Linux 自动备份脚本

写了个用于 Linux 备份的脚本。

Bash 脚本

其中几个用于配置的常量：

EXCLUDE：所有不备份的绝对路径，如果路径含有空格，应使用 '' 引用。
FILE_NAME_PREFIX：备份的文件名前缀。
FILE_NAME：备份的文件名（不含后缀），默认是文件名前缀+时间。
TEMP_PATH：临时文件（.tar）存放目录。
BACKUP_PATH：备份文件（.tar.xz）存放目录。
BACKUP_MAX：保留备份文件的最大数目，超过时将清理旧备份。
COPY_TO_REMOTE：是否复制到远程目录。
REMOTE_PATH：备份文件（.tar.xz）远程存放目录，例如使用 Rclone 挂载的路径。

#!/usr/bin/env bash

set -o pipefail

EXCLUDE="/dev /lost+found /media /mnt /proc /run /sys /tmp /var/cache /var/tmp /root/.cache"

FILE_NAME_PREFIX="debian"
FILE_NAME="${FILE_NAME_PREFIX}_$(date +%Y%m%d-%H%M)"

TEMP_PATH="/tmp"
BACKUP_PATH="/mnt/hdd/Backups"
BACKUP_MAX=5

COPY_TO_REMOTE=false
REMOTE_PATH="/mnt/onedrive/debian-backup"

log() {
  while read -r; do
    echo "[$(date +%F-%T)] $REPLY"
  done
}

# CHECK

[ "$(whoami)" != root ] && echo "Please run as root" | log && exit 1

[ -d $TEMP_PATH ] || (rm -f $TEMP_PATH && mkdir -p $TEMP_PATH)

[ -d $BACKUP_PATH ] || (rm -f $BACKUP_PATH && mkdir -p $BACKUP_PATH)

check_space() {
  local avail
  avail=$(df -k "$1" | sed -n "2p" | awk '{ print $4 }')
  # reserve 512MiB
  [ $((avail - $2)) -gt $((1024 * 512)) ]
}

EXCLUDE_OPT=""
for f in $EXCLUDE; do EXCLUDE_OPT="$EXCLUDE_OPT--exclude='$f' "; done

_du="du -sh $EXCLUDE_OPT /"
size=$(eval "$_du" | awk '{ print $1 }')
echo "Total size: $size" | log
_du="du -sk $EXCLUDE_OPT /"
size=$(eval "$_du" | awk '{ print $1 }')

tmp_path=$TEMP_PATH
if ! check_space $tmp_path "$size"; then
  echo "No enough space: $tmp_path" | log
  echo "Using: $BACKUP_PATH" | log
  tmp_path=$BACKUP_PATH
  if ! check_space $tmp_path "$size"; then
    echo "No enough space: $tmp_path" | log
    df -h $tmp_path | log
    exit 1
  fi
fi

# ARCHIVE

TARBALL="$tmp_path/$FILE_NAME.tar"
_tar="tar -cpf '$TARBALL' $EXCLUDE_OPT /"
echo "+ $_tar" | log
if (eval "$_tar" 2>&1) | log; then
  echo "Successfully saved as:" | log
  ls -lh "$TARBALL" | log
else
  echo "Failed to archive" | log
  rm -f "$TARBALL"
  exit 1
fi

# COMPRESS

BACKUP="$BACKUP_PATH/$FILE_NAME.tar.xz"
_xz="xz '$TARBALL' -T 0 -M 80% -c >'$BACKUP'"
echo "+ $_xz" | log
if eval "$_xz"; then
  echo "Successfully saved as:" | log
  ls -lh "$BACKUP" | log
  echo "Removing: $TARBALL" | log
  rm -f "$TARBALL"
else
  echo "Failed to compress" | log
  echo "Removing: $TARBALL" | log
  rm -f "$TARBALL"
  echo "Removing: $BACKUP" | log
  rm -f "$BACKUP"
  exit 1
fi

# REMOVE OUTDATED

count=$(find "$BACKUP_PATH" -maxdepth 1 -name "$FILE_NAME_PREFIX*.tar.xz" | wc -l)
for f in "$BACKUP_PATH/$FILE_NAME_PREFIX"*.tar.xz; do
  [ "$count" -le $BACKUP_MAX ] && break
  echo "Removing outdated: $f" | log
  rm -f "$f"
  ((count -= 1))
done

# COPY

[ $COPY_TO_REMOTE != true ] && exit 0

[ -d $REMOTE_PATH ] || (rm -f $REMOTE_PATH && mkdir -p $REMOTE_PATH)
size=$(du -sk "$BACKUP" | awk '{ print $1 }')
if ! check_space $REMOTE_PATH "$size"; then
  echo "No enough space:" | log
  df -h $REMOTE_PATH | log
  exit 1
fi

BACKUP_REMOTE="$REMOTE_PATH/$FILE_NAME.tar.xz"
echo "Copying to remote: $BACKUP_REMOTE" | log
if cp "$BACKUP" "$BACKUP_REMOTE" 2>&1 | log; then
  echo "Successfully saved as:" | log
  ls -lh "$BACKUP_REMOTE" | log
else
  echo "Failed to copy" | log
  rm -f "$BACKUP_REMOTE"
  exit 1
fi

总体过程是，先用 tar 打包存放到 TEMP_PATH，再使用 xz 压缩为最终备份文件并存放到 BACKUP_PATH，如果超出最大数则清理旧备份，最后复制到远程目录。

将脚本放到合适的路径下（如 /root/.local/bin/）并添加执行权限。

systemd 定时任务

~~将以下两个文件存放到 /lib/systemd/system/ 下：~~

linux-backup-script/backup.service

[Unit]
Description=Run backup

[Service]
ExecStart=/bin/bash -c "/root/bin/backup.sh >>/var/log/backup.log 2>&1 && echo >>/var/log/backup.log"

linux-backup-script/backup.timer

[Unit]
Description=Run backup weekly

[Timer]
OnCalendar=Mon, 03:00
AccuracySec=1h
Persistent=true

[Install]
WantedBy=timers.target

~~这里设置为每周一 03:00 开始备份，AccuracySec=1h 表示可以有一小时的容错，这是为了减少唤醒 CPU。~~

~~运行并使其开机启动：~~

$ sudo systemctl daemon-reload
$ sudo systemctl start backup.timer
$ sudo systemctl enable backup.timer

~~systemctl list-timers 可以查看所有定时任务。~~

cron 定时任务

systemd 太**难用了，还是 cron 好使，crontab -e 加入：

0 3 * * 1 /root/.local/bin/backup.sh >>/var/log/backup.log 2>&1