mysql 连接占满处理方式

1. 最近数据库莫名卡顿, 通过排查发现连接拉满

   mysql> show processlist;  # 查看线程情况
   +-----+-----------------+-----------------+-----------+---------+------+------------------------+------------------+
   | Id  | User            | Host            | db        | Command | Time | State                  | Info             |
   +-----+-----------------+-----------------+-----------+---------+------+------------------------+------------------+
   |   5 | event_scheduler | localhost       | NULL      | Daemon  | 1757 | Waiting on empty queue | NULL             |
   |  12 | root            | localhost:43514 | infosecqa | Sleep   |  538 |                        | NULL             |
   |  49 | root            | localhost:43540 | infosecqa | Sleep   |  480 |                        | NULL             |
   |  51 | root            | localhost:43544 | infosecqa | Sleep   |  479 | 
   

   mysql> show variables like "%max_connection%";  # 查看最大连接数
   +------------------------+-------+
   | Variable_name          | Value |
   +------------------------+-------+
   | max_connections        | 151   |  # mysql最大允许连接数
   | mysqlx_max_connections | 100   |
   +------------------------+-------+
   2 rows in set (0.00 sec)
   

   mysql> show status like "Thread%";  # 查看连接情况
   +-------------------+-------+
   | Variable_name     | Value |
   +-------------------+-------+
   | Threads_cached    | 4     |
   | Threads_connected | 143   |  # 已建立的连接数
   | Threads_created   | 148   |  # 打开的连接数
   | Threads_running   | 2     |  # 正在运行的连接数
   +-------------------+-------+
   4 rows in set (0.00 sec)
   

2. 根据show processlist中的某个连接

   • 比如这个看一下他在干什么: localhost:45686

   • [root@infosec ~]# ss -tp|grep mysql  # 发现说的连接都是2817538这个进程创建出来的
     ESTAB      0      0         [::ffff:127.0.0.1]:mysql                   [::ffff:127.0.0.1]:44620                 users:(("mysqld",pid=2817538,fd=165))
     ESTAB      0      0         [::ffff:127.0.0.1]:mysql                   [::ffff:127.0.0.1]:44352                 users:(("mysqld",pid=2817538,fd=117))
     ESTAB      0      0         [::ffff:127.0.0.1]:mysql                   [::ffff:127.0.0.1]:44540                 users:(("mysqld",pid=2817538,fd=150))
     ESTAB      0      0         [::ffff:127.0.0.1]:mysql                   [::ffff:127.0.0.1]:44910                 users:(("mysqld",pid=2817538,fd=205))
     

   • [root@infosec ~]# ps -ef|grep 281753[8]  # 查看当前进程是什么, 发现是mysql的后台的进程, 然我我就不知道该咋们办
     mysql    2817538 2817234  1 11:29 ?        00:01:38 /usr/local/mysql/bin/mysqld --basedir=/usr/local/mysql --datadir=/usr/local/mysql/data --plugin-dir=/usr/local/mysql/lib/plugin --user=mysql --log-error=xxx.err --pid-file=/usr/local/mysql/data/xxxx.pid --socket=/tmp/mysql.sock --port=xxxx
     

3. 突然发现ss查看到后面还有一个fd文件描述符

   • 我也不知道有了文件描述符我可以干嘛, 所有随便搜了一下linux 根据文件描述符它找到正在操作的文件, 这不就是我想要的么?

   • [root@infosec ~]# lsof -d 150  # ok根据这里看到同事写的代码有问题, 因为你已经知道是哪个文件哪个进程命令在操作, 直接让同事修bug吧
     COMMAND     PID  USER   FD   TYPE     DEVICE   SIZE/OFF       NODE NAME
     uwsgi      2510  root  150w   REG        8,2   33390455 1626717201 /XXX/base_xxx/base_xxx_httprunner.log (deleted)
     uwsgi   2443362  root  150w   REG        8,2 1140192047 1626762311 /XXX/base_test1/base_test1_httprunner_error.log
     mysqld  2817538 mysql  150u  IPv6 2663696291        0t0        TCP localhost:mysql->localhost:44540 (ESTABLISHED)
     uwsgi   2840143  root  150w   REG        8,2 1316132498 1626762312 /XXX/base_test1/base_test1_httprunner.log
     uwsgi   3911893  root  150w   REG        8,2  388836144  544195539 /XXX/base_NettAuthlite/base_NettAuthlite_httprunner.log (deleted)
     uwsgi   3911894  root  150w   REG        8,2  388836144  544195539 /XXXXXX/base_NettAuthlite/base_NettAuthlite_httprunner.log (deleted)
     

   • [root@infosec ~]# tail -fn 30  # 看看日志发生了什么, 直接定位到了错误
     /XXX/base_test1/base_test1_httprunner.log
     2022-04-13 10:40:06.850 | INFO     | HttpApi.HttprunDemo.runner:__run_step:324 - run step end: 参数错误_get <<<<<<
     2022-04-13 10:40:06.850 | INFO     | HttpApi.HttprunDemo.runner:__run_step:324 - run step end: 参数错误_get <<<<<<
     2022-04-13 10:40:06.850 | INFO     | HttpApi.HttprunDemo.runner:__run_step:324 - run step end: 参数错误_get <<<<<<
     2022-04-13 10:40:06.850 | INFO     | HttpApi.HttprunDemo.runner:__run_step:324 - run step end: 参数错误_get <<<<<<
     2022-04-13 10:40:06.850 | INFO     | HttpApi.HttprunDemo.runner:__run_step:324 - run step end: 参数错误_get <<<<<<
     

4. 先把占满的线程杀掉(mysql中)

   •  select concat("kill ", id, ";") from information_schema.processlist where COMMAND="Sleep" into outfile "/temp/a.txt";  # 拼接kill语句
     
      如果不能into 到文件就把结果复制到一个文件中使用awk手动切割, 按照|和空格切割文件, 并将结果重新写到b.txt
      awk -F '[\|| ]' '{print $3,$4}' a.txt > b.txt
      mysql中执行
      source /tmp/b.txt
     

参考:

• lsof命令
• 快速处理mysql连接数占满的问题