bug: EvaluationError failures from NodeReconciler not tracked in metrics

[Shreya2005-2005]

Describe the bug

In internal/controller/node_controller.go, when evaluateRuleForNode
fails inside processNodeAgainstAllRules, recordNodeFailure is called
but metrics.Failures counter is NOT incremented.

Line 155 (node_controller.go):

r.recordNodeFailure(rule, node.Name, "EvaluationError", err.Error())
// metrics.Failures.WithLabelValues(rule.Name, "EvaluationError").Inc() -- MISSING

Compare with processAllNodesForRule in nodereadinessrule_controller.go
line 275-276 which correctly does both:

r.recordNodeFailure(rule, node.Name, "EvaluationError", err.Error())
metrics.Failures.WithLabelValues(rule.Name, "EvaluationError").Inc()

Impact

Failures triggered via the NodeReconciler path (node condition changes)
are silently missing from node_readiness_failures_total Prometheus
metric, causing incomplete observability for operators monitoring
cluster health.

Expected behavior

metrics.Failures.WithLabelValues(rule.Name, "EvaluationError").Inc()
should be called alongside recordNodeFailure in node_controller.go.

File

internal/controller/node_controller.go line 155

Are you able to fix this issue?

Yes (I will propose a PR)

[sahitya-chandra]

I think their is already an open pr fixing this

[ajaysundark]

Looks like a duplicate to #219

/close

[k8s-ci-robot]

@ajaysundark: Closing this issue.

Details

In response to this:

Looks like a duplicate to #219

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository.