Observing Chaos: Real-Time Monitoring of AI-Driven Kubernetes Destr... Josh Halley & Ricardo Aravena

This talk, presented by Josh Halley from Cisco and Ricardo Aravena from CNCF, explores the evolution of chaos engineering within Kubernetes environments. The session highlights a feedback loop where reinforcement learning agents create increasingly complex chaos scenarios in a live Kubernetes cluster, utilizing integrations with ViZDoom and KubeDoom. Attendees will witness a live demonstration that combines gamification with chaos testing, as AI agents interact with real workloads by disrupting services and stressing infrastructure. Key topics include the use of OpenTelemetry for distributed tracing and Cilium for network observability, alongside insights on building resilient agent architectures and utilizing CNCF tools for enhanced chaos observability.