前几天跟一位朋友分析了一个死锁问题,所以有了这篇图文详细的博文,哈哈~
发生死锁了,如何排查和解决呢?本文将跟你一起探讨这个问题
数据库隔离级别:
mysql> select @@tx_isolation;
+-----------------+
| @@tx_isolation |
+-----------------+
| REPEATABLE-READ |
+-----------------+
1 row in set, 1 warning (0.00 sec)
自动提交关闭:
mysql> set autocommit=0;
Query OK, 0 rows affected (0.00 sec)
mysql> select @@autocommit;
+--------------+
| @@autocommit |
+--------------+
| 0 |
+--------------+
1 row in set (0.00 sec)
表结构:
//id是自增主键,name是非唯一索引,balance普通字段
CREATE TABLE `account` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`name` varchar(255) DEFAULT NULL,
`balance` int(11) DEFAULT NULL,
PRIMARY KEY (`id`),
KEY `idx_name` (`name`) USING BTREE
) ENGINE=InnoDB AUTO_INCREMENT=3 DEFAULT CHARSET=utf8;
表中的数据:
开启两个终端模拟事务并发情况,执行顺序以及实验现象如下:
1)事务A执行更新操作,更新成功
mysql> update account set balance =1000 where name ='Wei';
Query OK, 1 row affected (0.01 sec)
2)事务B执行更新操作,更新成功
mysql> update account set balance =1000 where name ='Eason';
Query OK, 1 row affected (0.01 sec)
3)事务A执行插入操作,陷入阻塞~
mysql> insert into account values(null,'Jay',100);
这时候可以用 select*frominformation_schema.innodb_locks;查看锁情况:
4)事务B执行插入操作,插入成功,同时事务A的插入由阻塞变为死锁error。
mysql> insert into account values(null,'Yan',100);
Query OK, 1 row affected (0.01 sec)
在分析死锁日志前,先做一下锁介绍,哈哈~
主要介绍一下兼容性以及锁模式类型的锁:
InnoDB 实现了标准的行级锁,包括两种:共享锁(简称 s 锁)、排它锁(简称 x 锁)。
如果事务 T1 持有行 r 的 s 锁,那么另一个事务 T2 请求 r 的锁时,会做如下处理:
如果 T1 持有 r 的 x 锁,那么 T2 请求 r 的 x、s 锁都不能被立即允许,T2 必须等待T1释放 x 锁才可以,因为X锁与任何的锁都不兼容。
比如:事务1在表1上加了S锁后,事务2想要更改某行记录,需要添加IX锁,由于不兼容,所以需要等待S锁释放;如果事务1在表1上加了IS锁,事务2添加的IX锁与IS锁兼容,就可以操作,这就实现了更细粒度的加锁。
InnoDB存储引擎中锁的兼容性如下表:
记录锁的事务数据(关键词:lock_mode X locks rec butnotgap),记录如下:
RECORD LOCKS space id 58 page no 3 n bits 72 index `PRIMARY` of table `test`.`t`
trx id 10078 lock_mode X locks rec but not gap
Record lock, heap no 2 PHYSICAL RECORD: n_fields 3; compact format; info bits 0
0: len 4; hex 8000000a; asc ;;
1: len 6; hex 00000000274f; asc 'O;;
2: len 7; hex b60000019d0110; asc ;;
间隙锁的事务数据(关键词:gap before rec),记录如下:
RECORD LOCKS space id 177 page no 4 n bits 80 index idx_name of table `test2`.`account`
trx id 38049 lock_mode X locks gap before rec
Record lock, heap no 6 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
0: len 3; hex 576569; asc Wei;;
1: len 4; hex 80000002; asc ;;
事务数据类似于下面:
RECORD LOCKS space id 31 page no 3 n bits 72 index `PRIMARY` of table `test`.`child`
trx id 8731 lock_mode X locks gap before rec insert intention waiting
Record lock, heap no 3 PHYSICAL RECORD: n_fields 3; compact format; info bits 0
0: len 4; hex 80000066; asc f;;
1: len 6; hex 000000002215; asc " ;;
2: len 7; hex 9000000172011c; asc r ;;...
