED BY 'password' clause for brevity, but you should include one if the account does not already exist, to avoid creating an insecure account that has no password.

Global Privileges

Global privileges are administrative or apply to all databases on a given server. To assign global privileges, use ON *.* syntax:

GRANT ALL ON *.* TO 'someuser'@'somehost';
GRANT SELECT, INSERT ON *.* TO 'someuser'@'somehost';

The CREATE TABLESPACE, CREATE USER, FILE, PROCESS, RELOAD, REPLICATION CLIENT, REPLICATION SLAVE, SHOW DATABASES, SHUTDOWN, and SUPER privileges are administrative and can only be granted globally.

Other privileges can be granted globally or at more specific levels.

MySQL stores global privileges in the mysql.user table.

Database Privileges

Database privileges apply to all objects in a given database. To assign database-level privileges, use ON db_name.* syntax:

GRANT ALL ON mydb.* TO 'someuser'@'somehost';
GRANT SELECT, INSERT ON mydb.* TO 'someuser'@'somehost';

If you use ON * syntax (rather than ON *.* and you have selected a default database, privileges are assigned at the database level for the default database. An error occurs if there is no default database.

The CREATE, DROP, EVENT, and GRANT OPTION privileges can be specified at the database level. Table or routine privileges also can be specified at the database level, in which case they apply to all tables or routines in the database.

MySQL stores database privileges in the mysql.db table.

Table Privileges

Table privileges apply to all columns in a given table. To assign table-level privileges, use ON db_name.tbl_name syntax:

GRANT ALL ON mydb.mytbl TO 'someuser'@'somehost';
GRANT SELECT, INSERT ON mydb.mytbl TO 'someuser'@'somehost';

If you specify tbl_name rather than db_name.tbl_name, the statement applies to tbl_name in the default database. An error occurs if there is no default database.

The permissible column-level priv_type values are ALTER, CREATE VIEW, CREATE, DELETE, DROP, GRANT OPTION, INDEX, INSERT, SELECT, SHOW VIEW, TRIGGER, and UPDATE.

MySQL stores table privileges in the mysql.tables_priv table.

Column Privileges

Column privileges apply to single columns in a given table. Each privilege to be granted at the column level must be followed by the column or columns, enclosed within parentheses.

GRANT SELECT (col1), INSERT (col1,col2) ON mydb.mytbl TO 'someuser'@'somehost';

The permissible priv_type values for a column (that is, when you use a column_list clause) are INSERT, SELECT, and UPDATE.

MySQL stores column privileges in the mysql.columns_priv table.

Stored Routine Privileges

The ALTER ROUTINE, CREATE ROUTINE, EXECUTE, and GRANT OPTION privileges apply to stored routines (procedures and functions). They can be granted at the global and database levels. Except for CREATE ROUTINE, these privileges can be granted at the routine level for individual routines.

GRANT CREATE ROUTINE ON mydb.* TO 'someuser'@'somehost';
GRANT EXECUTE ON PROCEDURE mydb.myproc TO 'someuser'@'somehost';

The permissible priv_type values at the routine level are ALTER ROUTINE, EXECUTE, and GRANT OPTION. CREATE ROUTINE is not a routine-level privilege because you must have this privilege to create a routine in the first place.

MySQL stores routine-level privileges in the mysql.procs_priv table.

Proxy User Privileges

The PROXY privilege enables one user to be a proxy for another. The proxy user impersonates or takes the identity of the proxied user.

GRANT PROXY ON 'localuser'@'localhost' TO 'externaluser'@'somehost';

When PROXY is granted, it must be the only privilege named in the GRANT statement, the REQUIRE clause cannot be given, and the only permitted WITH option is WITH GRANT OPTION.

Proxying requires that the proxy user authenticate through a plugin that returns the name of the proxied user to the server when the proxy user connects, and that the proxy user have the PROXY privilege for the proxied user. For details and examples, see Section 5.5.7, “Proxy Users”.

MySQL stores proxy privileges in the mysql.proxies_priv table.

For the global, database, table, and routine levels, GRANT ALL assigns only the privileges that exist at the level you are granting. For example, GRANT ALL ON db_name.* is a database-level statement, so it does not grant any global-only privileges such as FILE. Granting ALL does not assign the PROXY privilege.

The object_type clause, if present, should be specified as TABLE, FUNCTION, or PROCEDURE when the following object is a table, a stored function, or a stored procedure.

The privileges for a database, table, column, or routine are formed additively as the logical OR of the privileges at each of the privilege levels. For example, if a user has a global SELECT privilege, the privilege cannot be denied by an absence of the privilege at the database, table, or column level. Details of the privilege-checking procedure are presented in Section 5.4.5, “Access Control, Stage 2: Request Verification”.

MySQL enables you to grant privileges on databases or tables that do not exist. For tables, the privileges to be granted must include the CREATE privilege. This behavior is by design, and is intended to enable the database administrator to prepare user accounts and privileges for databases or tables that are to be created at a later time.

Important

MySQL does not automatically revoke any privileges when you drop a database or table. However, if you drop a routine, any routine-level privileges granted for that routine are revoked.

Account Names and Passwords

The user value indicates the MySQL account to which the GRANT statement applies. To accommodate granting rights to users from arbitrary hosts, MySQL supports specifying the user value in the form user_name@host_name. If a user_name or host_name value is legal as an unquoted identifier, you need not quote it. However, quotation marks are necessary to specify a user_name string containing special characters (such as “-”), or a host_name string containing special characters or wildcard characters (such as “%”); for example, 'test-user'@'%.com'. Quote the user name and host name separately.

You can specify wildcards in the host name. For example, user_name@'%.example.com' applies to user_name for any host in the example.com domain, and user_name@'192.168.1.%' applies to user_name for any host in the 192.168.1 class C subnet.

The simple form user_name is a synonym for user_name@'%'.

MySQL does not support wildcards in user names. To refer to an anonymous user, specify an account with an empty user name with the GRANT statement:

GRANT ALL ON test.* TO ''@'localhost' ...

In this case, any user who connects from the local host with the correct password for the anonymous user will be permitted access, with the privileges associated with the anonymous-user account.

For additional information about user name and host name values in account names, see Section 5.4.3, “Specifying Account Names”.

To specify quoted values, quote database, table, column, and routine names as identifiers. Quote user names and host names as identifiers or as strings. Quote passwords as strings. For string-quoting and identifier-quoting guidelines, see Section 8.1.1, “Strings”, and Section 8.2, “Schema Object Names”.

The “_” and “%” wildcards are permitted when specifying database names in GRANT statements that grant privileges at the global or database levels. This means, for example, that if you want to use a “_” character as part of a database name, you should specify it as “\_” in the GRANT statement, to prevent the user from being able to access additional databases matching the wildcard pattern; for example, GRANT ... ON `foo\_bar`.* TO ....

