Putting in place the right metrics in your monitoring system is critical when running a stable production environment. It enables you to identify problems faster and reduce your MTTR. Misconception and misinterpretation can truly lead you to either wasting cloud resources or to service instability. In this talk I will cover the essential metrics that you must include in your monitoring system if you are running a JVM workload on K8S. I will be showcasing our prometheus/grafana implementation ana for each metric explain what it means, how to expose it and what it the best way to visualize it. We will cover CGroup related metrics as well JVM CPU memory GC together with essential metrics for monitoring queues and pools. Going out of this talk will give you tools to dramatically improve the effectiveness of your monitoring system.