So just started looking into Docker Swarm with the new 1.12 release notes. My use case is that right now we have various schedule jobs that get fired off on two servers dedicated to that purpose. My idea was to convert this to a docker swarm and have a central point run all jobs as tasks in the swarm which would then schedule them between the available machines.
With the <1.11 documentation it seems this would work, but with the new service concept I’m struggling to see how that’d accomplish what I’m trying to do. I don’t think I’d want to define each job as a “service” as they should just run once and then complete (ideally removing their containers), and I’d never run more than one instance of it.
I certainly have other use cases where services would work well, but I’m failing to see how I’d translate this use case to swarm, or am I not thinking about something correctly?