Warning

If you permit anonymous users to connect to the MySQL server, you should also grant privileges to all local users as user_name@localhost. Otherwise, the anonymous user account for localhost in the mysql.user table (created during MySQL installation) is used when named users try to log in to the MySQL server from the local machine. For details, see Section 5.4.4, “Access Control, Stage 1: Connection Verification”.

To determine whether the preceding warning applies to you, execute the following query, which lists any anonymous users:

SELECT Host, User FROM mysql.user WHERE User='';

To avoid the problem just described, delete the local anonymous user account using this statement:

DROP USER ''@'localhost';

GRANT supports host names up to 60 characters long. Database, table, column, and routine names can be up to 64 characters. User names can be up to 16 characters.

Warning

The permissible length for user names cannot be changed by altering the mysql.user table. Attempting to do so results in unpredictable behavior which may even make it impossible for users to log in to the MySQL server. You should never alter any of the tables in the mysql database in any manner whatsoever except by means of the procedure described in Section 4.4.7, “mysql_upgrade — Check Tables for MySQL Upgrade”.

The user specification may indicate how the user should authenticate when connecting to the server, through inclusion of an IDENTIFIED BY or IDENTIFIED WITH clause. The syntax is the same as for the CREATE USER statement. See Section 12.4.1.1, “CREATE USER Syntax”.

When the IDENTIFIED BY clause is present and you have global grant privileges, the password becomes the new password for the account, even if the account exists and already has a password. With no IDENTIFIED BY clause, the account password remains unchanged.

If the NO_AUTO_CREATE_USER SQL mode is not enabled and the account named in a GRANT statement does not exist in the mysql.user table, GRANT creates it. If you specify no IDENTIFIED BY clause or provide an empty password, the user has no password. This is very insecure.

If NO_AUTO_CREATE_USER is enabled and the account does not exist, GRANT fails and does not create the account unless the IDENTIFIED BY clause is given to provide a nonempty password.

The NO_AUTO_CREATE_USER SQL mode has no effect for GRANT statements that include an IDENTIFIED WITH clause. That is, GRANT ... IDENTIFIED WITH creates nonexistent users regardless of the mode setting.

Important

GRANT may be recorded in server logs or in a history file such as ~/.mysql_history, which means that plaintext passwords may be read by anyone having read access to that information. See Section 5.3.2, “Password Security in MySQL”.

Other Account Characteristics

The WITH clause is used for several purposes:

  • To enable a user to grant privileges to other users

  • To specify resource limits for a user

  • To specify whether and how a user must use secure connections to the server

The WITH GRANT OPTION clause gives the user the ability to give to other users any privileges the user has at the specified privilege level. You should be careful to whom you give the GRANT OPTION privilege because two users with different privileges may be able to combine privileges!

You cannot grant another user a privilege which you yourself do not have; the GRANT OPTION privilege enables you to assign only those privileges which you yourself possess.

Be aware that when you grant a user the GRANT OPTION privilege at a particular privilege level, any privileges the user possesses (or may be given in the future) at that level can also be granted by that user to other users. Suppose that you grant a user the INSERT privilege on a database. If you then grant the SELECT privilege on the database and specify WITH GRANT OPTION, that user can give to other users not only the SELECT privilege, but also INSERT. If you then grant the UPDATE privilege to the user on the database, the user can grant INSERT, SELECT, and UPDATE.

For a nonadministrative user, you should not grant the ALTER privilege globally or for the mysql database. If you do that, the user can try to subvert the privilege system by renaming tables!

For additional information about security risks associated with particular privileges, see Section 5.4.1, “Privileges Provided by MySQL”.

Several WITH clause options specify limits on use of server resources by an account:

  • The MAX_QUERIES_PER_HOUR count, MAX_UPDATES_PER_HOUR count, and MAX_CONNECTIONS_PER_HOUR count limits restrict the number of queries, updates, and connections to the server permitted to this account during any given one-hour period. (Queries for which results are served from the query cache do not count against the MAX_QUERIES_PER_HOUR limit.) If count is 0 (the default), this means that there is no limitation for the account.

  • The MAX_USER_CONNECTIONS count limit restricts the maximum number of simultaneous connections to the server by the account. A nonzero count specifies the limit for the account explicitly. If count is 0 (the default), the server determines the number of simultaneous connections for the account from the global value of the max_user_connections system variable. If max_user_connections is also zero, there is no limit for the account.

To specify resource limits for an existing user without affecting existing privileges, use GRANT USAGE at the global level (ON *.*) and name the limits to be changed. For example:

GRANT USAGE ON *.* TO ...
  WITH MAX_QUERIES_PER_HOUR 500 MAX_UPDATES_PER_HOUR 100;

Limits not specified retain their current values.

For more information on restricting access to server resources, see Section 5.5.4, “Setting Account Resource Limits”.

MySQL can check X509 certificate attributes in addition to the usual authentication that is based on the user name and password. To specify SSL-related options for a MySQL account, use the REQUIRE clause of the GRANT statement. (For background information on the use of SSL with MySQL, see Section 5.5.8, “Using SSL for Secure Connections”.)

There are a number of different possibilities for limiting connection types for a given account:

  • REQUIRE NONE indicates that the account has no SSL or X509 requirements. This is the default if no SSL-related REQUIRE options are specified. Unencrypted connections are permitted if the user name and password are valid. However, encrypted connections can also be used, at the client's option, if the client has the proper certificate and key files. That is, the client need not specify any SSL command options, in which case the connection will be unencrypted. To use an encrypted connection, the client must specify either the --ssl-ca option, or all three of the --ssl-ca, --ssl-key, and --ssl-cert options.

  • The REQUIRE SSL option tells the server to permit only SSL-encrypted connections for the account.

    GRANT ALL PRIVILEGES ON test.* TO 'root'@'localhost'
      IDENTIFIED BY 'goodsecret' REQUIRE SSL;
    

    To connect, the client must specify the --ssl-ca option, and may additionally specify the --ssl-key and --ssl-cert options.

  • REQUIRE X509 means that the client must have a valid certificate but that the exact certificate, issuer, and subject do not matter. The only requirement is that it should be possible to verify its signature with one of the CA certificates.

    GRANT ALL PRIVILEGES ON test.* TO 'root'@'localhost'
      IDENTIFIED BY 'goodsecret' REQUIRE X509;
    

    To connect, the client must specify the --ssl-ca, --ssl-key, and --ssl-cert options. This is also true for ISSUER and SUBJECT because those REQUIRE options imply X509.

  • REQUIRE ISSUER 'issuer' places the restriction on connection attempts that the client must present a valid X509 certificate issued by CA 'issuer'. If the client presents a certificate that is valid but has a different issuer, the server rejects the connection. Use of X509 certificates always implies encryption, so the SSL option is unnecessary in this case.

    GRANT ALL PRIVILEGES ON test.* TO 'root'@'localhost'
      IDENTIFIED BY 'goodsecret'
      REQUIRE ISSUER '/C=FI/ST=Some-State/L=Helsinki/
        O=MySQL Finland AB/CN=Tonu Samuel/[email protected]';
    

    Note that the 'issuer' value should be entered as a single string.

  • REQUIRE SUBJECT 'subject' places the restriction on connection attempts that the client must present a valid X509 certificate containing the subject subject. If the client presents a certificate that is valid but has a different subject, the server rejects the connection.

    GRANT ALL PRIVILEGES ON test.* TO 'root'@'localhost'
      IDENTIFIED BY 'goodsecret'
      REQUIRE SUBJECT '/C=EE/ST=Some-State/L=Tallinn/
        O=MySQL demo client certificate/
        CN=Tonu Samuel/[email protected]';
    

    Note that the 'subject' value should be entered as a single string.

  • REQUIRE CIPHER 'cipher' is needed to ensure that ciphers and key lengths of sufficient strength are used. SSL itself can be weak if old algorithms using short encryption keys are used. Using this option, you can ask that a specific cipher method is used for a connection.

    GRANT ALL PRIVILEGES ON test.* TO 'root'@'localhost'
      IDENTIFIED BY 'goodsecret'
      REQUIRE CIPHER 'EDH-RSA-DES-CBC3-SHA';
    

The SUBJECT, ISSUER, and CIPHER options can be combined in the REQUIRE clause like this:

GRANT ALL PRIVILEGES ON test.* TO 'root'@'localhost'
  IDENTIFIED BY 'goodsecret'
  REQUIRE SUBJECT '/C=EE/ST=Some-State/L=Tallinn/
    O=MySQL demo client certificate/
    CN=Tonu Samuel/[email protected]'
  AND ISSUER '/C=FI/ST=Some-State/L=Helsinki/
    O=MySQL Finland AB/CN=Tonu Samuel/[email protected]'
  AND CIPHER 'EDH-RSA-DES-CBC3-SHA';

The order of the options does not matter, but no option can be specified twice. The AND keyword is optional between REQUIRE options.

If you are using table, column, or routine privileges for even one user, the server examines table, column, and routine privileges for all users and this slows down MySQL a bit. Similarly, if you limit the number of queries, updates, or connections for any users, the server must monitor these values.

MySQL and Standard SQL Versions of GRANT

The biggest differences between the MySQL and standard SQL versions of GRANT are:

  • MySQL associates privileges with the combination of a host name and user name and not with only a user name.

  • Standard SQL does not have global or database-level privileges, nor does it support all the privilege types that MySQL supports.

  • MySQL does not support the standard SQL UNDER privilege.

  • Standard SQL privileges are structured in a hierarchical manner. If you remove a user, all privileges the user has been granted are revoked. This is also true in MySQL if you use DROP USER. See Section 12.4.1.2, “DROP USER Syntax”.

  • In standard SQL, when you drop a table, all privileges for the table are revoked. In standard SQL, when you revoke a privilege, all privileges that were granted based on that privilege are also revoked. In MySQL, privileges can be dropped only with explicit DROP USER or REVOKE statements or by manipulating the MySQL grant tables directly.

  • In MySQL, it is possible to have the INSERT privilege for only some of the columns in a table. In this case, you can still execute INSERT statements on the table, provided that you insert values only for those columns for which you have the INSERT privilege. The omitted columns are set to their implicit default values if strict SQL mode is not enabled. In strict mode, the statement is rejected if any of the omitted columns have no default value. (Standard SQL requires you to have the INSERT privilege on all columns.) Section 5.1.7, “Server SQL Modes”, discusses strict mode. Section 10.1.4, “Data Type Default Values”, discusses implicit default values.

12.4.1.4. RENAME USER Syntax

RENAME USER old_user TO new_user
    [, old_user TO new_user] ...

The RENAME USER statement renames existing MySQL accounts. To use it, you must have the global CREATE USER privilege or the UPDATE privilege for the mysql database. An error occurs if any old account does not exist or any new account exists. Each account name uses the format described in Section 5.4.3, “Specifying Account Names”. For example:

RENAME USER 'jeffrey'@'localhost' TO 'jeff'@'127.0.0.1';

If you specify only the user name part of the account name, a host name part of '%' is used.

RENAME USER causes the privileges held by the old user to be those held by the new user. However, RENAME USER does not automatically drop or invalidate databases or objects within them that the old user created. This includes stored programs or views for which the DEFINER attribute names the old user. Attempts to access such objects may produce an error if they execute in definer security context. (For information about security context, see Section 19.6, “Access Control for Stored Programs and Views”.)

The privilege changes take effect as indicated in Section 5.4.6, “When Privilege Changes Take Effect”.

12.4.1.5. REVOKE Syntax

REVOKE
    priv_type [(column_list)]
      [, priv_type [(column_list)]] ...
    ON [object_type] priv_level
    FROM user [, user] ...

REVOKE ALL PRIVILEGES, GRANT OPTION
    FROM user [, user] ...

REVOKE PROXY ON user
    FROM user [, user] ...

The REVOKE statement enables system administrators to revoke privileges from MySQL accounts. Each account name uses the format described in Section 5.4.3, “Specifying Account Names”. For example:

REVOKE INSERT ON *.* FROM 'jeffrey'@'localhost';

If you specify only the user name part of the account name, a host name part of '%' is used.

For details on the levels at which privileges exist, the permissible priv_type and priv_level values, and the syntax for specifying users and passwords, see Section 12.4.1.3, “GRANT Syntax”

To use the first REVOKE syntax, you must have the GRANT OPTION privilege, and you must have the privileges that you are revoking.

To revoke all privileges, use the second syntax, which drops all global, database, table, column, and routine privileges for the named user or users:

REVOKE ALL PRIVILEGES, GRANT OPTION FROM user [, user] ...

To use this REVOKE syntax, you must have the global CREATE USER privilege or the UPDATE privilege for the mysql database.

REVOKE removes privileges, but does not drop mysql.user table entries. To remove a user account entirely, use DROP USER (see Section 12.4.1.2, “DROP USER Syntax”) or DELETE.

If the grant tables hold privilege rows that contain mixed-case database or table names and the lower_case_table_names system variable is set to a nonzero value, REVOKE cannot be used to revoke these privileges. It will be necessary to manipulate the grant tables directly. (GRANT will not create such rows when lower_case_table_names is set, but such rows might have been created prior to setting the variable.)

