昨天同事遇到的一个案例,这里简单描述一下:一个表里面有一个bit类型的字段,同事在优化相关SQL的过程中,给这个表的bit类型的字段新增了一个索引,然后测试验证
时,居然发现SQL语句执行结果跟不加索引不一样。加了索引后,SQL语句没有查询出一条记录,删除索引后,SQL语句就能查询出几十条记录。下面我们构造一个简单
的例子,重现一下这个案例
我们先创建表student_attend,初始化一些数据。这篇文章的测试环境为MySQL 8.0.35社区版。
CREATE TABLE `student_attend` (
`id` int NOT NULL AUTO_INCREMENT COMMENT '自增编号',
`std_id` int DEFAULT NULL COMMENT '学号',
`class_id` int DEFAULT NULL COMMENT '课程编号',
`is_attend` bit(1) DEFAULT b'1' COMMENT '是否缺陷考勤',
PRIMARY KEY (`id`)
) ENGINE=InnoDB;
insert into student_attend(std_id, class_id, is_attend)
select 1001, 1, 1 from dual union all
select 1001, 2, 0 from dual union all
select 1001, 3, 1 from dual union all
select 1001, 4, 1 from dual union all
select 1001, 5, 1 from dual union all
select 1001, 6, 0 from dual union all
select 1002, 1, 1 from dual union all
select 1002, 2, 1 from dual union all
select 1003, 1, 0 from dual union all
select 1003, 2, 0 from dual;
如下所示,假设我们要查询is_attend=1的所有学生信息,那么可以有下面三种写法
mysql> select * from student_attend where is_attend=1;
+----+--------+----------+----------------------+
| id | std_id | class_id | is_attend |
+----+--------+----------+----------------------+
| 1 | 1001 | 1 | 0x01 |
| 3 | 1001 | 3 | 0x01 |
| 4 | 1001 | 4 | 0x01 |
| 5 | 1001 | 5 | 0x01 |
| 7 | 1002 | 1 | 0x01 |
| 8 | 1002 | 2 | 0x01 |
+----+--------+----------+----------------------+
6 rows in set (0.00 sec)
mysql> select * from student_attend where is_attend=b'1';
+----+--------+----------+----------------------+
| id | std_id | class_id | is_attend |
+----+--------+----------+----------------------+
| 1 | 1001 | 1 | 0x01 |
| 3 | 1001 | 3 | 0x01 |
| 4 | 1001 | 4 | 0x01 |
| 5 | 1001 | 5 | 0x01 |
| 7 | 1002 | 1 | 0x01 |
| 8 | 1002 | 2 | 0x01 |
+----+--------+----------+----------------------+
6 rows in set (0.00 sec)
#遇到问题的SQL写法
mysql> select * from student_attend where is_attend='1';
+----+--------+----------+----------------------+
| id | std_id | class_id | is_attend |
+----+--------+----------+----------------------+
| 1 | 1001 | 1 | 0x01 |
| 3 | 1001 | 3 | 0x01 |
| 4 | 1001 | 4 | 0x01 |
| 5 | 1001 | 5 | 0x01 |
| 7 | 1002 | 1 | 0x01 |
| 8 | 1002 | 2 | 0x01 |
+----+--------+----------+----------------------+
6 rows in set (0.00 sec)
mysql>
接下来,我们在字段is_attend上创建索引ix_student_attend_n1,如下所示
create index ix_student_attend_n1 on student_attend(is_attend);
然后我们继续测试验证,就能出现我前文所说的情况,如需所示,最后一个SQL,它的返回记录数为0.
