Prometheus pods cause memory pressure on OpenShift 4 nodes

Issue

• Different OpenShift 4 clusters showing prometheus-k8s pods using too much memory
• The OpenShift Nodes (Infra Nodes) that run these pods end up in a MemoryPressure state
• Temporary increasing the OpenShift Nodes capacity (RAM) and reduce the Prometheus retention (from 15gg to 7gg to even 3gg) just delay the issue
• Temporary increasing storage capacity just delay the issue as well