To verify an account's privileges after a REVOKE operation, use SHOW GRANTS. See Section 12.4.5.22, “SHOW GRANTS Syntax”.

12.4.1.6. SET PASSWORD Syntax

SET PASSWORD [FOR user] =
    {
        PASSWORD('some password')
      | OLD_PASSWORD('some password')
      | 'encrypted password'
    }

The SET PASSWORD statement assigns a password to an existing MySQL user account.

If the password is specified using the PASSWORD() or OLD_PASSWORD() function, the literal text of the password should be given. If the password is specified without using either function, the password should be the already-encrypted password value as returned by PASSWORD().

With no FOR clause, this statement sets the password for the current user. Any client that has connected to the server using a nonanonymous account can change the password for that account.

With a FOR clause, this statement sets the password for a specific account on the current server host. Only clients that have the UPDATE privilege for the mysql database can do this. The user value should be given in user_name@host_name format, where user_name and host_name are exactly as they are listed in the User and Host columns of the mysql.user table entry. For example, if you had an entry with User and Host column values of 'bob' and '%.loc.gov', you would write the statement like this:

SET PASSWORD FOR 'bob'@'%.loc.gov' = PASSWORD('newpass');

That is equivalent to the following statements:

UPDATE mysql.user SET Password=PASSWORD('newpass')
  WHERE User='bob' AND Host='%.loc.gov';
FLUSH PRIVILEGES;

Another way to set the password is to use GRANT:

GRANT USAGE ON *.* TO 'bob'@'%.loc.gov' IDENTIFIED BY 'newpass';
Important

SET PASSWORD may be recorded in server logs or in a history file such as ~/.mysql_history, which means that plaintext passwords may be read by anyone having read access to that information. See Section 5.3.2, “Password Security in MySQL”.

Note

If you are connecting to a MySQL 4.1 or later server using a pre-4.1 client program, do not use the preceding SET PASSWORD or UPDATE statement without reading Section 5.3.2.3, “Password Hashing in MySQL”, first. The password format changed in MySQL 4.1, and under certain circumstances it is possible that if you change your password, you might not be able to connect to the server afterward.

To see which account the server authenticated you as, invoke the CURRENT_USER() function.

For more information about setting passwords, see Section 5.5.5, “Assigning Account Passwords”

12.4.2. Table Maintenance Statements

12.4.2.1. ANALYZE TABLE Syntax

ANALYZE [NO_WRITE_TO_BINLOG | LOCAL] TABLE
    tbl_name [, tbl_name] ...

ANALYZE TABLE analyzes and stores the key distribution for a table. During the analysis, the table is locked with a read lock for MyISAM and InnoDB. This statement works with MyISAM and InnoDB tables. For MyISAM tables, this statement is equivalent to using myisamchk --analyze.

For more information on how the analysis works within InnoDB, see Section 13.6.15, “Limits on InnoDB Tables”.

MySQL uses the stored key distribution to decide the order in which tables should be joined when you perform a join on something other than a constant. In addition, key distributions can be used when deciding which indexes to use for a specific table within a query.

This statement requires SELECT and INSERT privileges for the table.

ANALYZE TABLE is supported for partitioned tables, and you can use ALTER TABLE ... ANALYZE PARTITION to analyze one or more partitions; for more information, see Section 12.1.6, “ALTER TABLE Syntax”, and Section 18.3.3, “Maintenance of Partitions”.

ANALYZE TABLE returns a result set with the following columns.

ColumnValue
TableThe table name
OpAlways analyze
Msg_typestatus, error, info, note, or warning
Msg_textAn informational message

You can check the stored key distribution with the SHOW INDEX statement. See Section 12.4.5.23, “SHOW INDEX Syntax”.

If the table has not changed since the last ANALYZE TABLE statement, the table is not analyzed again.

By default, ANALYZE TABLE statements are written to the binary log so that they will be replicated to replication slaves. Logging can be suppressed with the optional NO_WRITE_TO_BINLOG keyword or its alias LOCAL.

12.4.2.2. CHECK TABLE Syntax

CHECK TABLE tbl_name [, tbl_name] ... [option] ...

option = {FOR UPGRADE | QUICK | FAST | MEDIUM | EXTENDED | CHANGED}

CHECK TABLE checks a table or tables for errors. CHECK TABLE works for MyISAM, InnoDB, ARCHIVE, and CSV tables. For MyISAM tables, the key statistics are updated as well.

To check a table, you must have some privilege for it.

CHECK TABLE can also check views for problems, such as tables that are referenced in the view definition that no longer exist.

CHECK TABLE is supported for partitioned tables, and you can use ALTER TABLE ... CHECK PARTITION to check one or more partitions; for more information, see Section 12.1.6, “ALTER TABLE Syntax”, and Section 18.3.3, “Maintenance of Partitions”.

CHECK TABLE returns a result set with the following columns.

ColumnValue
TableThe table name
OpAlways check
Msg_typestatus, error, info, note, or warning
Msg_textAn informational message

Note that the statement might produce many rows of information for each checked table. The last row has a Msg_type value of status and the Msg_text normally should be OK. If you don't get OK, or Table is already up to date you should normally run a repair of the table. See Section 6.6, “MyISAM Table Maintenance and Crash Recovery”. Table is already up to date means that the storage engine for the table indicated that there was no need to check the table.

The FOR UPGRADE option checks whether the named tables are compatible with the current version of MySQL. With FOR UPGRADE, the server checks each table to determine whether there have been any incompatible changes in any of the table's data types or indexes since the table was created. If not, the check succeeds. Otherwise, if there is a possible incompatibility, the server runs a full check on the table (which might take some time). If the full check succeeds, the server marks the table's .frm file with the current MySQL version number. Marking the .frm file ensures that further checks for the table with the same version of the server will be fast.

Incompatibilities might occur because the storage format for a data type has changed or because its sort order has changed. Our aim is to avoid these changes, but occasionally they are necessary to correct problems that would be worse than an incompatibility between releases.

Currently, FOR UPGRADE discovers these incompatibilities:

  • The indexing order for end-space in TEXT columns for InnoDB and MyISAM tables changed between MySQL 4.1 and 5.0.

  • The storage method of the new DECIMAL data type changed between MySQL 5.0.3 and 5.0.5.

  • If your table was created by a different version of the MySQL server than the one you are currently running, FOR UPGRADE indicates that the table has an .frm file with an incompatible version. In this case, the result set returned by CHECK TABLE contains a line with a Msg_type value of error and a Msg_text value of Table upgrade required. Please do "REPAIR TABLE `tbl_name`" to fix it!

