标签：
当我们租用Linux云主机时，如果服务即将到期，怎样彻底销毁你所租用Linux云主机呢？本文所转载的shell脚本可以非常彻底地销毁Linux云主机，并且对当前所使用硬盘采用执行N次覆盖写操作对数据进行彻底破坏（当然，对备份和快照没有作用）。
以下是shell脚本内容：
#!/bin/bash
#
# This script will absolutely kill a RHEL/CentOS/Fedora server. Use with extreme caution.
# Tested with several CentOS/RHEL versions only. Run as root user.
# 10.20.11 Paul Venezia (pvenezia@pvenezia.com)
#
zeroscript="/var/ramdisk/zeroscript.sh"
echo "*******************************************************************
** This will permanently kill this Linux system and erase every **
** local disk and filesystem. In other words, you better be **
** REALLY REALLY SURE you want to do this on this system. **
*******************************************************************"
echo -n "Are you absolutely sure you want to do this? [yes|no]: "; read yn
if [ -z $yn ] || [ $yn != "yes" ]; then
echo "Aborting"
exit 1
fi
echo -n "How many zeroing passes? "; read zeropass
if [ -z $zeropass ] || [ $zeropass -lt 1 ]; then
echo "Invalid number of passes specified. Aborting."
exit 1
fi
echo -n "Automatically shutdown? [yes|no] "; read asd
echo "Okay, here we go..."
echo "Making and populating ramdisk (512MB)..."
mkdir -p /var/ramdisk
mount -t tmpfs none /var/ramdisk -o size=512m # You may need to adjust this depending on the amount of RAM in the box
mkdir -p /var/ramdisk/var/run
for f in dev bin lib lib64 sbin etc; do
cp -pr /$f /var/ramdisk
done
cp -pr /var/run /var/ramdisk/var
echo "Stopping services, it‘s probably safe to ignore any errors..."
for s in httpd acpid anacron atd auditd autofs avahi-daemon bluetooth cpuspeed crond cups firstboot gpm haldaemon hidd hplip irqbalance iscsi iscsid kudzu lm_sensors lvm2-monitor mcstrans mdmonitor messagebus microcode_ctl netfs nfslock pcscd portmap rawdevices readahead_early restorecond rpcgssd rpcidmapd sendmail smartd sshd syslog vmware-tools xfs yum-updatesd; do
service $s stop
done
echo "Placing zeroing script..."
echo "#!/bin/bash" > $zeroscript
for i in `fdisk -l | grep Disk | awk ‘{print$2}‘ | sed -e s/:// | grep -v /dev/md`; do
DU=$DU" "$i
DSK=`basename $i`
BLKS=$((`grep -w $DSK /proc/partitions | awk ‘{print$3}‘` * 2)) # account for 512/1k blocksizes
BS=512
echo "echo \"Zeroing $i (dd if=/dev/zero of=$i bs=$BS count=$BLKS) ...\"" >> $zeroscript
for (( c=1; c<=$zeropass; c++ )); do
echo "echo \"Pass $c...\"" >> $zeroscript
echo "dd if=/dev/zero of=$i bs=$BS count=$BLKS" >> $zeroscript
done
echo "dd if=/dev/zero of=$i bs=512 count=1" >> $zeroscript # Just to make sure
done
echo "echo \"Disk(s)$DU have been zeroed $zeropass times\"" >> $zeroscript
if [ $asd = ‘yes‘ ]; then
echo "echo \"Shutting down...\"" >> $zeroscript
echo "sleep 5 && /sbin/poweroff -n -d -f" >> $zeroscript
fi
chmod +x $zeroscript
echo "Turning off swap..." && swapoff -a
echo "Entering chroot..."
chroot /var/ramdisk /`basename $zeroscript`
将上述shell保存为shell文件（如：destroyLinuxOS.sh）并授予可执行权限（chmod +x destroyLinuxOS.sh），执行该脚本时，要依次回答 “yes 、 2 、 yes” 才会真正开始彻底销毁Linux主机的程序，其中第一个 yes 表示同意开始执行shell（防止误执行），“2“为指定“覆盖写”的写次数（需要大于0），最后的 yes 表示完成“覆盖写”操作以后直接关机，如果回答不为 yes 则不关机。
分析上述脚本可知，首先（1）在内存中创建内存磁盘（可在执行上述脚本前根据当前系统的相关文件夹的磁盘占用情况修改上述脚本中的 mount 语句的 size 属性值为合适大小）；然后（2）从当前系统的相应目录中复制相关文件到内存磁盘中；之后（3）开始停止当前系统中的已启动服务（可根据你的主机服务情况进行修改，注意不要停止 sshd 服务）；最后（4）开始输出相关shell脚本到变量 zeroscript 指定的脚本文件中（该脚本用来（a）完成对磁盘进行覆盖写操作 [即：销毁磁盘数据]，根据第三个输入内容决定（b）是否直接关机）；在完成前面四项操作以后，（5）切换系统根路径到内存磁盘，并执行在 变量 zeroscript 中指定的脚本文件。over...
PS: 上述脚本来源于《如何将你的Linux服务器清空》文章所转载的内容。
标签：
原文地址：http://my.oschina.net/yaray/blog/512516
