[ ] We should wait for database to be ready before starting cron.
	Else, we get an err- file!
[ ] Use 'qcow2 (prealloc metadata)', it is only 6.5MiB.
	Maybe is faster.
[ ] Use 'virsh domifaddrs' to find out the IP addresses?
[ ] Add an install method to stream the image dirrectly + copy-on-read?
[ ] Check other filesystems: nils2 etc for virtual machine.
[ ] Add sha1sum of the VM images
[ ] Sign vm images.
[ ] Control ext4 size of the inodes / number to obtain a smaller image.
	* experiment:
	truncate a.img --size 100M
	tune2fs -l a.img | grep "Inode count" # => 25688; Inode size = 128
	truncate a.img --size 200M
	resize2fs a.img
	tune2fs -l a.img | grep "Inode count" # => 49400
	guestfish: sparsesuper sounds nice
	mkfs.ext4 -N number_of_inodes
	* experiment:
	mkfs.ext4 -N 1000
	tune2fs -l a.img | grep "Inode count" # => 1000; Inode size = 128

	truncate a.img --size 300M
	resize2fs a.img => 1520 inodes!
	So, we must start with a number of inodes.
[ ]