mysql> select * from student_attend where is_attend=1;
+----+--------+----------+----------------------+
| id | std_id | class_id | is_attend |
+----+--------+----------+----------------------+
| 1 | 1001 | 1 | 0x01 |
| 3 | 1001 | 3 | 0x01 |
| 4 | 1001 | 4 | 0x01 |
| 5 | 1001 | 5 | 0x01 |
| 7 | 1002 | 1 | 0x01 |
| 8 | 1002 | 2 | 0x01 |
+----+--------+----------+----------------------+
6 rows in set (0.00 sec)
mysql> select * from student_attend where is_attend=b'1';
+----+--------+----------+----------------------+
| id | std_id | class_id | is_attend |
+----+--------+----------+----------------------+
| 1 | 1001 | 1 | 0x01 |
| 3 | 1001 | 3 | 0x01 |
| 4 | 1001 | 4 | 0x01 |
| 5 | 1001 | 5 | 0x01 |
| 7 | 1002 | 1 | 0x01 |
| 8 | 1002 | 2 | 0x01 |
+----+--------+----------+----------------------+
6 rows in set (0.00 sec)
mysql> select * from student_attend where is_attend='1';
Empty set (0.00 sec)
mysql>
其实第一次见到这种情况的时候,我还是有点震惊的,因为在我的观念中,索引只会影响执行计划,不会影响查询结果,但是现在的情况是
索引的存在影响了SQL的查询结果。那么为什么会出现这种情况呢?
首先看了一下执行计划,如下所示,从执行计划看,它既没有走全表扫描也没有走索引,仅仅有"message": "no matching row in const table"提示,如果仅仅分析
执行计划,我们得不到更多的有用信息
mysql> explain
-> select * from student_attend where is_attend='1';
+----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+--------------------------------+
| id | select_type | table | partitions | type | possible_keys | key | key_len | ref | rows | filtered | Extra |
+----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+--------------------------------+
| 1 | SIMPLE | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | NULL | no matching row in const table |
+----+-------------+-------+------------+------+---------------+------+---------+------+------+----------+--------------------------------+
1 row in set, 1 warning (0.00 sec)
mysql>
mysql> explain format=json
-> select * from student_attend where is_attend='1'\G
*************************** 1. row ***************************
EXPLAIN: {
"query_block": {
"select_id": 1,
"message": "no matching row in const table"
} /* query_block */
}
1 row in set, 1 warning (0.00 sec)
mysql> show warnings\G
*************************** 1. row ***************************
Level: Note
Code: 1003
Message: /* select#1 */ select `kerry`.`student_attend`.`id` AS `id`,`kerry`.`student_attend`.`std_id` AS `std_id`,`kerry`.`student_attend`.`class_id` AS `class_id`,`kerry`.`student_attend`.`is_attend` AS `is_attend` from `kerry`.`student_attend` where (`kerry`.`student_attend`.`is_attend` = '1')
1 row in set (0.00 sec)
mysql>
那么我们使用trace跟踪分析一下优化器如何选择执行计划。看看其详细执行过程,如下所示
mysql> SET OPTIMIZER_TRACE="enabled=on",END_MARKERS_IN_JSON=on;
Query OK, 0 rows affected (0.00 sec)
mysql> SET OPTIMIZER_TRACE_MAX_MEM_SIZE=1000000;
Query OK, 0 rows affected (0.00 sec)
mysql> select * from student_attend where is_attend='1';
Empty set (0.00 sec)
mysql> SELECT * FROM INFORMATION_SCHEMA.OPTIMIZER_TRACE \G;
*************************** 1. row ***************************
QUERY: select * from student_attend where is_attend='1'
TRACE: {
"steps": [
{
"join_preparation": {
"select#": 1,
"steps": [
{
"expanded_query": "/* select#1 */ select `student_attend`.`id` AS `id`,`student_attend`.`std_id` AS `std_id`,`student_attend`.`class_id` AS `class_id`,`student_attend`.`is_attend` AS `is_attend` from `student_attend` where (`student_attend`.`is_attend` = '1')"
