Why Your Kubernetes Pod Has Too Many Containers

Episode notes

In this episode of DevOps Daily, Lucas and Luna dig into a quietly catastrophic antipattern: stuffing multiple containers into a single Kubernetes pod. They break down a real-world case where a monitoring stack with five containers per pod caused a production meltdown—crashing the main application container because a sidecar container leaked memory. They explain why the pod is not a lightweight VM, how resource isolation really works in Kubernetes, and why the industry is moving toward sidecar-less architectures. Lucas cites data from a 2025 CNCF survey showing that 42% of teams with multi-container pods reported at least one priority-1 incident caused by container resource fights. They also offer practical heuristics: one container per pod for stateless microservices, strict sidecar patterns for logging and proxies, and what to do when you genuinely need an init container. Plus, a small note on why this conversation made them think about how listener support keeps the show going.