Harnessing Karpenter: Transforming Kubernetes Clusters with Argo Workflows
summary
Practical migration path from Cluster Autoscaler to Karpenter using Argo Workflows with code and a demo: automated extraction of node group metadata into node pool and node class YAMLs, Python SDK tasks that create Karpenter resources, respect PodDisruptionBudgets, and offer a roll back workflow to restore original node groups. A live run consolidates ~100 legacy nodes into 29 nodes, uses Karpenter’s cost-aware instance selection, supports gradual or all-at-once migration, and completes successfully in the demo.
AI-written summary · about
chapters (11)
- 0:00Karpenter and Argo Workflows intro
- 1:05Cluster autoscaler limitations
- 2:50Karpenter autoscaling benefits
- 4:42Karpenter NodePools config
- 7:35Migration challenges real-world
- 9:47Karpenter migration steps
- 11:32Demo setup overview
- 13:20Self-service migration with Argo
- 17:41Node group metadata extraction
- 19:56Argo Workflows for automation
- 21:00Resources and next steps
related talks
19mKarpenter, the Future of Kubernetes Node Management
22mKarpenter Efficient scaling of Kubernetes clusters
27mHarness the power of Karpenter to scale, optimize & upgrade K8s
22mUse Karpenter to scale, optimize & upgrade K8s
18mKarpenter : Your trusted Companion for Kubernetes Scaling
33mNavigating and Optimizing Karpenter - Avoiding Common Pitfalls
19mJust in time Worker Nodes with Karpenter
33mKarpenter: Best Practices and Cost Optimization Techniques