Script to clean up Ubuntu Vagrant box before packaging

vagrant-clean.sh

#!/bin/bash

# This script zeroes out any space not needed for packaging a new Ubuntu Vagrant base box.
# Run the following command in a root shell:
#
# bash <(curl -s https://gist.github.com/justindowning/5670884/raw/vagrant-clean.sh)

function print_green {
  echo -e "\e[32m${1}\e[0m"
}

print_green 'Clean Apt'
apt-get -y autoremove
aptitude clean
aptitude autoclean

print_green 'Cleanup bash history'
unset HISTFILE
[ -f /root/.bash_history ] && rm /root/.bash_history
[ -f /home/vagrant/.bash_history ] && rm /home/vagrant/.bash_history
 
print_green 'Cleanup log files'
find /var/log -type f | while read f; do echo -ne '' > $f; done
 
print_green 'Whiteout root'
count=`df --sync -kP / | tail -n1  | awk -F ' ' '{print $4}'`
let count--
dd if=/dev/zero of=/tmp/whitespace bs=1024 count=$count
rm /tmp/whitespace
 
print_green 'Whiteout /boot'
count=`df --sync -kP /boot | tail -n1 | awk -F ' ' '{print $4}'`
let count--
dd if=/dev/zero of=/boot/whitespace bs=1024 count=$count;
rm /boot/whitespace

print_green 'Whiteout swap'
swappart=`cat /proc/swaps | tail -n1 | awk -F ' ' '{print $1}'`
swapoff $swappart
dd if=/dev/zero of=$swappart
mkswap -f $swappart
swapon $swappart

print_green 'Zero out disk'
dd if=/dev/zero of=/EMPTY bs=1M
rm -f /EMPTY

print_green 'Vagrant cleanup complete!'

you might want to try a 'sync' between the dd and rm -f

Zero out the free space to save space in the final image:

dd if=/dev/zero of=/EMPTY bs=1M
sync
rm -f /EMPTY
sync

If you put @ in your print_green command instead of 1, then you can do

print_green this should output in green

and not have to do the very ugly

print_green 'this should output in green'

This script is like strapping a rocket to your back and going for a sled ride.

Main issues with it are:

1. The swap zeroing bit does not check that there is actually a configured swap device, if there isn't it will think the swap device is a file called Filename.
2. The script does not allow for partition as files instead of devices. If a swap file is encountered, it will be grown to the size of the available free space. While on a roll, multiple swap devices are rare but perfectly legit.

Fix for the above two might be something like:

for swappart in $(grep partition /proc/swaps | awk -F ' ' '{print $1}')
do
  swapoff $swappart
  dd if=/dev/zero of=$swappart
  mkswap -f $swappart
  swapon $swappart
done

(swap file support left as an execise to the reader)

3. The "whiteout of root" step is done over a file in /tmp. /tmp is not always on the root partition, indeed it can be in virtual memory using the tmpfs file system. Inexplicably, this one will try to whiteout 1 block less then available, meanwhile....
4. The 'Zero out disk' step is explicitly going to cause a dd: error writing ‘/EMPTY’: No space left on device error every time. This is actually expected and should be suppressed or warned as it's surprising. Indeed one of the commenter above was surprised that they ran into the error and thought that was what failed for them. Appending a 2>/dev/null to the command-line would suppress that shocking warning.