Issue Type

• Feature request

Molecule and Ansible details

ansible 2.8.7
  config file = /etc/ansible/ansible.cfg
  configured module search path = ['/home/user/.ansible/plugins/modules', '/usr/share/ansible/plugins/modules']
  ansible python module location = /usr/local/lib/python3.5/dist-packages/ansible
  executable location = /usr/local/bin/ansible
  python version = 3.5.2 (default, Oct  8 2019, 13:06:37) [GCC 5.4.0 20160609]
molecule, version 2.22

Molecule installation method (one of):

• pip Ansible installation method (one of):

• pip

Desired Behavior

I Would like to request/suggest new Strategies for the --destroy option in the test command.

In my development work-flow I usually want to inspect the Machines created by molecule, when a Failure in i.E. in the Converge Step occurred.

Having a on-failure, on-success or an interactive ask strategy would allow to accommodate for different work-flows.

The on-failure strategy would destroy the machines, if any of the previous steps failed, the on-success strategy would destroy the machines upon successfully running all Steps. the ask strategy would interactively ask for the desired action, upon failure.

If this feature/request is considered useful, I am willing to try and take on this Issue myself.

To me on-failure would seem to be the most useful default as well - at the moment when test fails you have to either add --destroy=never or go through the steps manually.

I would't change the default because that would break expectations in CI workflows (always tear it down). I do sympathise with the development workflow though and this is definitely an issue for beginners ("Where is the box gone? Oh what? I need to run it all over again!?"). One thought is to use the proposed ask strategy as default when we detect a tty (dev mode) but otherwise, destroy all. However, introducing an interactive prompt feels a bit meh... maybe we can at least introduce schema configuration options for new strategies as a first step? AFAIR, there is just --destroy=never on the CLI and nothing else?

Yeah fair enough re breaking CI! CLI currently only has =never and =always (see #3009). Personally I'd be fine with no autodetection that we're in a "dev mode" and I'd just learn to put --destroy=on-failure or on-error or whatever it is on the CLI. If it could get set via an environment var I guess that might help.

Then maybe keep =never to preserve existing expectations and add on-failure/on-success strategy options, also configurable by env and please send a bit of docs love with the change and I'll be sure to give it a review. That's my take on this anyway.