由不同的索引更新解决MySQL死锁套路

(编辑：jimmy 日期: 2024/11/26 浏览：3 次 )

前几篇文章介绍了用源码的方式来调试锁相关的信息，这里同样用这个工具来解决一个线上实际的死锁案例，也是我们介绍的第一个两条 SQL 就造成死锁的情况。因为线上的表结构比较复杂，做了一些简化以后如下

CREATE TABLE `t3` (
 `id` int(11) NOT NULL AUTO_INCREMENT,
 `a` varchar(5),
 `b` varchar(5),
 PRIMARY KEY (`id`),
 UNIQUE KEY `uk_a` (`a`),
 KEY `idx_b` (`b`) 
)
INSERT INTO `t3` (`id`, `a`, `b`) VALUES 
 (1,'1','2');
# sql语句如下

# 事务1：t1
update t3 set b = '' where a = "1";

# 事务2：t2
update t3 set b = '' where b = "2";

两条语句造成死锁的情况用手动的方式比较难复现，我们先来分析一下加锁的过程

第一条语句（通过唯一索引去更新记录）

update t3 set b = '' where a = "1";

整理一下，加了3个X锁，顺序分别是

序号索引锁类型 1 uk_a X 2 PRIMARY X 3 idx_b X

第二条语句

update t3 set b = '' where b = "2";

整理一下，加了 3 个 X 锁，顺序分别是

序号索引锁类型 1 idx_b X 2 PRIMARY X 3 idx_b X

两条语句从加锁顺序看起来就已经有构成死锁的条件了

手动是比较难模拟的，写个代码并发的去同时执行那两条 SQL 语句，马上就出现死锁了

------------------------
LATEST DETECTED DEADLOCK
------------------------
181102 12:45:05
*** (1) TRANSACTION:
TRANSACTION 50AF, ACTIVE 0 sec starting index read
mysql tables in use 1, locked 1
LOCK WAIT 3 lock struct(s), heap size 376, 2 row lock(s)
MySQL thread id 34, OS thread handle 0x70000d842000, query id 549 localhost 127.0.0.1 root Searching rows for update
update t3 set b = '' where b = "2"
*** (1) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 67 page no 3 n bits 72 index `PRIMARY` of table `d1`.`t3` trx id 50AF lock_mode X locks rec but not gap waiting
Record lock, heap no 2 PHYSICAL RECORD: n_fields 5; compact format; info bits 0
 0: len 4; hex 80000001; asc ;;
 1: len 6; hex 0000000050ae; asc P ;;
 2: len 7; hex 03000001341003; asc 4 ;;
 3: len 1; hex 31; asc 1;;
 4: len 0; hex ; asc ;;

*** (2) TRANSACTION:
TRANSACTION 50AE, ACTIVE 0 sec updating or deleting
mysql tables in use 1, locked 1
4 lock struct(s), heap size 1248, 3 row lock(s), undo log entries 1
MySQL thread id 35, OS thread handle 0x70000d885000, query id 548 localhost 127.0.0.1 root Updating
update t3 set b = '' where a = "1"
*** (2) HOLDS THE LOCK(S):
RECORD LOCKS space id 67 page no 3 n bits 72 index `PRIMARY` of table `d1`.`t3` trx id 50AE lock_mode X locks rec but not gap
Record lock, heap no 2 PHYSICAL RECORD: n_fields 5; compact format; info bits 0
 0: len 4; hex 80000001; asc ;;
 1: len 6; hex 0000000050ae; asc P ;;
 2: len 7; hex 03000001341003; asc 4 ;;
 3: len 1; hex 31; asc 1;;
 4: len 0; hex ; asc ;;

*** (2) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 67 page no 5 n bits 72 index `idx_b` of table `d1`.`t3` trx id 50AE lock_mode X locks rec but not gap waiting
Record lock, heap no 2 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
 0: len 1; hex 32; asc 2;;
 1: len 4; hex 80000001; asc ;;

*** WE ROLL BACK TRANSACTION (1)

分析一下死锁日志

*** (1) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 67 page no 3 n bits 72 index PRIMARY of table d1.t3 trx id 50AF lock_mode X locks rec but not gap waiting

事务2：想获取主键索引的 X 锁

*** (2) HOLDS THE LOCK(S):
RECORD LOCKS space id 67 page no 3 n bits 72 index PRIMARY of table d1.t3 trx id 50AE lock_mode X locks rec but not gap

事务1：持有主键索引的 X 锁

*** (2) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 67 page no 5 n bits 72 index idx_b of table d1.t3 trx id 50AE lock_mode X locks rec but not gap waiting

事务1：想获取普通索引 idx_b 的 X 锁

与我们分析的完全一致，也与线上的死锁日志一模一样

以上就是本文的全部内容，希望对大家的学习有所帮助，也希望大家多多支持。

上一篇：SQL语句中OR和AND的混合使用的小技巧
下一篇：关于SQL语句中的AND和OR执行顺序遇到的问题