  • Changes are sometimes made to character sets or collations that require table indexes to be rebuilt. For details about these changes and when FOR UPGRADE detects them, see Section 2.13.3, “Checking Whether Tables or Indexes Must Be Rebuilt”.

The other check options that can be given are shown in the following table. These options are passed to the storage engine, which may use them or not. MyISAM uses them; they are ignored for InnoDB tables and views.

TypeMeaning
QUICKDo not scan the rows to check for incorrect links.
FASTCheck only tables that have not been closed properly.
CHANGEDCheck only tables that have been changed since the last check or that have not been closed properly.
MEDIUMScan rows to verify that deleted links are valid. This also calculates a key checksum for the rows and verifies this with a calculated checksum for the keys.
EXTENDEDDo a full key lookup for all keys for each row. This ensures that the table is 100% consistent, but takes a long time.

If none of the options QUICK, MEDIUM, or EXTENDED are specified, the default check type for dynamic-format MyISAM tables is MEDIUM. This has the same result as running myisamchk --medium-check tbl_name on the table. The default check type also is MEDIUM for static-format MyISAM tables, unless CHANGED or FAST is specified. In that case, the default is QUICK. The row scan is skipped for CHANGED and FAST because the rows are very seldom corrupted.

You can combine check options, as in the following example that does a quick check on the table to determine whether it was closed properly:

CHECK TABLE test_table FAST QUICK;
Note

In some cases, CHECK TABLE changes the table. This happens if the table is marked as “corrupted” or “not closed properly” but CHECK TABLE does not find any problems in the table. In this case, CHECK TABLE marks the table as okay.

If a table is corrupted, it is most likely that the problem is in the indexes and not in the data part. All of the preceding check types check the indexes thoroughly and should thus find most errors.

If you just want to check a table that you assume is okay, you should use no check options or the QUICK option. The latter should be used when you are in a hurry and can take the very small risk that QUICK does not find an error in the data file. (In most cases, under normal usage, MySQL should find any error in the data file. If this happens, the table is marked as “corrupted” and cannot be used until it is repaired.)

FAST and CHANGED are mostly intended to be used from a script (for example, to be executed from cron) if you want to check tables from time to time. In most cases, FAST is to be preferred over CHANGED. (The only case when it is not preferred is when you suspect that you have found a bug in the MyISAM code.)

EXTENDED is to be used only after you have run a normal check but still get strange errors from a table when MySQL tries to update a row or find a row by key. This is very unlikely if a normal check has succeeded.

Use of CHECK TABLE ... EXTENDED might influence the execution plan generated by the query optimizer.

Some problems reported by CHECK TABLE cannot be corrected automatically:

  • Found row where the auto_increment column has the value 0.

    This means that you have a row in the table where the AUTO_INCREMENT index column contains the value 0. (It is possible to create a row where the AUTO_INCREMENT column is 0 by explicitly setting the column to 0 with an UPDATE statement.)

    This is not an error in itself, but could cause trouble if you decide to dump the table and restore it or do an ALTER TABLE on the table. In this case, the AUTO_INCREMENT column changes value according to the rules of AUTO_INCREMENT columns, which could cause problems such as a duplicate-key error.

    To get rid of the warning, simply execute an UPDATE statement to set the column to some value other than 0.

  • If CHECK TABLE finds a problem for an InnoDB table, the server shuts down to prevent error propagation. Details of the error will be written to the error log.

12.4.2.3. CHECKSUM TABLE Syntax

CHECKSUM TABLE tbl_name [, tbl_name] ... [ QUICK | EXTENDED ]

CHECKSUM TABLE reports a table checksum. This statement requires the SELECT privilege for the table.

With QUICK, the live table checksum is reported if it is available, or NULL otherwise. This is very fast. A live checksum is enabled by specifying the CHECKSUM=1 table option when you create the table; currently, this is supported only for MyISAM tables. See Section 12.1.14, “CREATE TABLE Syntax”.

With EXTENDED, the entire table is read row by row and the checksum is calculated. This can be very slow for large tables.

If neither QUICK nor EXTENDED is specified, MySQL returns a live checksum if the table storage engine supports it and scans the table otherwise.

For a nonexistent table, CHECKSUM TABLE returns NULL and generates a warning.

The checksum value depends on the table row format. If the row format changes, the checksum also changes. For example, the storage format for VARCHAR changed between MySQL 4.1 and 5.0, so if a 4.1 table is upgraded to MySQL 5.0, the checksum value may change.

Important

If the checksums for two tables are different, then it is almost certain that the tables are different in some way. However, because the hashing function used by CHECKSUM TABLE is not guaranteed to be collision-free, there is a slight chance that two tables which are not identical can produce the same checksum.

12.4.2.4. OPTIMIZE TABLE Syntax

OPTIMIZE [NO_WRITE_TO_BINLOG | LOCAL] TABLE
    tbl_name [, tbl_name] ...

OPTIMIZE TABLE should be used if you have deleted a large part of a table or if you have made many changes to a table with variable-length rows (tables that have VARCHAR, VARBINARY, BLOB, or TEXT columns). Deleted rows are maintained in a linked list and subsequent INSERT operations reuse old row positions. You can use OPTIMIZE TABLE to reclaim the unused space and to defragment the data file. After extensive changes to a table, this statement may also improve performance of statements that use the table, sometimes significantly.

This statement requires SELECT and INSERT privileges for the table.

OPTIMIZE TABLE is supported for partitioned tables, and you can use ALTER TABLE ... OPTIMIZE PARTITION to optimize one or more partitions; for more information, see Section 12.1.6, “ALTER TABLE Syntax”, and Section 18.3.3, “Maintenance of Partitions”.

OPTIMIZE TABLE works only for MyISAM, InnoDB, and ARCHIVE tables. It does not work for tables created using any other storage engine.

For MyISAM tables, OPTIMIZE TABLE works as follows:

  1. If the table has deleted or split rows, repair the table.

  2. If the index pages are not sorted, sort them.

  3. If the table's statistics are not up to date (and the repair could not be accomplished by sorting the index), update them.

For InnoDB tables, OPTIMIZE TABLE is mapped to ALTER TABLE, which rebuilds the table to update index statistics and free unused space in the clustered index. This is displayed in the output of OPTIMIZE TABLE when you run it on an InnoDB table, as shown here:

mysql> OPTIMIZE TABLE foo;
+----------+----------+----------+-------------------------------------------------------------------+
| Table    | Op       | Msg_type | Msg_text                                                          |
+----------+----------+----------+-------------------------------------------------------------------+
| test.foo | optimize | note     | Table does not support optimize, doing recreate + analyze instead |
| test.foo | optimize | status   | OK                                                                |
+----------+----------+----------+-------------------------------------------------------------------+

You can make OPTIMIZE TABLE work on other storage engines by starting mysqld with the --skip-new or --safe-mode option. In this case, OPTIMIZE TABLE is just mapped to ALTER TABLE.

