MySQL表排序规则不同错误问题分析
MySQL多表join时报错如下：[Err]1267–Illegalmixofcollations(utf8_general_ci,IMPLICIT)and(utf8_unicode_ci,IMPLICIT)foroperation‘=
就是说两个表的排序规则（COLLATION）不同，无法完成比较。COLLATION是用在排序，大小比较上，一个字符集有一个或多种COLLATION，并且以_ci（大小写不敏感）、_cs（大小写敏感）或_bin（二进制）结束。在做比较时，应该确保两个表的字符排序相同。一般建表的时候不指定，可以走默认的，全是默认的就没什么问题了。
下面来模拟一下各种场景，表结构如下（utf8默认排序规则为utf8_general_ci）：
mysql>showcreatetabletest.cs\G ***************************1.row*************************** Table:cs CreateTable:CREATETABLE`cs`( `id`int(11)DEFAULTNULL, `name`varchar(10)DEFAULTNULL )ENGINE=InnoDBDEFAULTCHARSET=utf8 1rowinset(0.01sec)
查看表默认排序规则集
mysql>selectTABLE_SCHEMA,TABLE_NAME,TABLE_COLLATIONfrominformation_schema.tableswheretable_name='cs'; +--------------+------------+-----------------+ |TABLE_SCHEMA|TABLE_NAME|TABLE_COLLATION| +--------------+------------+-----------------+ |test|cs|utf8_general_ci| +--------------+------------+-----------------+ 1rowinset(0.00sec)
查看列排序规则集
mysql>selectTABLE_SCHEMA,TABLE_NAME,COLUMN_NAME,COLLATION_NAMEfrominformation_schema.COLUMNSwhereTABLE_NAME='cs'; +--------------+------------+-------------+-----------------+ |TABLE_SCHEMA|TABLE_NAME|COLUMN_NAME|COLLATION_NAME| +--------------+------------+-------------+-----------------+ |test|cs|id|NULL| |test|cs|name|utf8_general_ci| +--------------+------------+-------------+-----------------+ 2rowsinset(0.00sec)
从utf8升级为utf8mb4是不支持onlineddl的，如下：
mysql>ALTERTABLEcsCONVERTTOCHARACTERSETutf8mb4,ALGORITHM=INPLACE,LOCK=NONE; ERROR1846(0A000):ALGORITHM=INPLACEisnotsupported.Reason:CannotchangecolumntypeINPLACE.TryALGORITHM=COPY.
从utf8.utf8_general_ci变更为utf8.utf8_unicode_ci是不支持onlineddl的，如下：
mysql>ALTERTABLEcsCONVERTTOCHARACTERSETutf8collateutf8_unicode_ci,ALGORITHM=INPLACE,LOCK=NONE; ERROR1846(0A000):ALGORITHM=INPLACEisnotsupported.Reason:CannotchangecolumntypeINPLACE.TryALGORITHM=COPY.
如果使用下面这种方式修改字符集，你会发现，只更改了表级的，没有更改列级的。
mysql>ALTERTABLEcsCHARACTERSETutf8collateutf8_unicode_ci; QueryOK,0rowsaffected(0.01sec) Records:0Duplicates:0Warnings:0 mysql>selectTABLE_SCHEMA,TABLE_NAME,TABLE_COLLATIONfrominformation_schema.tableswheretable_name='cs'; +--------------+------------+-----------------+ |TABLE_SCHEMA|TABLE_NAME|TABLE_COLLATION| +--------------+------------+-----------------+ |test|cs|utf8_unicode_ci| +--------------+------------+-----------------+ 1rowinset(0.00sec) mysql>selectTABLE_SCHEMA,TABLE_NAME,COLUMN_NAME,COLLATION_NAMEfrominformation_schema.COLUMNSwhereTABLE_NAME='cs'; +--------------+------------+-------------+-----------------+ |TABLE_SCHEMA|TABLE_NAME|COLUMN_NAME|COLLATION_NAME| +--------------+------------+-------------+-----------------+ |test|cs|id|NULL| |test|cs|name|utf8_general_ci| +--------------+------------+-------------+-----------------+ 2rowsinset(0.00sec)
所以真正改字符集的时候别忘了加上CONVERTTO，如下：
mysql>ALTERTABLEcsCONVERTTOCHARACTERSETutf8collateutf8_unicode_ci; QueryOK,5rowsaffected(0.06sec) Records:5Duplicates:0Warnings:0 mysql>selectTABLE_SCHEMA,TABLE_NAME,COLUMN_NAME,COLLATION_NAMEfrominformation_schema.COLUMNSwhereTABLE_NAME='cs'; +--------------+------------+-------------+-----------------+ |TABLE_SCHEMA|TABLE_NAME|COLUMN_NAME|COLLATION_NAME| +--------------+------------+-------------+-----------------+ |test|cs|id|NULL| |test|cs|name|utf8_unicode_ci| +--------------+------------+-------------+-----------------+ 2rowsinset(0.00sec)
要仅仅改变一个表的默认字符集，应使用此语句：
mysql>ALTERTABLEcsdefaultCHARACTERSETutf8collateutf8_general_ci,ALGORITHM=INPLACE,LOCK=NONE; QueryOK,0rowsaffected(0.00sec) Records:0Duplicates:0Warnings:0 mysql>selectTABLE_SCHEMA,TABLE_NAME,TABLE_COLLATIONfrominformation_schema.tableswheretable_name='cs'; +--------------+------------+-----------------+ |TABLE_SCHEMA|TABLE_NAME|TABLE_COLLATION| +--------------+------------+-----------------+ |test|cs|utf8_general_ci| +--------------+------------+-----------------+ 1rowinset(0.00sec) mysql>selectTABLE_SCHEMA,TABLE_NAME,COLUMN_NAME,COLLATION_NAMEfrominformation_schema.COLUMNSwhereTABLE_NAME='cs'; +--------------+------------+-------------+-----------------+ |TABLE_SCHEMA|TABLE_NAME|COLUMN_NAME|COLLATION_NAME| +--------------+------------+-------------+-----------------+ |test|cs|id|NULL| |test|cs|name|utf8_unicode_ci| +--------------+------------+-------------+-----------------+ 2rowsinset(0.00sec)
可以发现列字符集没有改变，并且只有新的列才会默认继承表的字符集（utf8.utf8_general_ci）。
总结
以上就是本文关于MySQL表排序规则不同错误问题分析的全部内容，希望对大家有所帮助。感兴趣的朋友可以参阅：几个比较重要的MySQL变量、MySQL声明变量及存储过程分析、MySQL主库binlog(master-log)与从库relay-log关系代码详解、MySQLprepare原理详解等，有什么问题可以随时留言，互相交流，共同进步。