Local Autoscaling Limits

BookNest's cluster has one worker, a container sharing the host's four CPUs with the control plane and everything else. kind 14,561 cannot add a node: when the HPA wants more Pods than the worker's allocatable CPU holds, the extras stay Pending with Insufficient cpu (Scheduler), the signal a cloud node autoscaler would act on. More replicas on one machine share the same cores, so they spread load across processes, not hardware. Metrics and the HPA each tick every 15 seconds on a shared host, so read the timings in Autoscaling the API as approximate. The control loop itself, its arithmetic and its windows behave exactly as on a hundred nodes.