How Kubernetes Custom Resource Definitions Cause Controller Memory Leaks
from DevOps Daily with Fexingo: CI/CD, Kubernetes, and Modern Software Operations Hosted by Fexingo
Episode 83 of DevOps Daily with Fexingo dives into a subtle but costly Kubernetes issue: Custom Resource Definitions (CRDs) and the controllers built to manage them can silently leak memory over weeks of operation. Lucas and Luna walk through a real incident at a mid-sized fintech company where in-memory caches of CRD objects grew unbounded, leading to OOM-killed controllers and cascading failures. They explain why the problem stems from the way client-go's informer cache interacts with custom resources under high-churn workloads, and what operators can do to catch it before it hits production. If you manage a Kubernetes cluster with custom resources, this episode will change how you monitor controller memory. #Kubernetes #CRD #CustomResourceDefinitions #MemoryLeak #Controller #ClientGo #InformerCache #GoProgramming #OOMKill #Fintech #ProductionIncident #KubernetesOperator #CloudNative #SiteReliabilityEngineering #DevOps #Technology #FexingoBusiness #BusinessPodcast Keep every episode free: buymeacoffee.com/fexingo
Embed this episode
No transcript yet.