Kubernetes Networking & Troubleshooting in Production

Your pods are healthy, your CPU is flat, and checkout is still returning 500s. The dashboard says nothing is wrong. The packet disagrees.

Kubernetes Networking & Troubleshooting in Production is the field manual for the failures dashboards don't show: pod-to-pod dead while pod-to-internet works, MTU black holes that read as "slow", DNS resolution loops, and NetworkPolicies that deny traffic they should allow.

Every chapter runs on the same six-node EKS cluster, so the commands you copy are the commands that worked:

• Chapter 1 maps the four packet paths - pod to pod, pod to Service, pod to ingress, pod to the internet - and the exact hop where each one dies
• Chapter 2 pits Calico, Cilium, and the cloud-provider CNIs head-to-head: what each one actually programs, and the MTU math that follows
• Chapters 3 and 4: CoreDNS drops and resolution loops, then ingress-nginx SSL termination and the headers it loses
• Chapter 5 adds the service mesh only where it earns its keep; Chapter 6 turns NetworkPolicy into real zero-trust microsegmentation
• Chapter 7 traces dropped packets in real time with eBPF - no tcpdump required
• Chapter 8 connects a second cluster over a transit gateway without a CIDR collision
• Chapter 9: ten production war stories, in incident order - what the page said, what the packet showed, what fixed it
• Chapter 10: 14 alert thresholds and a 10-minute first-response runbook you can deploy the day you finish

Written by an SRE with five years of on-call experience. No cluster redefined between chapters, no "just add a service mesh", no hand-waving. If you're the person the page comes to, this is the book you reach for.