锁模式兼容矩阵(横向是已持有锁,纵向是正在请求的锁):
可以用 show engine innodb status,查看最近一次死锁日志哈~,执行后,死锁日志如下:
2020-04-11 00:35:55 0x243c
*** (1) TRANSACTION:
TRANSACTION 38048, ACTIVE 92 sec inserting
mysql tables in use 1, locked 1
LOCK WAIT 4 lock struct(s), heap size 1136, 4 row lock(s), undo log entries 2
MySQL thread id 53, OS thread handle 2300, query id 2362 localhost ::1 root update
insert into account values(null,'Jay',100)
*** (1) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 177 page no 4 n bits 80 index idx_name of table `test2`.`account`
trx id 38048 lock_mode X locks gap before rec insert intention waiting
Record lock, heap no 6 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
0: len 3; hex 576569; asc Wei;;
1: len 4; hex 80000002; asc ;;
*** (2) TRANSACTION:
TRANSACTION 38049, ACTIVE 72 sec inserting, thread declared inside InnoDB 5000
mysql tables in use 1, locked 1
5 lock struct(s), heap size 1136, 4 row lock(s), undo log entries 2
MySQL thread id 52, OS thread handle 9276, query id 2363 localhost ::1 root update
insert into account values(null,'Yan',100)
*** (2) HOLDS THE LOCK(S):
RECORD LOCKS space id 177 page no 4 n bits 80 index idx_name of table `test2`.`account`
trx id 38049 lock_mode X locks gap before rec
Record lock, heap no 6 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
0: len 3; hex 576569; asc Wei;;
1: len 4; hex 80000002; asc ;;
*** (2) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 177 page no 4 n bits 80 index idx_name of table `test2`.`account`
trx id 38049 lock_mode X insert intention waiting
Record lock, heap no 1 PHYSICAL RECORD: n_fields 1; compact format; info bits 0
0: len 8; hex 73757072656d756d; asc supremum;;
*** WE ROLL BACK TRANSACTION (1)
我们如何分析以上死锁日志呢?
1)找到关键词TRANSACTION,事务38048
2)查看正在执行的SQL
insert into account values(null,'Jay',100)
3)正在等待锁释放(WAITING FOR THIS LOCK TO BE GRANTED),插入意向排他锁(lockmode X locks gap before rec insert intention waiting),普通索引(idxname),物理记录(PHYSICAL RECORD),间隙区间(未知,Wei);
1)找到关键词TRANSACTION,事务38049
2)查看正在执行的SQL
insert into account values(null,'Yan',100)
3)持有锁(HOLDS THE LOCK),间隙锁(lockmode X locks gap before rec),普通索引(index idxname),物理记录(physical record),区间(未知,Wei);
4)正在等待锁释放(waiting for this lock to be granted),插入意向锁(lockmode X insert intention waiting),普通索引上(index idxname),物理记录(physical record),间隙区间(未知,+ );
5)事务1回滚(we roll back transaction 1);
查看日志可得:
这里面,有些朋友可能有疑惑,
我们接下来一小节详细分析一波,一个一个问题来~
为了方便记录,例子用W表示Wei,J表示Jay,E表示Eason哈~
update account set balance =1000 where name ='Wei';
间隙锁:
记录锁
Next-Key锁
综上所述,事务A执行完update更新语句,会持有锁:
insert into account values(null,'Jay',100);
间隙锁:
插入意向锁(Insert Intention)
因此,事务A的update语句和insert语句执行完,它是持有了 (E,W]的 Next-Key锁,(W,+ )的Gap锁,想拿到 (E,W)的插入意向排它锁,等待的锁跟死锁日志是对上的,哈哈~
update account set balance =1000 where name ='Eason';
间隙锁:
记录锁
Next-Key锁
综上所述,事务B执行完update更新语句,会持有锁:
insert into account values(null,'Yan',100);
间隙锁:
插入意向锁(Insert Intention)
所以,事务B的update语句和insert语句执行完,它是持有了 (- ,E]的 Next-Key锁,(E,W)的Gap锁,想拿到 (W,+ )的间隙锁,即插入意向排它锁,加锁情况跟死锁日志也是对上的~
接下来呢,让我们一起还原死锁真相吧~哈哈~
表结构:
CREATE TABLE `song_rank` (
