Ensure one VM ("gateway") is running before allowing the others

Vagrantfile

# Check to see that we have a gateway running before we allow the `up` of another VM
# This is a filthy hack
if ARGV[0] != 'up'
  GATEWAY_STATUS = 'running'  # this is irrelevant for everything but `vagrant up`
  # if we don't do this, we could fork forever in the else block, here
else
  GATEWAY_STATUS = `vagrant status --machine-readable | grep 'gateway,state,' | awk -F, {'print $4'}`.strip!
  if GATEWAY_STATUS != 'running'
    puts "*** The gateway is not running. You can not 'up' any other VMs."
  end
end

Vagrant.configure(2) do |config|
  if GATEWAY_STATUS == 'running'
    config.vm.define :othervm, primary: false, autostart: false do |config|

      bootstrap config.vm  # custom

    end
  end

  if GATEWAY_STATUS == 'running'
    config.vm.define :yetanothervm, primary: false, autostart: false do |config|

      bootstrap config.vm  # custom

    end
  end


  config.vm.define :gateway, primary: true, autostart: true do |config|

    bootstrap config.vm  # custom

  end


end

Someone tell me why this is a bad idea, and preferably: how to do it via internal APIs. (-:

Does it matter if the gateway is up? What you really care about is connectivity over your VPN.Though I suppose the conditional is the really ugly part.

Ha. Just noticed your comment @nicksloan. It does matter because we run apt-cacher-ng on the gateway, so other nodes "need" it to be running.