Autoscaling Can Be Reliable - Running Cluster Autoscaler in Production

Explore the intricacies of running Cluster Autoscaler reliably in production environments in this informative conference talk by Maciej Pytel from Google. Gain insights into managing Kubernetes cluster nodes across 25+ supported cloud providers using Cluster-Autoscaler. Learn about potential challenges in node management, including boot-up failures, resource quota issues, and provider capacity limitations. Discover which problems Cluster Autoscaler can handle autonomously and which require manual intervention. Benefit from Pytel's extensive experience with the GKE team, overseeing thousands of Cluster Autoscaler instances. Acquire knowledge on essential metrics for monitoring Cluster Autoscaler in single or multiple clusters, techniques for swift issue identification, log analysis methods, and common Cluster Autoscaler issues to avoid. This talk focuses on universal Cluster Autoscaler challenges applicable across various cloud providers, providing valuable insights for optimizing your Kubernetes infrastructure.

Autoscaling Can Be Reliable: Running Cluster Autoscaler in Prod - Maciej Pytel, Google