-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
None
-
0.42
-
False
-
-
False
-
None
-
---
-
---
-
-
CNV I/U Operators Sprint 262
-
None
Description of problem:
As part of testing LowReadyVirtControllersCount alert, the expression kubevirt_virt_controller_ready < kubevirt_virt_controller_up need to happen. the metric kubevirt_virt_controller_ready shows wrong value of ready virt-controller pods, the metric shows 2 even when 0 is ready.
Version-Release number of selected component (if applicable):
4.17
How reproducible:
100%
Steps to Reproduce:
1.Disable virt-operator and olm-operator oc scale deployment virt-operator -n openshift-cnv --replicas=0 oc scale deployment olm-operator -n openshift-operator-lifecycle-manager --replicas=0 2.Change under readinesProbe, the httpGet (or port) to something fake like /fake-leader 3.Delete virt-controller ReplicaSet 4.Look at the value of kubevirt_virt_controller_ready and compare it with the output of the oc command to get the pods. oc get pods -n openshift-cnv | grep virt-controller
Actual results:
The oc command shows that the virt-controller is not ready but up while the metric shows that the pods is ready.
Expected results:
The metric should correlate to the real status of the virt-controller pods which is not ready in this situation meaning the value should be 0 in this case.
Additional info:
Added screenshots
- is related to
-
CNV-43141 [spike] Update virt components unready alerts changes
- New