`id` int(11) NOT NULL AUTO_INCREMENT,
`songId` int(11) NOT NULL,
`weight` int(11) NOT NULL DEFAULT '0',
PRIMARY KEY (`id`),
UNIQUE KEY `songId_idx` (`songId`) USING BTREE
) ENGINE=InnoDB DEFAULT CHARSET=utf8;
隔离级别:
mysql> select @@tx_isolation;
+-----------------+
| @@tx_isolation |
+-----------------+
| REPEATABLE-READ |
+-----------------+
1 row in set, 1 warning (0.00 sec)
数据库版本:
+------------+
| @@version |
+------------+
| 5.7.21-log |
+------------+
1 row in set (0.00 sec)
关闭自动提交:
mysql> select @@autocommit;
+--------------+
| @@autocommit |
+--------------+
| 1 |
+--------------+
1 row in set (0.00 sec)
mysql> set autocommit=0;
Query OK, 0 rows affected (0.00 sec)
mysql> select @@autocommit;
+--------------+
| @@autocommit |
+--------------+
| 0 |
+--------------+
1 row in set (0.00 sec)
表中的数据:
mysql> select * from song_rank;
+----+--------+--------+
| id | songId | weight |
+----+--------+--------+
| 1 | 10 | 30 |
| 2 | 20 | 30 |
+----+--------+--------+
2 rows in set (0.01 sec)
死锁案发原因:
并发环境下,执行insert into … on duplicate key update…导致死锁
死锁模拟复现:
事务一执行:
mysql> begin; //第一步
Query OK, 0 rows affected (0.00 sec)
mysql> insert into song_rank(songId,weight) values(15,100) on duplicate key update weight=weight+1; //第二步
Query OK, 1 row affected (0.00 sec)
mysql> rollback; //第七步
Query OK, 0 rows affected (0.00 sec)
事务二执行:
mysql> begin; //第三步
Query OK, 0 rows affected (0.00 sec)
mysql> insert into song_rank(songId,weight) values(16,100) on duplicate key update weight=weight+1; // 第四步
Query OK, 1 row affected (40.83 sec)
事务三执行:
mysql> begin; //第五步
Query OK, 0 rows affected (0.00 sec)
mysql> insert into song_rank(songId,weight) values(18,100) on duplicate key update weight=weight+1; //第六步
事务一,事务二,事务三执行:
步骤 | 事务一 | 事务二 | 事务三 |
第一步 | begin; | ||
第二步 | insert into song_rank(songId,weight) values(15,100) on duplicate key update weight=weight+1; (Query OK, 1 row affected (0.00 sec) ) | ||
第三步 | begin; | ||
第四步 | insert into song_rank(songId,weight) values(16,100) on duplicate key update weight=weight+1; //被阻塞 | ||
第五步 | begin; | ||
第六步 | insert into song_rank(songId,weight) values(18,100) on duplicate key update weight=weight+1; //被阻塞 | ||
第七步 | rollback; | ||
结果 | Query OK, 1 row affected (40.83 sec) | ERROR 1213 (40001): Deadlock found when trying to get lock; try restarting transaction |
死锁浮出水面:
ERROR 1213 (40001): Deadlock found when trying to get lock; try restarting transaction
遇到死锁问题时,我们应该怎么处理呢?分一下几个步骤
当数据库发生死锁时,可以通过以下命令获取死锁日志:
show engine innodb status;
上面例子insert on duplicate死锁问题的日志如下:
*** (1) TRANSACTION:
TRANSACTION 27540, ACTIVE 19 sec inserting
mysql tables in use 1, locked 1
LOCK WAIT 3 lock struct(s), heap size 1136, 2 row lock(s), undo log entries 1
MySQL thread id 23, OS thread handle 14896, query id 582 localhost ::1 root update
insert into song_rank(songId,weight) values(18,100) on duplicate key update weight=weight+1
*** (1) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 116 page no 4 n bits 72 index songId_idx of table `test2`.`song_rank` trx id 27540 lock_mode X
locks gap before rec insert intention waiting
