OpenRAG is a comprehensive, single package Retrieval-Augmented Generation platform built on Langflow, Docling, and Opensearch.
Python
feat: crd report scaleddown phase status (#1846)
ming793d5a50e728
diff --git a/kubernetes/operator/internal/controller/openrag_controller.go b/kubernetes/operator/internal/controller/openrag_controller.goindex 6b3ff2371..35e19f3ea 100644--- a/kubernetes/operator/internal/controller/openrag_controller.go+++ b/kubernetes/operator/internal/controller/openrag_controller.go@@ -43,6 +43,7 @@ const (// Phase valuesphaseReconciled = "Reconciled"+ phaseScaledDown = "ScaledDown"phaseRunning = "Running"phaseError = "Error")diff --git a/kubernetes/operator/internal/controller/status.go b/kubernetes/operator/internal/controller/status.goindex b07ad32e5..d3999df58 100644--- a/kubernetes/operator/internal/controller/status.go+++ b/kubernetes/operator/internal/controller/status.go@@ -20,6 +20,10 @@ import (func (r *OpenRAGReconciler) updateStatusSuccess(ctx context.Context, instance *openragv1alpha1.OpenRAG, targetNS string) (ctrl.Result, error) {logger := log.FromContext(ctx)+ scaledDown := ((instance.Spec.Backend.Replicas != nil && *instance.Spec.Backend.Replicas == 0) &&+ (instance.Spec.Frontend.Replicas != nil && *instance.Spec.Frontend.Replicas == 0) &&+ (instance.Spec.Langflow.Replicas != nil && *instance.Spec.Langflow.Replicas == 0))+// Update backend condition firstbackendConditionStatus, err := r.updateBackendCondition(ctx, instance, targetNS)if err != nil {@@ -27,10 +31,19 @@ func (r *OpenRAGReconciler) updateStatusSuccess(ctx context.Context, instance *o// Continue anyway, don't fail reconciliation}- // Determine phase based on backend readiness+ // Determine phase based on scale state and backend readinessphase := phaseReconciledmessage := "All resources reconciled successfully, waiting for backend pod to be ready"- if backendConditionStatus == metav1.ConditionTrue {+ if scaledDown {+ if backendConditionStatus == metav1.ConditionTrue {+ // Desired state is scaled down and backend is healthy (scaled to 0), treat as running+ phase = phaseScaledDown+ message = "All components scaled to 0 replicas"+ } else {+ // Desired state is scaled down but backend condition is not true, treat as scaled down but not ready+ message = "Scaling down all components to 0 replicas (in progress)"+ }+ } else if backendConditionStatus == metav1.ConditionTrue {phase = phaseRunningmessage = "All resources reconciled successfully and backend is running"}@@ -93,6 +106,30 @@ func (r *OpenRAGReconciler) updateBackendCondition(ctx context.Context, instancereturn metav1.ConditionUnknown, fmt.Errorf("failed to get backend deployment: %w", err)}+ // Backend intentionally scaled to 0 — desired state, treat as healthy+ if instance.Spec.Backend.Replicas != nil && *instance.Spec.Backend.Replicas == 0 {+ if deployment.Status.Replicas == 0 {+ meta.SetStatusCondition(&instance.Status.Conditions, metav1.Condition{+ Type: conditionBackendReady,+ Status: metav1.ConditionTrue,+ Reason: "ScaledDown",+ Message: "Backend is scaled to 0 replicas",+ ObservedGeneration: instance.Generation,+ })+ return metav1.ConditionTrue, nil+ } else {+ meta.SetStatusCondition(&instance.Status.Conditions, metav1.Condition{+ Type: conditionBackendReady,+ Status: metav1.ConditionFalse,+ Reason: "ScaledDownNotReady",+ Message: "Backend is scaled to 0 replicas but deployment has non-zero replicas",+ ObservedGeneration: instance.Generation,+ })+ logger.Info("Backend scaled to 0 but deployment has non-zero replicas", "desiredReplicas", *instance.Spec.Backend.Replicas, "deploymentReplicas", deployment.Status.Replicas)+ return metav1.ConditionFalse, nil+ }+ }+reportedStatus := metav1.ConditionFalse// Check if backend pod is ready (single replica)if deployment.Status.ReadyReplicas > 0 && deployment.Status.ReadyReplicas == deployment.Status.Replicas {