Every flow traces one event end to end — from the command you type to the container that runs, and in Part E, across a whole fleet of clusters. Each has numbered steps, a diagram, and the ways it fails in production.
Flow 8 is the master flow: other chapters in Parts A–D zoom into their segment of it and say so; Part E's flows run above it, across clusters.
Chapter 1 — Architecture Big Picture
Chapter 2 — API Server & etcd Internals
- What happens when two controllers write the same object
- What happens when a watch is established — and falls behind
- What happens when an admission webhook is down
Chapter 3 — Scheduler Internals
- What happens when a pod is created and a node is chosen
- What happens when no node fits
- What happens when a taint is applied to a node
Chapter 4 — Kubelet, Pods & the Node
- What happens when a pod goes from created to Running (master flow)
- What happens when a pod is deleted
- What happens when a node is cordoned
- What happens when a node is drained
- What happens when a node dies
- What happens when a probe fails
- What happens when a node comes under memory pressure
Chapter 5 — Controller Fundamentals
- What happens when you edit or scale a Deployment
- What happens when the HPA scales a Deployment
- What happens when you delete an object that owns others
Chapter 6 — Writing Controllers Well
- What happens when one reconcile runs, end to end
- What happens when a controller pod restarts
- What happens when leader election hands over
- What happens when a pod gets its network (CNI ADD)
- What happens when a pod sends a request to a ClusterIP Service
- What happens when a pod becomes Ready (endpoint propagation)
- What happens when a PVC is created and a pod uses it
- What happens when a pod with a volume moves to another node
Chapter 9 — Runtime & Device Standards: CRI, Device Plugins, DRA