OPTIMIZE TABLE returns a result set with the following columns.

ColumnValue
TableThe table name
OpAlways optimize
Msg_typestatus, error, info, note, or warning
Msg_textAn informational message

Note that MySQL locks the table during the time OPTIMIZE TABLE is running.

By default, OPTIMIZE TABLE statements are written to the binary log so that they will be replicated to replication slaves. Logging can be suppressed with the optional NO_WRITE_TO_BINLOG keyword or its alias LOCAL.

OPTIMIZE TABLE does not sort R-tree indexes, such as spatial indexes on POINT columns. (Bug#23578)

12.4.2.5. REPAIR TABLE Syntax

REPAIR [NO_WRITE_TO_BINLOG | LOCAL] TABLE
    tbl_name [, tbl_name] ...
    [QUICK] [EXTENDED] [USE_FRM]

REPAIR TABLE repairs a possibly corrupted table. By default, it has the same effect as myisamchk --recover tbl_name. REPAIR TABLE works for MyISAM, ARCHIVE, and CSV tables. See Section 13.5, “The MyISAM Storage Engine”, and Section 13.12, “The ARCHIVE Storage Engine”, and Section 13.13, “The CSV Storage Engine”

This statement requires SELECT and INSERT privileges for the table.

REPAIR TABLE is supported for partitioned tables. However, the USE_FRM option cannot be used with this statement on a partitioned table.

You can use ALTER TABLE ... REPAIR PARTITION to repair one or more partitions; for more information, see Section 12.1.6, “ALTER TABLE Syntax”, and Section 18.3.3, “Maintenance of Partitions”.

Normally, you should never have to run REPAIR TABLE. However, if disaster strikes, this statement is very likely to get back all your data from a MyISAM table. If your tables become corrupted often, you should try to find the reason for it, to eliminate the need to use REPAIR TABLE. See Section C.5.4.2, “What to Do If MySQL Keeps Crashing”, and Section 13.5.4, “MyISAM Table Problems”.

Caution

It is best to make a backup of a table before performing a table repair operation; under some circumstances the operation might cause data loss. Possible causes include but are not limited to file system errors.

Warning

If the server crashes during a REPAIR TABLE operation, it is essential after restarting it that you immediately execute another REPAIR TABLE statement for the table before performing any other operations on it. In the worst case, you might have a new clean index file without information about the data file, and then the next operation you perform could overwrite the data file. This is an unlikely but possible scenario that underscores the value of making a backup first.

REPAIR TABLE returns a result set with the following columns.

ColumnValue
TableThe table name
OpAlways repair
Msg_typestatus, error, info, note, or warning
Msg_textAn informational message

The REPAIR TABLE statement might produce many rows of information for each repaired table. The last row has a Msg_type value of status and Msg_test normally should be OK. If you do not get OK for a MyISAM table, you should try repairing it with myisamchk --safe-recover. (REPAIR TABLE does not implement all the options of myisamchk.) With myisamchk --safe-recover, you can also use options that REPAIR TABLE does not support, such as --max-record-length.

If you use the QUICK option, REPAIR TABLE tries to repair only the index file, and not the data file. This type of repair is like that done by myisamchk --recover --quick.

If you use the EXTENDED option, MySQL creates the index row by row instead of creating one index at a time with sorting. This type of repair is like that done by myisamchk --safe-recover.

The USE_FRM option is available for use if the .MYI index file is missing or if its header is corrupted. This option tells MySQL not to trust the information in the .MYI file header and to re-create it using information from the .frm file. This kind of repair cannot be done with myisamchk.

Note

Use the USE_FRM option only if you cannot use regular REPAIR modes! Telling the server to ignore the .MYI file makes important table metadata stored in the .MYI unavailable to the repair process, which can have deleterious consequences:

  • The current AUTO_INCREMENT value is lost.

  • The link to deleted records in the table is lost, which means that free space for deleted records will remain unoccupied thereafter.

  • The .MYI header indicates whether the table is compressed. If the server ignores this information, it cannot tell that a table is compressed and repair can cause change or loss of table contents. This means that USE_FRM should not be used with compressed tables. That should not be necessary, anyway: Compressed tables are read only, so they should not become corrupt.

Caution

If you use USE_FRM for a table that was created by a different version of the MySQL server than the one you are currently running, REPAIR TABLE will not attempt to repair the table. In this case, the result set returned by REPAIR TABLE contains a line with a Msg_type value of error and a Msg_text value of Failed repairing incompatible .FRM file.

If USE_FRM is not used, REPAIR TABLE checks the table to see whether an upgrade is required. If so, it performs the upgrade, following the same rules as CHECK TABLE ... FOR UPGRADE. See Section 12.4.2.2, “CHECK TABLE Syntax”, for more information. REPAIR TABLE without USE_FRM upgrades the .frm file to the current version.

By default, REPAIR TABLE statements are written to the binary log so that they will be replicated to replication slaves. Logging can be suppressed with the optional NO_WRITE_TO_BINLOG keyword or its alias LOCAL.

Important

In the event that a table on the master becomes corrupted and you run REPAIR TABLE on it, any resulting changes to the original table are not propagated to slaves.

You may be able to increase REPAIR TABLE performance by setting certain system variables. See Section 7.6.4, “Speed of REPAIR TABLE Statements”.

12.4.3. Plugin and User-Defined Function Statements

12.4.3.1. Installing and Uninstalling Plugins

This section compares the various methods of installing and uninstalling server plugins.

INSTALL PLUGIN statement:

  • Loads the plugin code to activate it.

  • Registers the plugin in the mysql.plugin table. Registration causes the server to load the plugin automatically for subsequent restarts. That is, it effects an installation that persists until the plugin is uninstalled explicitly.

  • Requires the INSERT privilege for the mysql.plugin table.

--plugin-load option:

  • Loads the plugin code to activate it.

  • Does not register the plugin in mysql.plugin. For subsequent restarts, the server loads the plugin again only if --plugin-load is given again. That is, this option effects a one-time installation that persists only for one server invocation.

UNINSTALL PLUGIN statement:

  • Unloads the plugin code to deactivate it.

  • Unregisters the plugin from mysql.plugin. The server will not load it automatically for subsequent restarts.

  • Requires the DELETE privilege for the mysql.plugin table.

