Redundant Cluster For Concurrent Jobs

I have a bunch of physical servers that I want to run together to run batch jobs. If one of these servers goes down I don’t want it to affect the overall operation of the jobs. The container images would be built using Jenkins (preferably in a redundant Jenkins cluster. Essentially I would be using the physical servers as horse power to do heavy calculations.

I’ve looked at Swarm and Mesos but not sure how these would work (together?).

What’s the quickest way to build a cluster of containers across multiple hosts so that they are resilient and redundant?