I am curious as to how this sort of setup would work for processes that take a bit to load - such as when building an in-memory index. In my case, I have several services that take ~30s+ to start, and as such do a rolling restart. envconsul would just sigterm an entire tier, causing outages.

envconsul is a fairly simple UNIX tool. Instead of executing your application directly with it, you can put a rolling restart script on it to do this sort of thing. This answer is a bit of a cop out, but it is something that works today. Example:

    envconsul -reload prefix bin/restart-wrapper bin/rails server

We have more advanced orchestration features planned in the future, but we're not ready to discuss them yet.