Comment 2 for bug 961226

I've decided to turn this into a config option and leave it off by default (block on resize2fs).

this is pretty much worst possible scenario for io
 * on disk file was stored as 1.4G ish (possibly larger, but sparse)
 * hypervisor grows it to some larger size (10 or 20G or something) but probably leaves it sparse.
 * initramfs grew the partition table on the disk
 * resize2fs is run, which does loads of reads and writes
 * at the same time the system is under intense IO to boot itself.

The end result is that it changes people's expectation and possibly doesn't save much (my very small number of non-scientific tests showed small savings at best). But clearly, if you were resizing to 100G, that might take 5 minutes, and you might be able to finish boot and use the system during 4 of that 5.

So, I've just committed a change to allow this to be set by a user by setting cloud-config 'resize_rootfs' to a value of 'noblock'.