Description of problem:

After the upgrade, while watching the unavailability in Jupyterhub, we found that Graphana shows a metric that doesn't match with the prometheus metric and manual metric of availability.

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

We've found it upgrading from v180-14 to v190-7

Steps to Reproduce

1. Upgrade RHODS
2. Comparing the unavailability with Prometheus, manually by counting second and Graphana metric

Actual results:

Graphana shows an unavailability different than the unavailability shown in Prometheus.

Expected results:

Find the same unavailability time in both tools in the upgrade

Reproducibility (Always/Intermittent/Only Once):

Always

Build Details:

Workaround:

Additional info: