Alan Rominger

@matburt I want to ask your thoughts on how we might implement this because it touches back to the original design of instance group preferences. This would be a non-obvious...

Yes, that's exactly why. I am assuming that a merge of `task_manager_refactor` branch is imminent and that most things in this file have moved.

From manual testing:  But this quickly raised some new questions - in particular, if I were to _disable_ the node awx_1, the job should _not_ be...

No, this still has hazards, because if `execution_node` is dead, then we need to actually cancel the job (or else it may flip right back to successful/error). Perhaps, for now,...

Closing b/c this is too risky and not developed enough, and not possible to finish near-term.

@fosterseth after looking into the other option, I believe PR is the right path forward. I will wait for your work to land and then rebase this.

After this merges, we should revisit https://github.com/ansible/awx/issues/1823, as it may become trivial.

I've been working with @jay-steurer to develop complete integration testing of nested git submodules, which we've never had before. We currently have testing of a few permutations of submodules, but...

Structure of unit tests look great, I'd love for @fosterseth to see them

It's known behavior that deleting hosts can take a long time (or even [error](https://github.com/ansible/awx/issues/12176)) due to the related job host summaries, as well as related events. The host summaries give...