Kubernetes OOMKilled -- understanding memory limits

Pod is killed with OOMKilled status. Container restarts repeatedly. Application works locally but crashes in Kubernetes.

The container exceeded its memory limit. Debug: (1) kubectl describe pod -- check Last State: OOMKilled. (2) Check actual memory usage: kubectl top pod. (3) Common causes: memory limit too low (increase resources.limits.memory), memory leak in app, JVM not respecting container limits (use -XX:MaxRAMPercentage=75), Node.js needs --max-old-space-size. (4) Set requests close to actual usage, limits 20-50% higher. (5) Use metrics-server or Prometheus to track actual consumption over time.

Kubernetes enforces memory limits via cgroups. When a container exceeds its limit, the kernel OOM killer terminates it immediately (no graceful shutdown).