[原创] CentOS7 MySQL 无法启动

环境说明：
CentOS Linux release 7.3.1611 (Core) 64位
mysql  Ver 14.14 Distrib 5.7.19

现象：
安装完 Wordpress 及 DB 后，在 Wordpress 中删除了默认的文章，之后就突然莫名其妙的提示无法连接DB。（之前也可能做过了一些其它 DB 操作及修改，有点儿记不清了）

原因调查：
首先发现的是，不知道为什么，MySQL 被停止了，并且提示如下错误：

# systemctl status mysqld.service
● mysqld.service - MySQL Server
   Loaded: loaded (/usr/lib/systemd/system/mysqld.service; enabled; vendor preset: disabled)
   Active: failed (Result: start-limit) since 五 2017-10-20 09:13:16 CST; 633ms ago
     Docs: man:mysqld(8)
           http://dev.mysql.com/doc/refman/en/using-systemd.html
  Process: 10382 ExecStart=/usr/sbin/mysqld --daemonize --pid-file=/var/run/mysqld/mysqld.pid $MYSQLD_OPTS (code=exited, status=1/FAILURE)
  Process: 10365 ExecStartPre=/usr/bin/mysqld_pre_systemd (code=exited, status=0/SUCCESS)

10月 20 09:13:16 67-216-220-130 systemd[1]: Failed to start MySQL Server.
10月 20 09:13:16 67-216-220-130 systemd[1]: Unit mysqld.service entered failed state.
10月 20 09:13:16 67-216-220-130 systemd[1]: mysqld.service failed.
10月 20 09:13:16 67-216-220-130 systemd[1]: mysqld.service holdoff time over, scheduling restart.
10月 20 09:13:16 67-216-220-130 systemd[1]: start request repeated too quickly for mysqld.service
10月 20 09:13:16 67-216-220-130 systemd[1]: Failed to start MySQL Server.
10月 20 09:13:16 67-216-220-130 systemd[1]: Unit mysqld.service entered failed state.
10月 20 09:13:16 67-216-220-130 systemd[1]: mysqld.service failed.

之后查看 MySQL 日志：

# tail -n 100 /var/log/mysqld.log
......
2017-10-20T01:09:48.647161Z 0 [Note] InnoDB: Buffer pool(s) load completed at 171020  9:09:48
2017-10-20T01:09:48.647236Z 0 [Note] InnoDB: Starting shutdown...
2017-10-20T01:09:48.747574Z 0 [Note] InnoDB: Dumping buffer pool(s) to /var/lib/mysql/ib_buffer_pool
2017-10-20T01:09:48.747891Z 0 [ERROR] InnoDB: Cannot open '/var/lib/mysql/ib_buffer_pool.incomplete' for writing: Permission denied
2017-10-20T01:09:50.258312Z 0 [Note] InnoDB: Shutdown completed; log sequence number 4076122
2017-10-20T01:09:50.258371Z 0 [ERROR] InnoDB: Operating system error number 13 in a file operation.
2017-10-20T01:09:50.258378Z 0 [ERROR] InnoDB: The error means mysqld does not have the access rights to the directory.
......

日志显示 '/var/lib/mysql/ib_buffer_pool.incomplete' 没有写权限。看来是权限的问题。于是执行如下命令，查看所有者及目前的权限：

# ls -ld /var/lib/mysql
drwxr-x--x 6 mysql mysql 4096 10月 20 09:17 /var/lib/mysql

看来所有者没有问题，那么尝试为其子目录和文件修改所有者：

# chown mysql:mysql -R /var/lib/mysql

之后，重新启动 MySQL 服务：

# systemctl restart mysqld

MySQL 可以正常启动了。

============================================================
若查看 MySQL Log 提示的错误像下面这样：

# tail -n 100 /var/log/mysqld.log
......
2017-10-20T02:58:09.562816Z 0 [Note] InnoDB: Initializing buffer pool, total size = 96M, instances = 1, chunk size = 96M
2017-10-20T02:58:09.562860Z 0 [ERROR] InnoDB: mmap(103071744 bytes) failed; errno 12
2017-10-20T02:58:09.562867Z 0 [ERROR] InnoDB: Cannot allocate memory for the buffer pool
2017-10-20T02:58:09.562872Z 0 [ERROR] InnoDB: Plugin initialization aborted with error Generic error
2017-10-20T02:58:09.562878Z 0 [ERROR] Plugin 'InnoDB' init function returned error.
2017-10-20T02:58:09.562882Z 0 [ERROR] Plugin 'InnoDB' registration as a STORAGE ENGINE failed.
2017-10-20T02:58:09.562886Z 0 [ERROR] Failed to initialize plugins.
2017-10-20T02:58:09.562889Z 0 [ERROR] Aborting

2017-10-20T02:58:09.562907Z 0 [Note] Binlog end
......

说明 MySQL 启动时初始化的缓存大小太大了（也就是说可能是你的 VPS 内在太小了导致的）。默认的缓存大小是 128M，但是我已经改成了 96M 还是报错，说明还是设置的太大了（我的 VPS 只有 512M 内存，但是上面运行了好多的服务，因此可用内存不足）。

因此继续缩小缓存大小：

# vim /etc/my.cnf

修改如下值 innodb_buffer_pool_size = 64M（设置成合适的大小）

之后重新启动 MySQL 就可以了。