Kubernetes Scheduling Visualized: The Filter-Score Pipeline

Scheduling looks like magic until you watch the pipeline. This video walks through what happens between kubectl apply and a Pod's first container start: how the scheduler watches the API server for unscheduled Pods, filters candidate Nodes, scores them, and writes the binding. We visualize the real mechanics — node conditions, resource requests versus limits, taints and tolerations, affinity and anti-affinity, and how the kubelet's admission plays out against the schedule. Then we look at what breaks: scheduler starvation, resource fragmentation, and how preemption interacts with priorities.

Topics covered:

• The scheduler's filter-then-score pipeline
• Requests, limits, and how resources are accounted
• Taints, tolerations, affinity, and anti-affinity
• Preemption, starvation, and scheduling failures