}
] /* steps */
} /* join_preparation */
},
{
"join_optimization": {
"select#": 1,
"steps": [
{
"condition_processing": {
"condition": "WHERE",
"original_condition": "(`student_attend`.`is_attend` = '1')",
"steps": [
{
"transformation": "equality_propagation",
"resulting_condition": "(`student_attend`.`is_attend` = '1')"
},
{
"transformation": "constant_propagation",
"resulting_condition": "(`student_attend`.`is_attend` = '1')"
},
{
"transformation": "trivial_condition_removal",
"resulting_condition": "(`student_attend`.`is_attend` = '1')"
}
] /* steps */
} /* condition_processing */
},
{
"substitute_generated_columns": {
} /* substitute_generated_columns */
},
{
"table_dependencies": [
{
"table": "`student_attend`",
"row_may_be_null": false,
"map_bit": 0,
"depends_on_map_bits": [
] /* depends_on_map_bits */
}
] /* table_dependencies */
},
{
"ref_optimizer_key_uses": [
{
"table": "`student_attend`",
"field": "is_attend",
"equals": "'1'",
"null_rejecting": true
}
] /* ref_optimizer_key_uses */
},
{
"rows_estimation": [
{
"table": "`student_attend`",
"range_analysis": {
"table_scan": {
"rows": 10,
"cost": 3.35
} /* table_scan */,
"potential_range_indexes": [
{
"index": "PRIMARY",
"usable": false,
"cause": "not_applicable"
},
{
"index": "ix_student_attend_n1",
"usable": true,
"key_parts": [
"is_attend",
"id"
] /* key_parts */
}
] /* potential_range_indexes */,
"setup_range_conditions": [
{
"impossible_condition": {
"cause": "value_out_of_range"
} /* impossible_condition */
}
] /* setup_range_conditions */,
"impossible_range": true
} /* range_analysis */,
"rows": 0,
"cause": "impossible_where_condition"
}
] /* rows_estimation */
}
] /* steps */,
"empty_result": {
"cause": "no matching row in const table"
} /* empty_result */
} /* join_optimization */
},
{
"join_execution": {
"select#": 1,
"steps": [
] /* steps */
} /* join_execution */
}
] /* steps */
}
MISSING_BYTES_BEYOND_MAX_MEM_SIZE: 0
INSUFFICIENT_PRIVILEGES: 0
1 row in set (0.00 sec)
ERROR:
No query specified
mysql>
mysql> SET optimizer_trace="enabled=off";
Query OK, 0 rows affected (0.01 sec)
从trace的详细信息看,这个过程中发生了隐式转换:下面这个过程就是发生了类型转换
由于发生类型转换过程中(字符串转换为bit类型)遇到了数据截断错误(从value_out_of_range等信息就可以看出),如下截图所示
而优化器应该是根据一定的逻辑判断,得到这个值不存在索引中,从而就判断没有匹配的记录,直接返回空的结果集了,根本不去走扫描全表或走索引查找等操作。
"empty_result": {
"cause": "no matching row in const table"
} /* empty_result */
当然这里仅仅是根据trace的信息做的一个判断,如有错误或不谨慎的地方,敬请谅解。毕竟没有深入分析过源码。
那么为什么没有索引的话,SQL语句的结果就是正确的呢? 难道没有发生类型转换吗? 难道没有发生数据截断错误吗?那么我们就继续trace跟踪分析看看,如下所示
mysql> drop index ix_student_attend_n1 on student_attend;
Query OK, 0 rows affected (0.03 sec)
Records: 0 Duplicates: 0 Warnings: 0
mysql> SET OPTIMIZER_TRACE="enabled=on",END_MARKERS_IN_JSON=on;
Query OK, 0 rows affected (0.00 sec)
mysql> SET OPTIMIZER_TRACE_MAX_MEM_SIZE=1000000;
Query OK, 0 rows affected (0.00 sec)
mysql> select * from student_attend where is_attend='1';
+----+--------+----------+----------------------+
| id | std_id | class_id | is_attend |
+----+--------+----------+----------------------+
| 1 | 1001 | 1 | 0x01 |
| 3 | 1001 | 3 | 0x01 |
| 4 | 1001 | 4 | 0x01 |
| 5 | 1001 | 5 | 0x01 |
| 7 | 1002 | 1 | 0x01 |
| 8 | 1002 | 2 | 0x01 |
+----+--------+----------+----------------------+
6 rows in set (0.00 sec)
mysql> SELECT * FROM INFORMATION_SCHEMA.OPTIMIZER_TRACE \G;
