死锁案例三-CFANZ编程社区

1、环境说明

MySQL5.6.33，隔离级别是RR。表结构及数据：

create table t2 (
  id int primary key auto_increment ,
  c1 int ,
  c2 int ,
  key (c1)
) engine=innodb ;

insert into t2 values(24,3,4),(25,3,4),(26,3,4),(30,5,8);

2、测试用例

session1	session2
begin;	begin;
select *from t2 where id=30 for update;
	update t2 set c2=8 where c1=5;
delete from t2 where id=30;
	deadlock

3、死锁日志

LATEST DETECTED DEADLOCK
------------------------
2018-07-14 06:45:32 a3522b90
*** (1) TRANSACTION:
TRANSACTION 8976, ACTIVE 104 sec starting index read
mysql tables in use 1, locked 1
LOCK WAIT 3 lock struct(s), heap size 320, 2 row lock(s)
MySQL thread id 2, OS thread handle 0xa34f1b90, query id 28 localhost root Searching rows for update
update t2 set c2=8 where c1=5
*** (1) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 19 page no 3 n bits 72 index `PRIMARY` of table `yzs`.`t2` trx id 8976 lock_mode X locks rec but not gap waiting
Record lock, heap no 5 PHYSICAL RECORD: n_fields 5; compact format; info bits 32
 0: len 4; hex 8000001e; asc     ;;
 1: len 6; hex 00000000230f; asc     # ;;
 2: len 7; hex 0d000001410943; asc     A C;;
 3: len 4; hex 80000005; asc     ;;
 4: len 4; hex 80000008; asc     ;;

*** (2) TRANSACTION:
TRANSACTION 8975, ACTIVE 251 sec updating or deleting
mysql tables in use 1, locked 1
3 lock struct(s), heap size 320, 2 row lock(s), undo log entries 1
MySQL thread id 1, OS thread handle 0xa3522b90, query id 29 localhost root updating
delete from t2 where id=30
*** (2) HOLDS THE LOCK(S):
RECORD LOCKS space id 19 page no 3 n bits 72 index `PRIMARY` of table `yzs`.`t2` trx id 8975 lock_mode X locks rec but not gap
Record lock, heap no 5 PHYSICAL RECORD: n_fields 5; compact format; info bits 32
 0: len 4; hex 8000001e; asc     ;;
 1: len 6; hex 00000000230f; asc     # ;;
 2: len 7; hex 0d000001410943; asc     A C;;
 3: len 4; hex 80000005; asc     ;;
 4: len 4; hex 80000008; asc     ;;

*** (2) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 19 page no 4 n bits 72 index `c1` of table `yzs`.`t2` trx id 8975 lock_mode X locks rec but not gap waiting
Record lock, heap no 5 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
 0: len 4; hex 80000005; asc     ;;
 1: len 4; hex 8000001e; asc     ;;

*** WE ROLL BACK TRANSACTION (1)

4、分析死锁日志

TRANSACTION 8975：

执行完delete from t2 where id=30后，拥有的锁是聚集索引记录(30,5,8)的X类型的记录锁lock_mode X locks rec but not gap，申请等待的锁是二级索引记录(5,30)的X类型的记录锁lock_mode X locks rec but not gap waiting。

TRANSACTION 8976：

执行完update t2 set c2=8 where c1=5语句时，显示申请等待的锁是聚集索引记录(30,5,8)的X类型的记录锁lock_mode X locks rec but not gap waiting。

只从死锁日志上看不出造成死锁的原因。未提交的事务中，有已执行过但未提交的SQL语句锁造成拥有的锁没有显示出来。

5、加锁原理

6、解析

session1执行完select *from t2 where id=30 for update;后，搜索到主键30，即对(30,5,8)加上X类型的记录锁

session2执行完update t2 set c2=8 where c1=5;后，在search阶段先搜索二级索引，搜索到(5,30)，对此记录加X类型的next-key锁 lock_mode X；然后搜索聚集索引记录30，需要对聚集索引记录(30,5,8)加X类型记录锁，因为和session1互斥，所以锁等待。

session1再次执行delete from t2 where id=30;，对记录锁聚集索引记录(30,5,8)申请X类型的记录锁，因为事务前面已申请过，所以不再申请；再对二级索引记录(5,30)加X类型的记录锁，和session2互斥。

这样彼此发生锁等待，造成死锁。

7、解决方法

杨奇龙兄建议可以将session2的update t2 set c2=8 where c1=5改成按主键更新 update tx set c2=x where id=30。这样加锁顺序都是先聚集再二级索引，避免主键和二级索引加锁顺序交叉造成死锁。

8、参考

http://blog.itpub.net/22664653/viewspace-2152274/