Exercise 2: Why Ansible Doesn't Run Continuously Like Kubernetes — Possible Solution ==================================================================== Explanation: This comes down to a genuine architectural difference the chapter names directly, not a missing feature Ansible simply hasn't gotten around to building. Kubernetes' own reconciliation loop is designed for managing a LIVE, continuously running system -- pods can crash, get rescheduled, or drift from their desired replica count at any moment, and something needs to be constantly watching and correcting that in real time, with no human involved, for the cluster to stay healthy. That's the specific problem Kubernetes exists to solve, and its own architecture (a controller running permanently, always comparing desired vs. actual state) is built around solving it continuously. Ansible is solving a different problem: applying a defined CONFIGURATION to a set of hosts, typically as a deliberate, reviewed action -- either run on demand by a person, or on a schedule via CI. Ansible's own idempotent design means it's SAFE to run repeatedly without unwanted side effects, but "safe to run repeatedly" is a different guarantee than "runs constantly and catches drift the instant it happens." Building Ansible to run continuously in the background would mean turning it into something closer to an agent-based tool with a persistent daemon on every managed host -- which is exactly the architecture ansible1-1 already contrasted Ansible against (Puppet/Chef). Ansible's whole agentless, push-based design is a deliberate tradeoff: no persistent process anywhere, at the cost of not catching drift until the next time someone explicitly runs a playbook. WHY THIS WORKS AS AN ANSWER ------------------------------ This explains the difference by connecting it to what each tool was actually built to manage (a live running cluster vs. a config-management task run on demand), and ties the "why not run continuously" question back to ansible1-1's own agentless-vs-agent-based architecture tradeoff rather than treating it as an arbitrary design choice.