Record lock, heap no 3 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
0: len 4; hex 80000014; asc ;;
1: len 4; hex 80000002; asc ;;
*** (2) TRANSACTION:
TRANSACTION 27539, ACTIVE 41 sec inserting, thread declared inside InnoDB 1
mysql tables in use 1, locked 1
4 lock struct(s), heap size 1136, 3 row lock(s), undo log entries 1
MySQL thread id 22, OS thread handle 6976, query id 580 localhost ::1 root update
insert into song_rank(songId,weight) values(16,100) on duplicate key update weight=weight+1
*** (2) HOLDS THE LOCK(S):
RECORD LOCKS space id 116 page no 4 n bits 72 index songId_idx of table `test2`.`song_rank` trx id 27539 lock_mode X
locks gap before rec
Record lock, heap no 3 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
0: len 4; hex 80000014; asc ;;
1: len 4; hex 80000002; asc ;;
*** (2) WAITING FOR THIS LOCK TO BE GRANTED:
RECORD LOCKS space id 116 page no 4 n bits 72 index songId_idx of table `test2`.`song_rank` trx id 27539 lock_mode X
locks gap before rec insert intention waiting
Record lock, heap no 3 PHYSICAL RECORD: n_fields 2; compact format; info bits 0
0: len 4; hex 80000014; asc ;;
1: len 4; hex 80000002; asc ;;
如何分析死锁日志呢? 分享一下我的思路
从日志我们可以看到事务1正在执行的SQL为:
insert into song_rank(songId,weight) values(18,100) on duplicate key update weight=weight+1
该条语句正在等待索引songId_idx的插入意向排他锁:
lock_mode X locks gap before rec insert intention waiting
从日志我们可以看到事务2正在执行的SQL为:
insert into song_rank(songId,weight) values(16,100) on duplicate key update weight=weight+1
该语句持有一个索引songId_idx的间隙锁:
lock_mode X locks gap before rec
该条语句正在等待索引songId_idx的插入意向排他锁:
lock_mode X locks gap before rec insert intention waiting
考虑到有些读者可能对上面insert intention锁等不太熟悉,所以这里这里补一小节锁相关概念。官方文档
InnoDB 锁类型思维导图:
我们主要介绍一下兼容性以及锁模式类型的锁
1.共享锁与排他锁:
InnoDB 实现了标准的行级锁,包括两种:共享锁(简称 s 锁)、排它锁(简称 x 锁)。
如果事务 T1 持有行 r 的 s 锁,那么另一个事务 T2 请求 r 的锁时,会做如下处理:
如果 T1 持有 r 的 x 锁,那么 T2 请求 r 的 x、s 锁都不能被立即允许,T2 必须等待T1释放 x 锁才可以,因为X锁与任何的锁都不兼容。
2.意向锁
比如:事务1在表1上加了S锁后,事务2想要更改某行记录,需要添加IX锁,由于不兼容,所以需要等待S锁释放;如果事务1在表1上加了IS锁,事务2添加的IX锁与IS锁兼容,就可以操作,这就实现了更细粒度的加锁。
InnoDB存储引擎中锁的兼容性如下表:
兼容性 | IS | IX | S | X |
IS | 兼容 | 兼容 | 兼容 | 不兼容 |
IX | 兼容 | 兼容 | 不兼容 | 不兼容 |
S | 兼容 | 不兼容 | 兼容 | 不兼容 |
X | 不兼容 | 不兼容 | 不兼容 | 不兼容 |
3.记录锁(Record Locks)
记录锁的事务数据(关键词: lock_mode X locks rec butnotgap),记录如下:
RECORD LOCKS space id 58 page no 3 n bits 72 index `PRIMARY` of table `test`.`t`
trx id 10078 lock_mode X locks rec but not gap
Record lock, heap no 2 PHYSICAL RECORD: n_fields 3; compact format; info bits 0
0: len 4; hex 8000000a; asc ;;
1: len 6; hex 00000000274f; asc 'O;;
2: len 7; hex b60000019d0110; asc ;;
4.间隙锁(Gap Locks)
5.Next-Key Locks
6.插入意向锁(Insert Intention)
事务数据类似于下面:
RECORD LOCKS space id 31 page no 3 n bits 72 index `PRIMARY` of table `test`.`child`
trx id 8731 lock_mode X locks gap before rec insert intention waiting
Record lock, heap no 3 PHYSICAL RECORD: n_fields 3; compact format; info bits 0
0: len 4; hex 80000066; asc f;;
1: len 6; hex 000000002215; asc " ;;
2: len 7; hex 9000000172011c; asc r ;;...