While a plugin is loaded, information about it is available at runtime from these sources:

If --plugin-load is given at startup for a plugin already installed with INSTALL PLUGIN during a previous server invocation, the server starts but writes these messages to the error log:

100310 19:15:44 [ERROR] Function 'plugin_name' already exists
100310 19:15:44 [Warning] Couldn't load plugin named 'plugin_name'
with soname 'plugin_object_file'.

UNINSTALL PLUGIN can unload plugins regardless of whether they were loaded with INSTALL PLUGIN or --plugin-load. In the latter case, UNINSTALL PLUGIN does not unregister the plugin from mysql.plugin because --plugin-load did not register it.

UNINSTALL PLUGIN cannot unload plugins that are compiled into the server. (These can be identified as those that have a library name of NULL in the output from INFORMATION_SCHEMA.PLUGINS or SHOW PLUGINS.)

If the server is started with the --skip-grant-tables option, it does not consult the mysql.plugins table and thus does not load plugins installed with the INSTALL PLUGIN statement. --plugin-load enables plugins to be loaded even when --skip-grant-tables is given. --plugin-load also enables plugins to be loaded at startup under configurations when plugins cannot be loaded at runtime.

12.4.3.2. CREATE FUNCTION Syntax for User-Defined Functions

CREATE [AGGREGATE] FUNCTION function_name RETURNS {STRING|INTEGER|REAL|DECIMAL}
    SONAME shared_library_name

A user-defined function (UDF) is a way to extend MySQL with a new function that works like a native (built-in) MySQL function such as ABS() or CONCAT().

function_name is the name that should be used in SQL statements to invoke the function. The RETURNS clause indicates the type of the function's return value. DECIMAL is a legal value after RETURNS, but currently DECIMAL functions return string values and should be written like STRING functions.

shared_library_name is the basename of the shared object file that contains the code that implements the function. The file must be located in the plugin directory. This directory is given by the value of the plugin_dir system variable. For more information, see Section 23.3.2.5, “Compiling and Installing User-Defined Functions”.

To create a function, you must have the INSERT privilege for the mysql database. This is necessary because CREATE FUNCTION adds a row to the mysql.func system table that records the function's name, type, and shared library name. If you do not have this table, you should run the mysql_upgrade command to create it. See Section 4.4.7, “mysql_upgrade — Check Tables for MySQL Upgrade”.

An active function is one that has been loaded with CREATE FUNCTION and not removed with DROP FUNCTION. All active functions are reloaded each time the server starts, unless you start mysqld with the --skip-grant-tables option. In this case, UDF initialization is skipped and UDFs are unavailable.

For instructions on writing user-defined functions, see Section 23.3.2, “Adding a New User-Defined Function”. For the UDF mechanism to work, functions must be written in C or C++ (or another language that can use C calling conventions), your operating system must support dynamic loading and you must have compiled mysqld dynamically (not statically).

An AGGREGATE function works exactly like a native MySQL aggregate (summary) function such as SUM or COUNT(). For AGGREGATE to work, your mysql.func table must contain a type column. If your mysql.func table does not have this column, you should run the mysql_upgrade program to create it (see Section 4.4.7, “mysql_upgrade — Check Tables for MySQL Upgrade”).

Note

To upgrade the shared library associated with a UDF, issue a DROP FUNCTION statement, upgrade the shared library, and then issue a CREATE FUNCTION statement. If you upgrade the shared library first and then use DROP FUNCTION, the server may crash.

12.4.3.3. DROP FUNCTION Syntax

DROP FUNCTION function_name

This statement drops the user-defined function (UDF) named function_name.

To drop a function, you must have the DELETE privilege for the mysql database. This is because DROP FUNCTION removes a row from the mysql.func system table that records the function's name, type, and shared library name.

Note

To upgrade the shared library associated with a UDF, issue a DROP FUNCTION statement, upgrade the shared library, and then issue a CREATE FUNCTION statement. If you upgrade the shared library first and then use DROP FUNCTION, the server may crash.

DROP FUNCTION is also used to drop stored functions (see Section 12.1.21, “DROP PROCEDURE and DROP FUNCTION Syntax”).

12.4.3.4. INSTALL PLUGIN Syntax

INSTALL PLUGIN plugin_name SONAME 'shared_library_name'

This statement installs a server plugin. It requires the INSERT privilege for the mysql.plugin table.

plugin_name is the name of the plugin as defined in the plugin descriptor structure contained in the library file (see Section 23.2.4, “Plugin Data Structures and Functions”). Plugin names are not case sensitive. For maximal compatibility, plugin names should be limited to ASCII letters, digits, and underscore because they are used in C source files, shell command lines, M4 and Bourne shell scripts, and SQL environments.

shared_library_name is the name of the shared library that contains the plugin code. The name includes the file name extension (for example, libmyplugin.so, libmyplugin.dll, or libmyplugin.dylib).

The shared library must be located in the plugin directory (the directory named by the plugin_dir system variable). The library must be in the plugin directory itself, not in a subdirectory. By default, plugin_dir is the plugin directory under the directory named by the pkglibdir configuration variable, but it can be changed by setting the value of plugin_dir at server startup. For example, set its value in a my.cnf file:

[mysqld]
plugin_dir=/path/to/plugin/directory

If the value of plugin_dir is a relative path name, it is taken to be relative to the MySQL base directory (the value of the basedir system variable).

INSTALL PLUGIN loads and initializes the plugin code to make the plugin available for use. A plugin is initialized by executing its initialization function, which handles any setup that the plugin must perform before it can be used. When the server shuts down, it executes the deinitialization function for each plugin that is loaded so that the plugin has a change to perform any final cleanup.

INSTALL PLUGIN also registers the plugin by adding a line that indicates the plugin name and library file name to the mysql.plugin table. At server startup, the server loads and initializes any plugin that is listed in the mysql.plugin table. This means that a plugin is installed with INSTALL PLUGIN only once, not every time the server starts. Plugin loading at startup does not occur if the server is started with the --skip-grant-tables option.

A plugin library can contain multiple plugins. For each of them to be installed, use a separate INSTALL PLUGIN statement. Each statement names a different plugin, but all of them specify the same library name.

INSTALL PLUGIN causes the server to read option (my.cnf) files just as during server startup. This enables the plugin to pick up any relevant options from those files. It is possible to add plugin options to an option file even before loading a plugin (if the loose prefix is used). It is also possible to uninstall a plugin, edit my.cnf, and install the plugin again. Restarting the plugin this way enables it to the new option values without a server restart.

For options that control individual plugin loading at server startup, see Section 5.1.3, “Server Options for Loading Plugins”. If you need to load plugins for a single server startup when the --skip-grant-tables option is given (which tells the server not to read system tables), use the --plugin-load option. See Section 5.1.2, “Server Command Options”.

To remove a plugin, use the UNINSTALL PLUGIN statement.

