job: ensure that job is running

[rodrigol-chan]

Refs #1. It's sometimes handy to use the UI to stop jobs e.g. during a maintenance event. In this specific case, the provider should check if the job is no longer running and resubmit it. I'm open to suggestions on how to approach this better.

[hashicorp-cla]

All committers have signed the CLA.

[rodrigol-chan]

@lgfa29 Sorry for the ping, but is there interest in this? We would really like to solve the "ensure job is running" problem.

[lgfa29]

Hi @rodrigol-chan 👋

Apologies for the delay in review the PR. We're wrapping work for Nomad 1.7, which has kept us busy for the past weeks. I will try to review this some time next week.

Thanks for raising it!

[lgfa29]

Thanks for the PR @rodrigol-chan, and apologies again for the delay here.

I think the changes look good, but I do worry about this breaking backwards compatibility in case someone is relying on the existing behaviour.

I added a new flag called rerun_if_dead to allow for this new behaviour to be opt-in but I couldn't push the change to your branch 😅

Since this PR is atomic enough I will merge it as-is and then open a follow-up PR for the new attribute.

Thank you again for the contribution!