I installed Prometheus operator Helm chart on my k8s cluster. I had a look at Grafana and could see the node CPU and memory usage. However Prometheus doesn't manage to collect pod CPU/memory usage.
For instance sum(rate(container_cpu_usage_seconds_total{container_name!="POD",pod_name!=""}[5m]))
would return no data
.kubectl top pod
works fine so I cannot understand why Prometheus won't manage to get this data. Or is the rule above wrong perhaps?
I'm using: