Deconstructing K8s Networking for Knowledge and Profit
summary
Claims that managed Kubernetes is just VMs and networking is interfaces plus routing rules (mostly iptables), demonstrates differences between kubenet and Azure CNI (route tables present in kubenet, absent in CNI), and walks through a real outage where kubectl, crictl, nsenter, tcpdump, tshark/Wireshark and the hey load generator trace a load-balanced service failing due to the app listening on 127.0.0.1 instead of 0.0.0.0; switching to 0.0.0.0 restores traffic. Troubleshooting favors bottom-up kernel inspection.
AI-written summary · about
chapters (16)
- 0:00Intro
- 2:29Kes Networking: Key Components Kubelet
- 4:23K8s Networking: Key Components
- 7:49Networking Components in Action
- 8:26Time to dig in on a networking issue...
- 8:57High Level Environment Assessment
- 11:11Sidebar: Kubenet vs. Azure CNI
- 14:19Walking the stack - Load Balancer
- 15:06Walking the stack - Service iptables
- 15:51Walking the stack - Endpoints
- 16:54Walking the stack - Pod IP
- 18:17Load Balancer Metrics
- 19:29Packet Flow Test - tshark
- 20:20Packet Flow Test - tcpdump Client
- 20:39Checking for listening ports
- 21:09Application Code
related talks
29mInside Kubernetes Networking
20mKubernetes Networking 101
51mHow Kubernetes Networking Really Works: A Packet’s Journey Across Pods and Nodes
32mUnderstand and Troubleshoot the “Magic” of Kubernetes Networking - Minhan Xia & Rohit Ramkumar
32mThe Easy--Don't Drive Yourself Crazy--Way to Kubernetes Networking
40mCloud, Kubernetes, and Service Networking - Taming the Turtles
34mIt's Not Always the Network (But Here's How to Prove It): Kubernetes Packet...
22mNavigating the Kubernetes Networking Maze