For additional information about plugin loading, see Section 12.4.3.1, “Installing and Uninstalling Plugins”.

To see what plugins are installed, use the SHOW PLUGINS statement or query the INFORMATION_SCHEMA.PLUGINS table.

If you recompile a plugin library and need to reinstall it, you can use either of the following methods:

  • Use UNINSTALL PLUGIN to uninstall all plugins in the library, install the new plugin library file in the plugin directory, and then use INSTALL PLUGIN to install all plugins in the library. This procedure has the advantage that it can be used without stopping the server. However, if the plugin library contains many plugins, you must issue many INSTALL PLUGIN and UNINSTALL PLUGIN statements.

  • Stop the server, install the new plugin library file in the plugin directory, and restart the server.

12.4.3.5. UNINSTALL PLUGIN Syntax

UNINSTALL PLUGIN plugin_name

This statement removes an installed server plugin. It requires the DELETE privilege for the mysql.plugin table.

plugin_name must be the name of some plugin that is listed in the mysql.plugin table. The server executes the plugin's deinitialization function and removes the row for the plugin from the mysql.plugin table, so that subsequent server restarts will not load and initialize the plugin. UNINSTALL PLUGIN does not remove the plugin's shared library file.

You cannot uninstall a plugin if any table that uses it is open.

Plugin removal has implications for the use of associated tables. For example, if a full-text parser plugin is associated with a FULLTEXT index on the table, uninstalling the plugin makes the table unusable. Any attempt to access the table results in an error. The table cannot even be opened, so you cannot drop an index for which the plugin is used. This means that uninstalling a plugin is something to do with care unless you do not care about the table contents. If you are uninstalling a plugin with no intention of reinstalling it later and you care about the table contents, you should dump the table with mysqldump and remove the WITH PARSER clause from the dumped CREATE TABLE statement so that you can reload the table later. If you do not care about the table, DROP TABLE can be used even if any plugins associated with the table are missing.

For additional information about plugin loading, see Section 12.4.3.1, “Installing and Uninstalling Plugins”.

12.4.4. SET Syntax

SET variable_assignment [, variable_assignment] ...

variable_assignment:
      user_var_name = expr
    | [GLOBAL | SESSION] system_var_name = expr
    | [@@global. | @@session. | @@]system_var_name = expr

The SET statement assigns values to different types of variables that affect the operation of the server or your client. Older versions of MySQL employed SET OPTION, but this syntax is deprecated in favor of SET without OPTION.

This section describes use of SET for assigning values to system variables or user variables. For general information about these types of variables, see Section 5.1.4, “Server System Variables”, and Section 8.4, “User-Defined Variables”. System variables also can be set at server startup, as described in Section 5.1.5, “Using System Variables”.

Some variants of SET syntax are used in other contexts:

The following discussion shows the different SET syntaxes that you can use to set variables. The examples use the = assignment operator, but you can also use the := assignment operator for this purpose.

A user variable is written as @var_name and can be set as follows:

SET @var_name = expr;

Many system variables are dynamic and can be changed while the server runs by using the SET statement. For a list, see Section 5.1.5.2, “Dynamic System Variables”. To change a system variable with SET, refer to it as var_name, optionally preceded by a modifier:

  • To indicate explicitly that a variable is a global variable, precede its name by GLOBAL or @@global.. The SUPER privilege is required to set global variables.

  • To indicate explicitly that a variable is a session variable, precede its name by SESSION, @@session., or @@. Setting a session variable requires no special privilege, but a client can change only its own session variables, not those of any other client.

  • LOCAL and @@local. are synonyms for SESSION and @@session..

  • If no modifier is present, SET changes the session variable.

A SET statement can contain multiple variable assignments, separated by commas. If you set several system variables, the most recent GLOBAL or SESSION modifier in the statement is used for following variables that have no modifier specified.

Examples:

SET sort_buffer_size=10000;
SET @@local.sort_buffer_size=10000;
SET GLOBAL sort_buffer_size=1000000, SESSION sort_buffer_size=1000000;
SET @@sort_buffer_size=1000000;
SET @@global.sort_buffer_size=1000000, @@local.sort_buffer_size=1000000;

The @@var_name syntax for system variables is supported for compatibility with some other database systems.

If you change a session system variable, the value remains in effect until your session ends or until you change the variable to a different value. The change is not visible to other clients.

If you change a global system variable, the value is remembered and used for new connections until the server restarts. (To make a global system variable setting permanent, you should set it in an option file.) The change is visible to any client that accesses that global variable. However, the change affects the corresponding session variable only for clients that connect after the change. The global variable change does not affect the session variable for any client that is currently connected (not even that of the client that issues the SET GLOBAL statement).

To prevent incorrect usage, MySQL produces an error if you use SET GLOBAL with a variable that can only be used with SET SESSION or if you do not specify GLOBAL (or @@global.) when setting a global variable.

To set a SESSION variable to the GLOBAL value or a GLOBAL value to the compiled-in MySQL default value, use the DEFAULT keyword. For example, the following two statements are identical in setting the session value of max_join_size to the global value:

SET max_join_size=DEFAULT;
SET @@session.max_join_size=@@global.max_join_size;

Not all system variables can be set to DEFAULT. In such cases, use of DEFAULT results in an error.

You can refer to the values of specific global or sesson system variables in expressions by using one of the @@-modifiers. For example, you can retrieve values in a SELECT statement like this:

SELECT @@global.sql_mode, @@session.sql_mode, @@sql_mode;

When you refer to a system variable in an expression as @@var_name (that is, when you do not specify @@global. or @@session.), MySQL returns the session value if it exists and the global value otherwise. (This differs from SET @@var_name = value, which always refers to the session value.)

Note

Some variables displayed by SHOW VARIABLES may not be available using SELECT @@var_name syntax; an Unknown system variable occurs. As a workaround in such cases, you can use SHOW VARIABLES LIKE 'var_name'.

Suffixes for specifying a value multiplier can be used when setting a variable at server startup, but not to set the value with SET at runtime. On the other hand, with SET you can assign a variable's value using an expression, which is not true when you set a variable at server startup. For example, the first of the following lines is legal at server startup, but the second is not:

shell> mysql --max_allowed_packet=16M
shell> mysql --max_allowed_packet=16*1024*1024

Conversely, the second of the following lines is legal at runtime, but the first is not:

mysql> SET GLOBAL max_allowed_packet=16M;
mysql> SET GLOBAL max_allowed_packet=16*1024*1024;

To display system variables names and values, use the SHOW VARIABLES statement. (See Section 12.4.5.40, “SHOW VARIABLES Syntax”.)

The following list describes SET options that have nonstandard syntax (that is, options that are not set with name = value syntax).