While triaging failure of kubemacpooldown test, I found out that the test is failing because the metric kubevirt_hco_system_health_status that should get the value 2 because it is a critical one, get the value 3, which is not like the design of the metric, not mentioned in the documentation and the UI doesn't know to present it because the values should only 0/1/2.

Metrics should report only -  healthy (0), warning (1), or error (2)
The metric is used to asses the Operator health and causes a bug that is very visible in the also in the UI for the Operator heath in the OCP Overview page.

100%

1.Scale down “cluster-network-addons-operator” deployment to zero (Otherwise, it will revert the changes on kubemacpool-mac-controller-manager)

oc -n openshift-cnv scale deployment cluster-network-addons-operator --replicas=0 

2.Scale down “kubemacpool-mac-controller-manager” deployment to zero

oc -n openshift-cnv scale deployment kubemacpool-mac-controller-manager --replicas=0

Observe the alert
check severity=critical
check operator_health_impact=critical

3.Check for kubevirt_hyperconverged_operator_health_status metric value 

3

2

This bug affects 4.18