Does declarative config actually replace an RMM?

An RMM is necessary software. If you’re managing a fleet of machines there isn’t really an alternative to a good one, and I’d rather not pretend the declarative-everything story changes that.

For my own homelab fleet, Colmena does the job. Every host is a closure I deploy and roll back from one place, so I don’t reach for an RMM at all. But Colmena isn’t an RMM, and it’s worth being honest about why. It manages machines I own, built to be uniform, that I understand top to bottom. It says nothing about a fleet of normal users’ machines that each do one specific job and still need monitoring, remote hands, and support.

That gap is the idea NixRMM was circling: an enterprise-style configuration-management tool for ultra-scoped machines, boxes that only do one job, run by normal users. After living with Colmena I’m honestly not sure what’s left for it to be, or whether a declarative model buys enough over a real RMM at that scale to be worth building.

So the open question, genuinely unresolved: is there a version of declarative fleet management that competes with a good RMM for normal-user machines, or does the declarative approach only win when you already own and understand every box?