We need to separate out the node creation call from the call that initiates introspection.

If introspection fails, for example because we mistyped the ipmi password, then it leaves behind a node object on the Director that will never come up but will block any other attempts to register that machine because the mac address conflicts. Even if we had a way to delete nodes (see issue 44) there's not a good way to know which node to delete because we never had the uuid returned, just the exception from introspection. If the two calls are separated then we can be more intelligent about retrying introspection if that makes sense or deleting the node and allowing the user to re-enter information.

We need a way to delete a node through egon. I tried adding a simple method to nodes.rb, but that didn't work, so I suspect there's some magic going on here that I'm not aware of.

def delete_node(node_id)
  service('Baremetal').nodes.delete(node_id)
end

When I do a vagrant up on our devel environment I see these errors. I suspect it's complaining about wanting a #!/usr/bin/env ruby or something like that.

==> default: WARNING:  no description specified
==> default: WARNING:  bin/undercloud-install-local.rb is missing #! line
==> default: WARNING:  bin/undercloud-install-satellite.rb is missing #! line
==> default: WARNING:  bin/undercloud-install-vanilla-rhel.rb is missing #! line