有趣的insert死锁
昨天看到一个很有意思的死锁，拿来记录下：
环境:deadlock on
事务隔离级别: read commited
表结构：
root@test 08:34:01>show create table lingluo\G *************************** 1. row *************************** Table: lingluo Create Table: CREATE TABLE `lingluo` ( `a` int(11) NOT NULL DEFAULT '0', `b` int(11) DEFAULT NULL, `c` int(11) DEFAULT NULL, `d` int(11) DEFAULT NULL, PRIMARY KEY (`a`), UNIQUE KEY `uk_bc` (`b`,`c`) ) ENGINE=InnoDB DEFAULT CHARSET=gbk 1 row in set (0.00 sec)
session 1:
root@test 08:45:51>select * from lingluo; +--------+------+------+------+ | a | b | c | d | +--------+------+------+------+ | 1 | 2 | 3 | 4 | | 500 | 100 | 200 | 43 | | 1000 | 10 | 20 | 43 | | 10001 | 21 | 21 | 32 | | 100202 | 213 | 213 | 312 | | 100212 | 214 | 214 | 312 | +--------+------+------+------+ 6 rows in set (0.00 sec) root@test 08:46:38>begin; Query OK, 0 rows affected (0.00 sec) root@test 08:47:04>insert into lingluo values(100213,215,215,312); Query OK, 1 row affected (0.00 sec)
session 2:
root@test 08:46:02>begin; Query OK, 0 rows affected (0.00 sec) root@test 08:47:20>insert into lingluo values(100214,215,215,312); Query OK, 1 row affected (12.77 sec)
session3:
root@test 08:46:24>begin; Query OK, 0 rows affected (0.00 sec) root@test 08:47:23>insert into lingluo values(100215,215,215,312);
session 1 rollback前:
---TRANSACTION 4F3D6F33, ACTIVE 3 sec inserting mysql tables in use 1, locked 1 LOCK WAIT 2 lock struct(s), heap size 376, 1 row lock(s), undo log entries 1 MySQL thread id 18124715, OS thread handle 0x7fea34912700, query id 1435660081 localhost root update insert into lingluo values(100215,215,215,312) ------- TRX HAS BEEN WAITING 3 SEC FOR THIS LOCK TO BE GRANTED: RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6F33 lock mode S waiting ------------------ TABLE LOCK table `test`.`lingluo` trx id 4F3D6F33 lock mode IX RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6F33 lock mode S waiting ---TRANSACTION 4F3D6D24, ACTIVE 5 sec inserting mysql tables in use 1, locked 1 LOCK WAIT 2 lock struct(s), heap size 376, 1 row lock(s), undo log entries 1 MySQL thread id 18124702, OS thread handle 0x7fe706fdf700, query id 1435659684 localhost root update insert into lingluo values(100214,215,215,312) ------- TRX HAS BEEN WAITING 5 SEC FOR THIS LOCK TO BE GRANTED: RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6D24 lock mode S waiting ------------------ TABLE LOCK table `test`.`lingluo` trx id 4F3D6D24 lock mode IX RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6D24 lock mode S waiting ---TRANSACTION 4F3D4423, ACTIVE 33 sec 2 lock struct(s), heap size 376, 1 row lock(s), undo log entries 1 MySQL thread id 18124692, OS thread handle 0x7fe73c89a700, query id 1435651549 localhost root TABLE LOCK table `test`.`lingluo` trx id 4F3D4423 lock mode IX RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D4423 lock_mode X locks rec but not gap
/****
session 1上的转为显式锁：lock_mode X locks rec but not gap
session 2等待的锁：lock mode S waiting
session 3等待的锁：lock mode S waiting
***/
session 1 rollback
session 2插入成功
session 3:
ERROR 1213 (40001): Deadlock found when trying to get lock; try restarting transaction
这个时候show engine innodb status:
5 lock struct(s), heap size 1248, 3 row lock(s), undo log entries 1 MySQL thread id 18124702, OS thread handle 0x7fe706fdf700, query id 1435659684 localhost root TABLE LOCK table `test`.`lingluo` trx id 4F3D6D24 lock mode IX RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6D24 lock mode S RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6D24 lock mode S RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6D24 lock_mode X insert intention RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6D24 lock mode S locks gap before rec
死锁信息：
------------------------ LATEST DETECTED DEADLOCK ------------------------ 130701 20:47:57 *** (1) TRANSACTION: TRANSACTION 4F3D6D24, ACTIVE 13 sec inserting, thread declared inside InnoDB 1 mysql tables in use 1, locked 1 LOCK WAIT 4 lock struct(s), heap size 1248, 2 row lock(s), undo log entries 1 MySQL thread id 18124702, OS thread handle 0x7fe706fdf700, query id 1435659684 localhost root update insert into lingluo values(100214,215,215,312) *** (1) WAITING FOR THIS LOCK TO BE GRANTED: RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6D24 lock_mode X insert intention waiting *** (2) TRANSACTION: TRANSACTION 4F3D6F33, ACTIVE 11 sec inserting, thread declared inside InnoDB 1 mysql tables in use 1, locked 1 4 lock struct(s), heap size 1248, 2 row lock(s), undo log entries 1 MySQL thread id 18124715, OS thread handle 0x7fea34912700, query id 1435660081 localhost root update insert into lingluo values(100215,215,215,312) *** (2) HOLDS THE LOCK(S): RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6F33 lock mode S *** (2) WAITING FOR THIS LOCK TO BE GRANTED: RECORD LOCKS space id 3351 page no 4 n bits 80 index `uk_bc` of table `test`.`lingluo` trx id 4F3D6F33 lock_mode X insert intention waiting *** WE ROLL BACK TRANSACTION (2)
原因:
s1 , type_mode=1059 //s2为s1转换隐式锁为显式锁
s2, type_mode=1282 //检查重复键，需要加共享锁，被s1 block住，等待S锁
s3, type_mode=1282 // 被s1 block住，等待S锁
s1, type_mode=547 //s1回滚，删除记录，lock_update_delete锁继承，
s2, type_mode=546 //创建s锁 LOCK_GAP | LOCK_REC | LOCK_S
s3, type_mode=546 //创建s锁 LOCK_GAP | LOCK_REC | LOCK_S
s2, type_mode=2819 // LOCK_X | LOCK_GAP | LOCK_INSERT_INTENTION
s3, type_mode=2819 // LOCK_X | LOCK_GAP | LOCK_INSERT_INTENTION
当s1回滚后，s2和s3获得s锁，但随后s2和s3又先后请求插入意向锁，因此锁队列为：
s2(S GAP)<—s3(S GAP)<—s2(插入意向锁)<–s3(插入意向锁) s3(s锁),s2(x锁),s3(x锁)形成死锁。
这样的死锁不光出现在unique key，还包括primary key（unique key的特殊形式）
印风的博客里有更详细的代码级别的记录，有兴趣的可以看下
另外如果：deadlock off的话,即使session 1rollback了，session 2和session 3还是处于等待的状态，除非超过了innodb_lock_wait_timeout的时间，报
ERROR 1205 (HY000): Lock wait timeout exceeded; try restarting transaction
这个情况和官方的这篇文章是一样的http://dev.mysql.com/doc/refman/5.5/en/innodb-locks-set.html
