Pod Scheduling: How Kubernetes Decides Where Pods Run

A Pod stuck Pending with 0/3 nodes are available in the events is a scheduling failure, and the scheduler is trying to tell you exactly which control blocked it. There are four orthogonal placement controls: resource requests (capacity filter), node affinity (attraction), taints and tolerations (repulsion), and priority classes (preemption order). Each operates at a different point in the scheduler pipeline, and getting one wrong doesn't fail loudly — the Pod just stays Pending.

This lesson maps the full scheduling landscape. Each of the four controls gets an overview here; the three largest topics — resource management, affinity, and taints — have dedicated deep-dive lessons that follow.