Peeking Under the Hood - Observability of Kubernetes gRPC Internals

Explore advanced observability techniques for Kubernetes internal gRPC communications in this conference talk that reveals how to instrument and trace kubelet-to-container runtime interactions without modifying Kubernetes source code. Learn to build fine-grained visibility into the typically opaque communication layer between kubelet and containerd using a combination of eBPF, OpenTelemetry, and gRPC interceptors. Discover practical methods for tracing failures in image pulls, analyzing container start latency, and monitoring volume mount operations at the gRPC protocol level where kubelet and Container Runtime Interface (CRI) communicate. Gain insights into debugging mysterious pod startup delays and node-level runtime issues by implementing distributed tracing within Kubernetes' own internal systems. Walk away with a actionable blueprint for closing critical observability gaps in Kubernetes infrastructure and understanding the hidden gRPC magic that powers container orchestration behind the scenes.