Run Cluster Autoscaler on Your Own Cloud; or, Why Do We Need Cluster API?
summary
Explains exactly what must be implemented to run Cluster Autoscaler on an unsupported cloud: a cloud provider module that uses Kubernetes API plus the cloud’s API, ensures nodes have correct providerID, can list cluster instances, and can create and delete specific VMs. Covers extra features: scale-from-zero via a retained fake node template and a pricing expander for cost-aware scaling. Describes dependency and vendoring pain in the binary and why Cluster API would help once it supports atomic deletion of a specific machine.
AI-written summary · about
chapters (5)
- 0:00Start
- 0:25What Is Cluster Autoscaler
- 2:57Problem One Is How Do We Merge Node to a Vm Instance
- 11:24Critical Requirements
- 20:48Dependencies
related talks
39mCluster API & Cluster Autoscaler
30mBuild Your Own Cluster API Provider the Easy Way
32mBeyond CAS: Why the world needs another Kubernetes Cluster Autoscaler
21mWhy should you bother about cluster autoscaling- Karpenter
25mIntroduction To Cluster API
25mCluster API: IaC for Kubernetes
34mInfrastructure Provider for Cluster API: From Zero to Open Source
36mCluster Autoscaler Evolution