引言

最近博客又抽風了，打開主頁后提示 Error Establishing a Database Connection 。仔細想想，應該就是數據庫服務器 mariadb 掛了；以前也遇到過類似的問題。經過分析日志，并結合網上的資料最終解決了問題。

日志

以下是 mariadb 服務器掛掉時的比較關鍵的日志信息，從下面的日志信息中，我們可以很容易地看出由于內存不足，從而導致數據庫服務器啟動時崩潰。

InnoDB: Starting crash recovery.
InnoDB: Reading tablespace information from the .ibd files...
InnoDB: Restoring possible half-written data pages from the doublewrite
InnoDB: buffer...
160919  2:47:12  InnoDB: Waiting for the background threads to start
160919  2:47:13 Percona XtraDB (http://www.percona.com) 5.5.46-MariaDB-37.6 started; log sequence number 352718445
160919  2:47:13 [ERROR] mysqld: Out of memory (Needed 128917504 bytes)
160919  2:47:13 [Note] Plugin 'FEEDBACK' is disabled.
160919  2:47:13 [Note] Server socket created on IP: '0.0.0.0'.
160919  2:47:13 [Note] Event Scheduler: Loaded 0 events
160919  2:47:13 [Note] /usr/libexec/mysqld: ready for connections.
Version: '5.5.47-MariaDB'  socket: '/var/lib/mysql/mysql.sock'  port: 3306  MariaDB Server
160919 02:47:35 mysqld_safe Number of processes running now: 0
160919 02:47:35 mysqld_safe mysqld restarted
160919  2:47:35 [Note] /usr/libexec/mysqld (mysqld 5.5.47-MariaDB) starting as process 28614 ...
160919  2:47:35 InnoDB: The InnoDB memory heap is disabled
160919  2:47:35 InnoDB: Mutexes and rw_locks use GCC atomic builtins
160919  2:47:35 InnoDB: Compressed tables use zlib 1.2.7
160919  2:47:35 InnoDB: Using Linux native AIO
160919  2:47:35 InnoDB: Initializing buffer pool, size = 128.0M
InnoDB: mmap(137756672 bytes) failed; errno 12
160919  2:47:35 InnoDB: Completed initialization of buffer pool
160919  2:47:35 InnoDB: Fatal error: cannot allocate memory for the buffer pool
160919  2:47:35 [ERROR] Plugin 'InnoDB' init function returned error.
160919  2:47:35 [ERROR] Plugin 'InnoDB' registration as a STORAGE ENGINE failed.
160919  2:47:35 [ERROR] mysqld: Out of memory (Needed 128917504 bytes)
160919  2:47:35 [ERROR] mysqld: Out of memory (Needed 96681984 bytes)
160919  2:47:35 [ERROR] mysqld: Out of memory (Needed 72499200 bytes)
160919  2:47:35 [Note] Plugin 'FEEDBACK' is disabled.
160919  2:47:35 [ERROR] Unknown/unsupported storage engine: InnoDB
160919  2:47:35 [ERROR] Aborting

解決

在使用 free -m 查看內存信息時，發現 swap 分區大小為 0。難怪說數據庫服務器無法啟動呢，在內存不夠用的情況下，又無法使用 swap 分區，自然崩潰了。由于 VPS 使用了 SSD，性能自然不錯。下面我們給服務器系統 CentOS 7 添加 1024M 的 swap 分區，采用的方法是創建一個 swap 文件：

使用下面的命令創建 swapfile ：