I've been operating a Kubernetes cluster for a while now, and I'm quite happy with it. It does what I want. However, the complexity cost is quite high. That's especially true when developers who without much k8s experience need to interact with a local or staging cluster somehow to debug applications. And of course, the YAML and package management is hell despite (and partly because of) the many templating tools out there.
I'm wondering if Nomad could remove some of that complexity or whether it's still early stage and a production setup is just as complex. I'm also wondering how easy it is to deploy common applications (Postgres, Traefik, Prometheus, etc) on Nomad - easy Helm installations are a big selling point of k8s.
We run a local instance of send.firefox.com, redis, metabase, docker registry, a speedtest tool, a bunch of batch/periodic jobs, some internal custom services, all of our web hosting, etc directly out of nomad.
Our CI/CD runs make which builds the code, generates .nomad files and runs them, so it's super easy to re-deploy, rollback, etc.
k8s management on your own hardware is a full-time job. There are 2 of us that do all dev and ops for our product and there is 0% chance we could manage a k8s cluster w/ only 2 of us, but managing a nomad, consul and vault cluster is easy for us.
There are nomad templating tools, but we started using nomad back when it was brand new, so we just used sed for our meagre templating needs. Works a treat, and never saw any value to using some specialized tool(for our needs).
Totally recommend checking out nomad.