通过分析死锁日志,我们可以找到发生死锁的SQL,以及相关等待的锁,我们再对对应的SQL进行加锁分析,其实问题就迎刃而解了。
OK,我们回到对应的SQL,insert into song_rank(songId,weight) values(16,100) on duplicate key update weight=weight+1 执行过程到底加了什么锁呢?加锁机制官方文档
insert加锁策略:
insert语句会对插入的这条记录加排他记录锁,在加记录锁之前还会加一种 GAP 锁,叫做插入意向(insert intention)锁,如果出现唯一键冲突,还会加一个共享记录(S)锁。
(SQL加锁分析非常重要,在这里给大家推荐一篇文章,讲的非常好,解决死锁之路 - 常见 SQL 语句的加锁分析)
insert on duplicate key加锁验证
为了验证一下insert on duplicate key加锁情况,我们拿上面demo的事务1和2在走一下流程。 事务1:
mysql> begin; //第一步
Query OK, 0 rows affected (0.00 sec)
mysql> insert into song_rank(songId,weight) values(15,100) on duplicate key
update weight=weight+1; //第二步
Query OK, 1 row affected (0.00 sec)
事务2(另开窗口):
mysql> begin; //第三步
Query OK, 0 rows affected (0.00 sec)
mysql> insert into song_rank(songId,weight) values(16,100) on duplicate key
update weight=weight+1; // 第四步
使用show engine innodb status查看当前锁请求信息,如图:
有图可得:
事务2持有:IX锁(表锁),gap x锁,insert intention lock(在等待事务1的gap锁)
所以,insert on duplicate 执行过程会上这三把锁。
回归到本文开头介绍的死锁案发模拟现场(事务1,2,3)以及死锁日志现场,
案发后事务1的锁:
案发后事务2的锁:
案发复原路线:
1.首先,执行事务1执行: begin; insertintosong_rank(songId,weight)values(15,100)on duplicate key update weight=weight+1;会获得 gap锁(10,20),insert intention lock(插入意向锁)
2.接着,事务2执行: begin; insertintosong_rank(songId,weight)values(16,100)on duplicate key update weight=weight+1;会获得 gap锁(10,20),同时等待事务1的insert intention lock(插入意向锁)。
3.再然后,事务3执行: begin; insertintosong_rank(songId,weight)values(18,100)on duplicate key update weight=weight+1;会获得 gap锁(10,20),同时等待事务1的insert intention lock(插入意向锁)。
4.最后,事务1回滚(rollback),释放插入意向锁,导致事务2,3同时持有gap锁,等待insert intention锁,死锁形成!
锁模式兼容矩阵(横向是已持有锁,纵向是正在请求的锁):
兼容性 | Gap | Insert Intention | Record | Next-Key |
Gap | 兼容 | 兼容 | 兼容 | 兼容 |
Insert Intention | 冲突 | 兼容 | 兼容 | 冲突 |
Record | 兼容 | 兼容 | 冲突 | 冲突 |
Next-Key | 兼容 | 兼容 | 冲突 | 冲突 |
这是MySql5.7的一个bug
try{
insert();
}catch(DuplicateKeyException e){
update();
}
因为insert不会加gap锁,所以可以避免该问题。
既然这是MySql5.7的一个bug,那么可以考虑更改Mysql版本。
gap锁跟索引有关,并且unique key 和foreign key会引起额外的index检查,需要更大的开销,所以我们尽量减少使用不必要的索引。
本文介绍了MySql5.7死锁的一个bug。我们应该怎样去排查死锁问题呢?
最后,遇到死锁问题,我们应该怎么分析呢?
页面更新:2024-04-23
本站资料均由网友自行发布提供,仅用于学习交流。如有版权问题,请与我联系,QQ:4156828
© CopyRight 2008-2024 All Rights Reserved. Powered By bs178.com 闽ICP备11008920号-3
闽公网安备35020302034844号