Feature: set real source type in status

Author: kebe7jun

api/dataset/v1alpha1/dataset_types.go

@@ -174,14 +174,17 @@ type DatasetStatus struct {
 	// readOnly indicates whether the dataset is mounted as read-only.
 	ReadOnly     bool        `json:"readOnly,omitempty"`
 	LastSyncTime metav1.Time `json:"lastSyncTime,omitempty"`
+	// SourceType defaults to spec.source.type, unless it is a REFERENCE,
+	// in which case it will match the type of the original Dataset.
+	SourceType DatasetType `json:"sourceType,omitempty"`
 }
 
 // Dataset is the Schema for the datasets API
 // +genclient
 // +kubebuilder:object:root=true
 // +kubebuilder:subresource:status
 // +kubebuilder:resource:shortName=data
-// +kubebuilder:printcolumn:name="type",type=string,JSONPath=`.spec.source.type`
+// +kubebuilder:printcolumn:name="type",type=string,JSONPath=`.status.sourceType`
 // +kubebuilder:printcolumn:name="uri",type=string,JSONPath=`.spec.source.uri`
 // +kubebuilder:printcolumn:name="phase",type=string,JSONPath=`.status.phase`
 type Dataset struct {

config/crd/bases/dataset.baizeai.io_datasets.yaml

@@ -17,7 +17,7 @@ spec:
   scope: Namespaced
   versions:
   - additionalPrinterColumns:
-    - jsonPath: .spec.source.type
+    - jsonPath: .status.sourceType
       name: type
       type: string
     - jsonPath: .spec.source.uri
@@ -695,6 +695,11 @@ spec:
                 description: readOnly indicates whether the dataset is mounted as
                   read-only.
                 type: boolean
+              sourceType:
+                description: |-
+                  SourceType defaults to spec.source.type, unless it is a REFERENCE,
+                  in which case it will match the type of the original Dataset.
+                type: string
               syncRoundStatuses:
                 description: |-
                   syncRoundStatuses is a list of data sync round statuses.

internal/controller/dataset/dataset_controller.go

@@ -102,6 +102,7 @@ func (r *DatasetReconciler) Reconcile(ctx context.Context, req ctrl.Request) (ct
 			{typ: condTypeConfigMap, rec: r.reconcileConfigMap},
 			{typ: condTypeJob, rec: r.reconcileJob},
 			{typ: condTypeJobStatus, rec: r.reconcileJobStatus},
+			{typ: "", rec: r.reconcileStatusType},
 		}
 	}
 
@@ -110,7 +111,7 @@ func (r *DatasetReconciler) Reconcile(ctx context.Context, req ctrl.Request) (ct
 		err := rr.rec(ctx, ds)
 		ds.Status.Conditions = kubeutils.SetCondition(ds.Status.Conditions, rr.typ, err)
 		if err != nil {
-			log.Errorf("error reconciling dataset for %s/%s: %v", ds.Namespace, ds.Name, err)
+			log.Errorf("error reconciling dataset(%s) for %s/%s: %v", rr.typ, ds.Namespace, ds.Name, err)
 			break
 		}
 	}
@@ -726,6 +727,19 @@ func (r *DatasetReconciler) reconcileJob(ctx context.Context, ds *datasetv1alpha
 	return nil
 }
 
+func (r *DatasetReconciler) reconcileStatusType(ctx context.Context, ds *datasetv1alpha1.Dataset) error {
+	if ds.Spec.Source.Type != datasetv1alpha1.DatasetTypeReference {
+		ds.Status.SourceType = ds.Spec.Source.Type
+		return nil
+	}
+	srcDs, err := r.getSourceDataset(ctx, ds)
+	if err != nil {
+		return err
+	}
+	ds.Status.SourceType = srcDs.Status.SourceType
+	return nil
+}
+
 func (r *DatasetReconciler) reconcileJobStatus(ctx context.Context, ds *datasetv1alpha1.Dataset) error {
 	if !supportPreload(ds) {
 		ds.Status.LastSyncTime = ds.CreationTimestamp