Self-healing Clusters: Game of Nodes and the Scaling Throne

Talk by Tyler Gillson

In this session, we delve into how Descheduler, Node Problem Detector (NPD), and Autoscaler combine their strengths to deal with problematic nodes efficiently. We'll simulate a real-world scenario where a Kubernetes cluster faces turbulence caused by unhealthy nodes. Here, the Descheduler plays the role of an efficient manager, the NPD acts as the problem finder, and the Autoscaler balances the cluster's size. Together, they detect, remove, and maintain the right scale of the cluster.