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 '',
`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

有趣的insert死锁的更多相关文章

  1. InnoDB的锁机制浅析(五)—死锁场景(Insert死锁)

    可能的死锁场景 文章总共分为五个部分: InnoDB的锁机制浅析(一)-基本概念/兼容矩阵 InnoDB的锁机制浅析(二)-探索InnoDB中的锁(Record锁/Gap锁/Next-key锁/插入意 ...

  2. Mysql Innodb存储引擎 insert 死锁分析

    http://chenzhenianqing.cn/articles/1308.html

  3. mysql insert锁机制【转】

    最近再找一些MySQL锁表原因,整理出来一部分sql语句会锁表的,方便查阅,整理的不是很全,都是工作中碰到的,会持续更新 笔者能力有限,如果有不正确的,或者不到位的地方,还请大家指出来,方便你我,方便 ...

  4. Cassandra1.2文档学习(17)—— CQL数据模型(上)

    参考文档:http://www.datastax.com/documentation/cql/3.0/webhelp/index.html#cql/ddl/ddl_anatomy_table_c.ht ...

  5. Mysql innodb 间隙锁

    前段时间系统老是出现insert死锁,很是纠结.经过排查发现是间隙锁!间隙锁是innodb中行锁的一种, 但是这种锁锁住的却不止一行数据,他锁住的是多行,是一个数据范围.间隙锁的主要作用是为了防止出现 ...

  6. java面试——问题回溯

    背景:用来记录面试过程中遇到的问题,在这里进行记录,下次不要犯同样的错误. 迪普科技 Linux服务器下的top命令 #动态更新的虚拟文件实际上是许多其他内存相关工具(如:free / ps / to ...

  7. InnoDB的锁机制浅析(四)—不同SQL的加锁状况

    不同SQL的加锁状况 文章总共分为五个部分: InnoDB的锁机制浅析(一)-基本概念/兼容矩阵 InnoDB的锁机制浅析(二)-探索InnoDB中的锁(Record锁/Gap锁/Next-key锁/ ...

  8. InnoDB的锁机制浅析(三)—幻读

    文章总共分为五个部分: InnoDB的锁机制浅析(一)-基本概念/兼容矩阵 InnoDB的锁机制浅析(二)-探索InnoDB中的锁(Record锁/Gap锁/Next-key锁/插入意向锁) Inno ...

  9. InnoDB的锁机制浅析(二)—探索InnoDB中的锁(Record锁/Gap锁/Next-key锁/插入意向锁)

    Record锁/Gap锁/Next-key锁/插入意向锁 文章总共分为五个部分: InnoDB的锁机制浅析(一)-基本概念/兼容矩阵 InnoDB的锁机制浅析(二)-探索InnoDB中的锁(Recor ...

随机推荐

  1. SET Statements for SQLServer

    SET SHOWPLAN_ALL { ON | OFF } It will not execute the TSQL statements. It cannot be specified inside ...

  2. C语言语句分类:大致可分为六大类

    #include <stdio.h> void main() { int sign, i, n; float sum; printf("please input an integ ...

  3. iOS自动偏移64个像素

    自从iOS7开始,如果添加的scrollview是uiviewController第一个视图,系统会默认自动添加-64的偏移量,所以规避的方案就添加一个UIView之后再添加你的scrollview.

  4. log4j2.x 配置文件默认寻找顺序

    Automatic Configuration Log4j has the ability to automatically configure itself during initializatio ...

  5. log file switch (checkpoint incomplete)

    接手了一个新交接的库,alert日志频繁报告log file switch (checkpoint incomplete)事件 oracle文档解释: Waiting for a log switch ...

  6. mysql 插入重复值 INSERT ... ON DUPLICATE KEY UPDATE

    向数据库插入记录时,有时会有这种需求,当符合某种条件的数据存在时,去修改它,不存在时,则新增,也就是saveOrUpdate操作.这种控制可以放在业务层,也可以放在数据库层,大多数数据库都支持这种需求 ...

  7. 数dp多少个0-n多少0-9

    #include <bits/stdc++.h> using namespace std; const int N = 15; int n; int dp[N][N][N]; int a[ ...

  8. java基础(59):Eclipse把自动括号匹配改成C++那样的(强迫症,没办法)

    在Eclipse开发环境下依次选择: 1.Windows->Preferences->Java->Code Style->Formatter: 2.在Active Profil ...

  9. javaapi中的排序

    有的时候需要对数组里的element进行排序.当然可以自己编写合适的排序方法,但既然java包里有自带的Arrays.sort排序方法,在 数组元素比较少的时候为何不用? Sorting an Arr ...

  10. bzoj3192 [JLOI2013]删除物品

    用数组表示两个栈,将两个栈的栈顶并在一起,用树状数组维护一下操作即可. 代码 #include<cstdio> #include<algorithm> #include< ...