mysql 死锁简单分析

Posted 2021-01-25

1. mysql都有什么锁

MySQL有三种锁的级别：页级、表级、行级，内存级（latch）。

表级锁：开销小，加锁快；不会出现死锁；锁定粒度大，发生锁冲突的概率最高,并发度最低。

行级锁：开销大，加锁慢；会出现死锁；锁定粒度最小，发生锁冲突的概率最低,并发度也最高。

页面锁：开销和加锁时间界于表锁和行锁之间；会出现死锁；锁定粒度界于表锁和行锁之间，并发度一般

算法：

next KeyLocks锁，同时锁住记录(数据)，并且锁住记录前面的Gap

Gap锁，不锁记录，仅仅记录前面的Gap

Recordlock锁（锁数据，不锁Gap）

所以其实 Next-KeyLocks=Gap锁+ Recordlock锁

2. 什么情况下会造成死锁

所谓死锁 DeadLock 是指两个或两个以上的进程在执行过程中,
因争夺资源而造成的一种互相等待的现象,若无外力作用,它们都将无法推进下去.
此时称系统处于死锁状态或系统产生了死锁,这些永远在互相等竺的进程称为死锁进程.
表级锁不会产生死锁.所以解决死锁主要还是针对于最常用的InnoDB.

死锁的关键在于：两个(或以上)的Session加锁的顺序不一致。

那么对应的解决死锁问题的关键就是：让不同的session加锁有次序

4，下面就简单来重现一下死锁：

死锁重现：
事务A:
[email protected] 16:01>select connection_id();
+-----------------+
| connection_id() |
+-----------------+
| 47274 |
+-----------------+
1 row in set (0.01 sec)

[email protected] 16:02>set autocommit =0;
Query OK, 0 rows affected (0.00 sec)

[email protected] 16:02>select * from t where id =1 for update;
+----+
| id |
+----+
| 1 |
+----+
1 row in set (0.00 sec)

[email protected] 16:02>select * from t where id =2 for update;
ERROR 1213 (40001): Deadlock found when trying to get lock; try restarting transaction
[email protected] 16:03>

事务B：
[email protected] 16:02>select connection_id();
+-----------------+
| connection_id() |
+-----------------+
| 47272 |
+-----------------+
1 row in set (0.00 sec)

[email protected] 16:02>set autocommit =0;
Query OK, 0 rows affected (0.00 sec)

[email protected] 16:02>select * from t where id =2 for update;
+----+
| id |
+----+
| 2 |
+----+
1 row in set (0.00 sec)

[email protected] 16:03>select * from t where id =1 for update;
+----+
| id |
+----+
| 1 |
+----+
1 row in set (5.53 sec)

===========================================

死锁信息：

LATEST DETECTED DEADLOCK

2018-10-19 16:03:14 7f9612b6d700
(1) TRANSACTION:
TRANSACTION 870600, ACTIVE 11 sec starting index read
mysql tables in use 1, locked 1
LOCK WAIT 3 lock struct(s), heap size 360, 2 row lock(s)
MySQL thread id 47272, OS thread handle 0x7f9612e38700, query id 1112421 127.0.0.1 root statistics
select from t where id =1 for update
** (1) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 330 page no 3 n bits 72 index PRIMARY of table test.t trx id 870600 lock_mode X locks rec but not gap waiting
(2) TRANSACTION:
TRANSACTION 870599, ACTIVE 22 sec starting index read
mysql tables in use 1, locked 1
3 lock struct(s), heap size 360, 2 row lock(s)
MySQL thread id 47274, OS thread handle 0x7f9612b6d700, query id 1112422 127.0.0.1 root statistics
select * from t where id =2 for update
(2) HOLDS THE LOCK(S):
RECORD LOCKS space id 330 page no 3 n bits 72 index PRIMARY of table test.t trx id 870599 lock_mode X locks rec but not gap
(2) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 330 page no 3 n bits 72 index PRIMARY of table test.t trx id 870599 lock_mode X locks rec but not gap waiting
*** WE ROLL BACK TRANSACTION (2)

5分析：
1，这上面是显示是事务产生死锁的sql并打印出相应所持和等待的锁
2，上面的信息并没有输出事务死锁之前的sql，所以可以直接堆出两个事务执行的sql使他们相互持有了对方等待的锁
3，造成死锁是必然的，慢sql和不合理的业务的逻辑是造成死锁过多的主要原因

重要的事情说三遍：优化sql，优化业务，优化逻辑