1. Proposed title of this feature request

Multinode: Double scrape_interval for CMO controlled ServiceMonitors 

2. What is the nature and description of the request?

One of our Telco partner needs to have more flexibility about how the monitoring stack works, in order the reduce the CPU/Memory consumption. This RFE will focus on: To change the scrap interval as it was done for SNO.

Doubling the scrape_interval will make the CPU consumption decreases. Specially for Prometheus, Kubelet and node-exporter. Which are the more affected one in our partner.

This has been already done for SNOs:

https://issues.redhat.com/browse/MON-2384

And we would like to implement the same for multinode. This was implemented only for SNO, because, in principle, these are more constrained about resources. But multinode clusters are very much bigger, and therefore are more affected by a Prometheus scraping too often. Bigger servers have also more devices, more pods, and these makes the cluster more affected by Monitoring Stack consumption. Even higher than on SNO, as it has been seen in our partner's cluster.

3. Why does the customer need this? (List the business requirements here)

Telco, and specially RAN, have very special requirements about performance and resources consumption. The RAN Profile already contains different optimizations focused on CPU utilization like the PAO, accelerated booting, disable some systemd services, and other specific optimizations. But monitoring stack consumes some resources that our partners would use for their workloads.

We have made an experiment on one of their clusters. We have re-produced the intervals according to what it was implemented for SNO:

https://issues.redhat.com/browse/MON-2384

Getting very good results:

• kubelet
• node-exporter
• prometheus

They would like to have this same scraping intervals on SNO.

4. List any affected packages or components.

Prometheus, node-exporter, kubelet