为什么这个MySQL触发器会导致堆栈溢出？

Question

我认为有人试图模拟第二个auto_increment值.刚升级到MySQL 5.5.9

CREATE TABLE `job_title` (
  `job_id` int(11) NOT NULL AUTO_INCREMENT,
  `position_id` int(11) DEFAULT NULL,
  `title` varchar(255) COLLATE latin1_general_cs NOT NULL,
  `selectable` tinyint(4) NOT NULL DEFAULT '0',
  PRIMARY KEY (`job_id`),
  UNIQUE KEY `title` (`title`)
) ENGINE=InnoDB;

create trigger job_position_trigger
  before insert on job_title for each row
 begin
   if new.position_id is null then 
     set @position = (select max(position_id)+1 from job_title);
     set new.position_id = @position;
   end if;
 end

错误: Thread stack overrun: 9024 bytes used of a 131072 byte stack, and 128000 bytes needed. Use 'mysqld --thread_stack=#' to specify a bigger stack.' on query. Default database: 'mydb'. Query: 'insert ignore into job_title (title) values ('Morning Show Personality')

Answer 1

我今天遇到了同样的问题,每次触发都会导致堆栈溢出.原来我的Zend社区服务器安装附带了一个默认的my.cnf文件,其中thread_stack大小设置为128K,这导致每个线程中可用于堆栈的131072字节:

mysql> show variables where `Variable_name` = 'thread_stack';
+---------------+--------+
| Variable_name | Value  |
+---------------+--------+
| thread_stack  | 131072 |
+---------------+--------+

所以我在/usr/local/zend/mysql/data/my.cnf中注释了这一行,重新启动了mysql守护进程,瞧!该默认 192K是

mysql> show variables where `Variable_name` = 'thread_stack';
+---------------+--------+
| Variable_name | Value  |
+---------------+--------+
| thread_stack  | 196608 |
+---------------+--------+

现在你的桌子和tchester的触发器工作完美:)(尽管请注意分隔符)

mysql> CREATE TABLE `job_title` (
    ->   `job_id` int(11) NOT NULL AUTO_INCREMENT,
    ->   `position_id` int(11) DEFAULT NULL,
    ->   `title` varchar(255) COLLATE latin1_general_cs NOT NULL,
    ->   `selectable` tinyint(4) NOT NULL DEFAULT '0',
    ->   PRIMARY KEY (`job_id`),
    ->   UNIQUE KEY `title` (`title`)
    -> ) ENGINE=InnoDB;
Query OK, 0 rows affected (0.14 sec)

mysql> DELIMITER &&
mysql> create trigger job_position_trigger   
    ->   before insert on job_title for each row  
    -> begin    
    ->     if new.position_id is null then       
    ->        set @position = (select max(position_id)+1 from job_title);
    ->        if @position is null then set @position = 1; end if;
    ->        set new.position_id = @position;    
    ->     end if;  
    -> end; 
    -> &&
Query OK, 0 rows affected (0.29 sec)

mysql> DELIMITER ;
mysql> insert into job_title (title, selectable) values ("test", 1);
Query OK, 1 row affected (0.00 sec)

mysql> insert into job_title (title, selectable) values ("test2", 3);
Query OK, 1 row affected (0.00 sec)

mysql> select * from job_title;
+--------+-------------+-------+------------+
| job_id | position_id | title | selectable |
+--------+-------------+-------+------------+
|      1 |           1 | test  |          1 |
|      2 |           2 | test2 |          3 |
+--------+-------------+-------+------------+
2 rows in set (0.00 sec)

您获得的错误,131072字节堆栈使用的9024字节和所需的128000字节是有意义的:9024 + 128000> 131072.

Answer 2

在 MySQL 5.1 下，我无法导致堆栈溢出，但触发器从未设置 [position_id] 字段。它从未被设置的原因是，当您对空表或 [position_id] 列中只有 NULL 的表执行 MAX(position_id) + 1 时，您的 @position 变量将被设置为 NULL。我想知道这是否导致触发器重新评估自身（看到 [position_id] 仍然为空），因此它重复调用自身导致堆栈溢出。您可以尝试的一种选择是更改触发器以检查 null @position 值并在分配之前将其强制为“1”。

create trigger job_position_trigger   
  before insert on job_title for each row  
begin    
    if new.position_id is null then       
       set @position = (select max(position_id)+1 from job_title);
       if @position is null then set @position = 1; end if;
       set new.position_id = @position;    
    end if;  
end;