自动化备份方案

背景说明

网上有很多教程，写的都是从零搭建一个什么什么，基本上都是从无到有的教程，但是，很少有文章提及搭建好之后如何备份，这次通过请教GitHub Copilot Chat，生成几个备份脚本，以备后用。

注：本文涉及的所有脚本默认仅针对Linux环境。

备份MySQL

#!/bin/bash

# Set the database name, backup directory, host, username, and password
database=""
backup_directory=""
host=""
username=""
password=""

# Create the backup directory if it doesn't exist
mkdir -p $backup_directory

# Create a timestamped backup file name
backup_file="$backup_directory/$database-$(date +%Y-%m-%d-%H-%M-%S).sql"

# Backup the database to the backup file
mysqldump -h $host -u $username -p$password $database > $backup_file

# Delete backups older than 7 days
find $backup_directory -type f -name "$database-*.sql" -mtime +7 -delete

echo $backup_file

备份SQLite

#!/bin/bash

# Set the database file path and backup directory
database=""
backup_directory=""

# Create the backup directory if it doesn't exist
mkdir -p $backup_directory

# Create a timestamped backup file name
backup_file="$backup_directory/$(basename $database)-$(date +%Y-%m-%d-%H-%M-%S).db"

# Backup the database to the backup file using SQLite's .backup command
sqlite3 $database ".backup $backup_file"

# Delete backups older than 7 days
find $backup_directory -type f -name "*.db" -mtime +7 -delete

echo $backup_file

备份目录

#!/bin/bash

# Set the directory to backup and backup directory
directory=""
backup_directory=""

# Create the backup directory if it doesn't exist
mkdir -p $backup_directory

# Create a timestamped backup file name
backup_file="$backup_directory/$(basename $directory)-$(date +%Y-%m-%d-%H-%M-%S).tar.gz"

# Backup the directory to the backup file
tar -czf $backup_file $directory

# Delete backups older than 7 days
find $backup_directory -type f -name "*.tar.gz" -mtime +7 -delete

echo $backup_file

上传至对象存储

因为平时腾讯云用的比较多，对腾讯云的产品比较熟悉，所以这里就以腾讯云的对象存储为例。

#!/bin/bash
  
read file

prefix=$1
bucket=$2

coscli cp $file "cos://$bucket/$prefix/$(basename $file)"

COSCLI 是腾讯云对象存储（Cloud Object Storage，COS）提供的客户端命令行工具。具体安装、配置参考官方文档。当然也可以选择s5cmd这种兼容s3协议的客户端，支持各家对象存储。

划重点

上面的内容看起来平平无奇，我主要想说的是最后这一段。

上文中的备份脚本有个细节，最后会输出备份文件的路径，目的是为了结合最后一个脚本上传至对象存储。

整体使用逻辑是：备份脚本分别写，上传脚本共用同一个。

这里以备份 /data/test/目录至对象存储的/test/路径下为示例说明具体使用：

backup.sh

#!/bin/bash

directory="/data/test"
backup_directory="/data/backup/test"
mkdir -p $backup_directory
backup_file="$backup_directory/$(basename $directory)-$(date +%Y-%m-%d-%H-%M-%S).tar.gz"
cd $directory
tar -czf $backup_file .
find $backup_directory -type f -name "*.tar.gz" -mtime +7 -delete
echo $backup_file

upload.sh

#!/bin/bash
  
read file

prefix=$1
bucket=$2

coscli cp $file "cos://$bucket/$prefix/$(basename $file)"

备份命令如下：

./backup.sh | ./upload.sh test temp

结合crontab就可以实现定时备份了。

另外，对于对象存储上的内容，可以设置生命周期，自动删除旧的备份。如下图（腾讯的对象存储为例），设置了备份保留30天。