Vertical Pod Autoscaler — Right-Sizing Requests Automatically

Day 25 made you reason about requests and limits by hand — but how do you know the right numbers? Guess low and you OOMKill or get evicted; guess high and you waste the reservation on every node. The Vertical Pod Autoscaler watches a workload's real usage and tells you (or sets for you) the requests it should have. Install VPA, read its recommendation against a deliberately over-provisioned workload, watch it evict-and-resize in Recreate mode — then confront the one rule that matters: VPA and the Day 12 HPA must never both act on CPU, or they fight.