Lots of experimentation. The current focus is on:
- CoreOS, Fleet, and ETCD compatible
- make sure the components can be reused in other scheduling systems, such as AWS ECS, Kubernetes, Docker Storm, etc.
Please contribute thoughts, ideas, or PRs via the usual github channels. Thanks!