Let's look at using Ansible to remove this first VM. Exit out of the user ansible.

destroy_vm.yml

---
- hosts: localhost  # Run on the local machine
  connection: local
  gather_facts: false
  vars_files:
    - variables.yml
  vars:
    vm_name: my_vm

  tasks:
  - name: Remove any old VM with this name
    command: "{{ Global.vboxmanage_path }} unregistervm --delete {{ vm_name }}"

Try this with the VM running. Note how it fails.

ansible-playbook destroy_vm.yml

destroy_vm.yml

---
- hosts: localhost  # Run on the local machine
  connection: local
  gather_facts: false
  vars_files:
    - variables.yml
  vars:
    vm_name: my_vm

  tasks:
  - name: Shut down VM with this name
    command: "{{ Global.vboxmanage_path }} controlvm {{ vm_name }} acpipowerbutton"
  - name: Remove any old VM with this name
    command: "{{ Global.vboxmanage_path }} unregistervm --delete {{ vm_name }}"

Try this version with the VM running. Note how it fails to work as expected. The VM stops, but is not removed. If you run it a second time, it fails on the power off command.

What ideas do you have to make the script work, shutting down the VM if necessary before removing it? There are many ways.

Bring the VM back up using vboxmanage startvm my_vm.

:!:Important if you are using a remote SSH session, use vboxmanage startvm my_vm –type headless. You can't start a gui session over ssh.

Instead of the graceful poweroff method above, you can also experiment with vboxmanage controlvm my_vm poweroff. Does this take less time than a graceful shutdown?

Research another option, vboxmanage controlvm my_vm poweroff –type emergencystop. Why would you only use this option if you didn't care of the VM was corrupted?

This is a quick and dirty solution, no better than running the commands manually.

To run the playbook below: ansible-playbook destroy_vm.yml

destroy_vm.yml

---
- hosts: localhost  # Run on the local machine
  connection: local
  gather_facts: false
  vars_files:
    - variables.yml
  vars:
    vm_name: my_vm

  tasks:
  - name: Shut down VM with this name
    command: "{{ Global.vboxmanage_path }} controlvm {{ vm_name }} acpipowerbutton"
  - name: Pause 1 minute
    pause:
      minutes: 1
  - name: Remove any old VM with this name
    command: "{{ Global.vboxmanage_path }} unregistervm --delete {{ vm_name }}"

Oracle Virtualbox doesn't have the nice modules for Ansible that VMware vCenter has. The GUI interface is the easiest way to manage virtual machines, but we can do the changes by command line.

Ansible can do the logic we want, but it's not straightforward:

1. Is the VM powered on? How can we check that with Ansible?
   • If yes, power it off
2. Is the VM now off?
3. Delete the VM
4. And yes, error handling

wait_for is a useful command. You wait for an ssh probe on stop working on the server. Or we could store the VM servers we create in a local table by VM, hostname name and IP. Our scripts could use that data.

OK, we did the easy bit first. When you are ready for a challenge, continue to Deploy a Fleet of VMs.

Or back to Test the First Ubuntu VM on Ansible