*************************** 1. row ***************************
QUERY: select * from student_attend where is_attend='1'
TRACE: {
"steps": [
{
"join_preparation": {
"select#": 1,
"steps": [
{
"expanded_query": "/* select#1 */ select `student_attend`.`id` AS `id`,`student_attend`.`std_id` AS `std_id`,`student_attend`.`class_id` AS `class_id`,`student_attend`.`is_attend` AS `is_attend` from `student_attend` where (`student_attend`.`is_attend` = '1')"
}
] /* steps */
} /* join_preparation */
},
{
"join_optimization": {
"select#": 1,
"steps": [
{
"condition_processing": {
"condition": "WHERE",
"original_condition": "(`student_attend`.`is_attend` = '1')",
"steps": [
{
"transformation": "equality_propagation",
"resulting_condition": "(`student_attend`.`is_attend` = '1')"
},
{
"transformation": "constant_propagation",
"resulting_condition": "(`student_attend`.`is_attend` = '1')"
},
{
"transformation": "trivial_condition_removal",
"resulting_condition": "(`student_attend`.`is_attend` = '1')"
}
] /* steps */
} /* condition_processing */
},
{
"substitute_generated_columns": {
} /* substitute_generated_columns */
},
{
"table_dependencies": [
{
"table": "`student_attend`",
"row_may_be_null": false,
"map_bit": 0,
"depends_on_map_bits": [
] /* depends_on_map_bits */
}
] /* table_dependencies */
},
{
"ref_optimizer_key_uses": [
] /* ref_optimizer_key_uses */
},
{
"rows_estimation": [
{
"table": "`student_attend`",
"table_scan": {
"rows": 10,
"cost": 0.25
} /* table_scan */
}
] /* rows_estimation */
},
{
"considered_execution_plans": [
{
"plan_prefix": [
] /* plan_prefix */,
"table": "`student_attend`",
"best_access_path": {
"considered_access_paths": [
{
"rows_to_scan": 10,
"access_type": "scan",
"resulting_rows": 10,
"cost": 1.25,
"chosen": true
}
] /* considered_access_paths */
} /* best_access_path */,
"condition_filtering_pct": 100,
"rows_for_plan": 10,
"cost_for_plan": 1.25,
"chosen": true
}
] /* considered_execution_plans */
},
{
"attaching_conditions_to_tables": {
"original_condition": "(`student_attend`.`is_attend` = '1')",
"attached_conditions_computation": [
] /* attached_conditions_computation */,
"attached_conditions_summary": [
{
"table": "`student_attend`",
"attached": "(`student_attend`.`is_attend` = '1')"
}
] /* attached_conditions_summary */
} /* attaching_conditions_to_tables */
},
{
"finalizing_table_conditions": [
{
"table": "`student_attend`",
"original_table_condition": "(`student_attend`.`is_attend` = '1')",
"final_table_condition ": "(`student_attend`.`is_attend` = '1')"
}
] /* finalizing_table_conditions */
},
{
"refine_plan": [
{
"table": "`student_attend`"
}
] /* refine_plan */
}
] /* steps */
} /* join_optimization */
},
{
"join_execution": {
"select#": 1,
"steps": [
] /* steps */
} /* join_execution */
}
] /* steps */
}
MISSING_BYTES_BEYOND_MAX_MEM_SIZE: 0
INSUFFICIENT_PRIVILEGES: 0
1 row in set (0.00 sec)
ERROR:
No query specified
mysql> SET optimizer_trace="enabled=off";
Query OK, 0 rows affected (0.00 sec)
从上面trace信息来看,似乎执行计划先进行全表扫描,然后过滤记录,输出信息里面没有value_out_of_range这类信息,似乎没有发生数据截断。具体步骤跟之前的trace信息有很大不同。具体只看到了下面这些信息,但是更多信息我也看不出来。不清楚底层到底做了啥。
小结
关于bit类型的字段,我们写SQL的时候,不要使用字符串,避免发生隐式类型转换。正确的写法应该是下面这种方式
select * from student_attend where is_attend=b'1';
或
select * from student_attend where is_attend=1;
DBA在给bit类型创建索引的时候也必须谨慎处理,跟开发和Support人员多协商沟通,告知他们可能出现这种情况,因为你可能没法控制开发人员写出这样的SQL。