Sometimes we get to a point where a Vagrant VM just won't run, and there's nothing for it but to get rid of it.
First get the ID of the machine in question:
cat .vagrant
Output will be something like:
{"active":{"hadoop_client":"31c1a009-f1b0-4acc-8200-1f33b95a0e3b"}}
To find out which directory this VM is using, do:
vboxmanage list vms
Output will be something like:
"vagrant_1310474995" {31c1a009-f1b0-4acc-8200-1f33b95a0e3b}
Now ask VBox to shut the instance down (Vagrant can't manage it) and unregister the VM:
vboxmanage controlvm {31c1a009-f1b0-4acc-8200-1f33b95a0e3b} poweroff
vboxmanage unregistervm {31c1a009-f1b0-4acc-8200-1f33b95a0e3b}
You can now delete the directory vagrant_1310474995
and start again with a fresh instance.