fix: Don't propagate partial recovery healthy events to ND and FR (#296)

Author: KaivalyaMDabhadkar

fault-quarantine/pkg/reconciler/reconciler.go

@@ -434,10 +434,19 @@ func (r *Reconciler) handleAlreadyQuarantinedNode(
 		return nil
 	}
 
-	// Event will modify annotations, proceed with quarantine handling and propagate to ND/FR
-	var status model.Status
+	// Event will modify FQ annotations, proceed with quarantine handling
+	stayQuarantined := r.handleQuarantinedNode(ctx, event, ruleSetEvals)
 
-	if r.handleQuarantinedNode(ctx, event, ruleSetEvals) {
+	// Partial recovery: healthy event that doesn't fully unquarantine the node should
+	// not be propagated to ND/FR
+	if event.IsHealthy && stayQuarantined {
+		return nil
+	}
+
+	var status model.Status
+	if stayQuarantined {
+		// Only for an unhealthy event, set status to AlreadyQuarantined and
+		// propagate to ND/FR
 		status = model.AlreadyQuarantined
 	} else {
 		status = model.UnQuarantined

fault-quarantine/pkg/reconciler/reconciler_e2e_test.go

@@ -735,11 +735,11 @@ func TestE2E_EntityLevelTracking(t *testing.T) {
 		model.StatusInProgress,
 	)
 
-	t.Log("Verify status is AlreadyQuarantined (partial recovery, node stays quarantined)")
+	t.Log("Verify status is nil (partial recovery not propagated to ND/FR)")
 	require.Eventually(t, func() bool {
 		status := getStatus(eventID3)
-		return status != nil && *status == model.AlreadyQuarantined
-	}, statusCheckTimeout, statusCheckPollInterval, "Status should be AlreadyQuarantined for partial recovery")
+		return status == nil
+	}, statusCheckTimeout, statusCheckPollInterval, "Status should be nil for partial recovery")
 
 	require.Eventually(t, func() bool {
 		node, _ := e2eTestClient.CoreV1().Nodes().Get(ctx, nodeName, metav1.GetOptions{})