A few days ago I was doing some cleanup on a passive master database using the MySQL client. I didn’t want my commands to be replicated so I executed set sql_log_bin=0 in my session.

One of my queries dropped an unused schema that I knew was corrupt, so I wasn’t too surprised when the drop database command crashed the MySQL server. After the crash, the server came back up quickly, and my client automatically reconnected, so it was safe to keep running queries right?

Wrong.

When the client reconnected I lost my session state, so sql_log_bin reverted to 1, and any commands I ran from that point forward would be replicated, which I did not want.

This behavior makes sense, and it’s documented in the manual:

Automatic reconnection can be convenient because you need not implement your own reconnect code, but if a reconnection does occur, several aspects of the connection state are reset on the server side and your application will not know about it. The connection-related state is affected as follows:

  • Any active transactions are rolled back and autocommit mode is reset.
  • All table locks are released.
  • All TEMPORARY tables are closed (and dropped).
  • Session variables are reinitialized to the values of the corresponding variables. This also affects variables that are set implicitly by statements such as SET NAMES.
  • User variable settings are lost.
  • Prepared statements are released.
  • HANDLER variables are closed.
  • The value of LAST_INSERT_ID() is reset to 0.
  • Locks acquired with GET_LOCK() are released.

But it’s easy to overlook such details when working with automatic features like MySQL client auto-reconnect. In this specific case I didn’t execute any other commands in the reconnected session so I didn’t inadvertantly replicate anything, but this incident served as a good reminder to be vigilant about my session state when using the MySQL client.

Here’s a snippet from my session to show the value of sql_log_bin before and after the crash:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
mysql> set sql_log_bin = 0;
Query OK, 0 rows affected (0.00 sec) mysql> select @@sql_log_bin;
+---------------+
| @@sql_log_bin |
+---------------+
| 0 |
+---------------+
1 row in set (0.00 sec) mysql> drop database test;
Query OK, 1 row affected (0.19 sec) mysql> select @@sql_log_bin;
+---------------+
| @@sql_log_bin |
+---------------+
| 0 |
+---------------+
1 row in set (0.00 sec) mysql> drop database reports;
ERROR 2013 (HY000): Lost connection to MySQL server during query
mysql> select @@sql_log_bin;
ERROR 2006 (HY000): MySQL server has gone away
No connection. Trying to reconnect...
Connection id: 505
Current database: *** NONE *** +---------------+
| @@sql_log_bin |
+---------------+
| 1 |
+---------------+
1 row in set (0.00 sec)

https://mechanics.flite.com/blog/2013/05/03/the-downside-of-mysql-auto-reconnect/

The Downside of MySQL Auto-reconnect的更多相关文章

  1. mysql AUTO INCREMENT字段 语法

    mysql AUTO INCREMENT字段 语法 作用:在新记录插入表中时生成一个唯一的数字 说明:我们通常希望在每次插入新记录时,自动地创建主键字段的值.我们可以在表中创建一个 auto-incr ...

  2. mysql auto reset

    参数说明: •相关参数说明: •dataSource: 要连接的 datasource (通常我们不会定义在 server.xml) defaultAutoCommit: 对于事务是否 autoCom ...

  3. MySQL(Percona Server) 5.6 主从复制

    MySQL(Percona Server) 5.6.15 主库:192.168.2.21 从库:192.168.2.22 例如我们同步的数据库为:test. 如果需要同步多个数据库下面会有说明. My ...

  4. 使用 xtrabackup 进行MySQL数据库物理备份

    0. xtrabackup的功能 能实现的功能: 非阻塞备份innodb等事务引擎数据库. 备份myisam表会阻塞(需要锁). 支持全备.增量备份.压缩备份. 快速增量备份(xtradb,原理类似于 ...

  5. centos升级mysql至5.7

    1.备份原数据库 [root@www ~] #mysqldump -u root –p -E –all-database > /home/db-backup.sql 加-E是因为mysqldum ...

  6. mysql 双机热备注意事项

    上一篇文章已经介绍过    主从复制,   本文对主从复制只是简单描述,如对主从复制不清楚的,可以先看上一篇文章   主从复制  一:介绍 mysql版本:5.7.20 第一个主服服务器ip:192. ...

  7. .NET Core+MySql+Nginx 容器化部署

    .NET Core容器化@Docker .NET Core容器化之多容器应用部署@Docker-Compose .NET Core+MySql+Nginx 容器化部署 GitHub-Demo:Dock ...

  8. Mysql主从复制_模式之日志点复制

    MySQL数据复制的原理 MySQL复制基于主服务器在二进制日志中跟踪所有对数据库的更改(更新.删除等等).因此,要进行复制,必须在主服务器上启用二进制日志. 每个从服务器从主服务器接收主服务器已经记 ...

  9. 安装Percona版本的MySQL主从复制

    准备两台虚拟机,按顺序执行1.1节的公共部分 1.1 首先安装 cmake # yum –y install cmake     //也需要安装gcc-c++,openssl openssl-deve ...

随机推荐

  1. Vue + Element UI 实现权限管理系统 前端篇(二):Vue + Element 案例

    导入项目 打开 Visual Studio Code,File --> add Folder to Workspace,导入我们的项目. 安装 Element 安装依赖 Element 是国内饿 ...

  2. 百度全站 https FAQ:技术宅告诉你如何搜索更安全

    百度从 14 年开始对外开放了 https 的访问,并于 3 月初正式对全网用户进行了 https 跳转. 你也许会问,切换就切换呗,和我有啥关系?我平常用百度还不是照常顺顺当当的,没感觉到什么切换. ...

  3. VGG 论文研读

    VERY DEEP CONVOLUTIONAL NETWORKS FOR LARGE-SCALE IMAGE RECOGNITION 论文地址 摘要 研究主要贡献是通过非常小的3x3卷积核的神经网络架 ...

  4. 七台机器部署Hadoop2.6.5高可用集群

    1.HA架构注意事项 两个Namenode节点在某个时间只能有一个节点正常响应客户端请求,响应请求的节点状态必须是active standby状态要能够快速无缝切换成active状态,两个NN节点必须 ...

  5. SQL Server复制入门(二)----复制的几种模式

    简介 本系列文章的上一篇对复制是什么做了一个概述.本篇文章根据发布服务器,分发服务器和订阅服务器的组织方式和复制类型来讲述常用复制的几种模式. 模式的选择 选择复制的模式取决于多个方面.首先需要考虑具 ...

  6. 在VS2010中使用Git管理源代码

    前文我们讲了使用TortoiseGit管理源代码,但是对于使用VS2010的朋友来说,源代码管理起来还是不怎么方便.要是直接在VS2010中能使用Git就好了,下面我们就来看看怎么在VS2010中使用 ...

  7. order by与索引(转载)

    order by与索引   ORDER BY 通常会有两种实现方法,一个是利用有序索引自动实现,也就是说利用有序索引的有序性就不再另做排序操作了.另一个是把结果选好之后再排序. 用有序索引这种,当然是 ...

  8. 简单的SpringMVC经典案例

    主题:构建一个基于SpringMVC的HelloWord Web 项目 目的:快速体验什么是SpringMVC 方案: 1.创建工程,命名:SpringMVC 2.导包 3.在SRC下添加spring ...

  9. SORT AGAIN(hdu2523)

    SORT AGAIN Time Limit: 2000/1000 MS (Java/Others)    Memory Limit: 32768/32768 K (Java/Others) Total ...

  10. Mybatis逆向生成代碼

    Idea 单模块 1.在pom.xml中添加依赖 <build> <plugins> <plugin> <groupId>org.mybatis.gen ...