K8s NodeSet Managing Worker Nodes with Kubernetes
summary
Introduces a Kubernetes-native way to provision worker machines using a NodeClass resource and a NodeSet controller that speaks to cloud APIs via Docker Machine drivers (AWS, GCP, DigitalOcean, Packet/OpenMetal, OpenStack or bare metal). Demonstrates kubectl scaling from 1 to 21 nodes, integrates with autoscaling logic so nodes are created and deleted to match replicas and HPA pressure, and publishes the implementation as open source on GitHub where it already works in the demo.
AI-written summary · about
chapters (9)
- 0:00Start
- 2:15Managing the Nodes with Kubernetes
- 3:01Deploying Kubernetes
- 6:11Persistent Volumes
- 7:03Pets versus Cattle Semantics
- 9:27Node Controller
- 13:51Demo
- 24:57Questions
- 25:38The Horizontal Pod Autoscaler
related talks
38mLet a Kubernetes cluster manage its worker nodes by itself
28mScaling Node.js Applications with Kubernetes and Docker
31mKubernetes workloads with Karpenter & Amazon EKS
19mJust in time Worker Nodes with Karpenter
21mScaling Node.js applications with Kubernetes and Docker
23mVirtual nodes to auto-scale applications on Kubernetes
38mNode Operator: Kubernetes Node Management Made Simple
11mCost and Resource Optimization In Kubernetes
