Description of problem:

Spawning a medium size jupyter container, in a cluster without enough resources, is causing the notebook pod to remain in Pending state for 10 minutes, but the metrics cpu_requests_runtime and running cpu_limits_runtime are counting it as if it was running.

Prerequisites (if any, like setup, operators/versions):

rhods version v1.10.0-6

Steps to Reproduce

1. install RHODS in a cluster without enough resources and to have a pod in a pending state
2. Go to openshift-monitoring project
3. Write the metrics in prometheus cpu_requests_runtime and running cpu_limits_runtime

Actual results:

A pod in a pending state is counting into the metric.

Expected results:

Pods have to be counted when they are in a running state.

Reproducibility (Always/Intermittent/Only Once):

Always

Build Details:

Workaround:

Additional info: