Fixed bugs and validated current test cases

Author: shaikmoeed

kubeflow/trainer/backends/kubernetes/backend.py

@@ -83,7 +83,7 @@ def list_runtimes(self) -> list[types.Runtime]:
                 namespace_thread.get(constants.DEFAULT_TIMEOUT)
             )
 
-            if not (cluster_runtime_list and namespace_runtime_list):
+            if not (cluster_runtime_list or namespace_runtime_list):
                 return result
 
             for runtime in namespace_runtime_list.items + cluster_runtime_list.items:

kubeflow/trainer/backends/kubernetes/backend_test.py

@@ -339,6 +339,16 @@ def list_namespaced_custom_object_response(*args, **kwargs):
             models.TrainerV1alpha1TrainJobList(items=items),
             models.TrainerV1alpha1TrainJobList,
         )
+    elif args[3] == constants.TRAINING_RUNTIME_PLURAL:
+        # TODO: add test case for namespace scoped runtimes
+        # items = [
+        #     create_training_runtime(name="runtime-1"),
+        #     create_training_runtime(name="runtime-2"),
+        # ]
+        mock_thread.get.return_value = normalize_model(
+            models.TrainerV1alpha1TrainingRuntimeList(items=[]),
+            models.TrainerV1alpha1TrainingRuntimeList,
+        )
 
     return mock_thread
 
@@ -472,6 +482,37 @@ def create_cluster_training_runtime(
     )
 
 
+def create_training_runtime(
+    name: str,
+    namespace: str = "default",
+) -> models.TrainerV1alpha1TrainingRuntime:
+    """Create a mock namespaced TrainingRuntime object (not cluster-scoped)."""
+    return models.TrainerV1alpha1TrainingRuntime(
+        apiVersion=constants.API_VERSION,
+        kind="TrainingRuntime",
+        metadata=models.IoK8sApimachineryPkgApisMetaV1ObjectMeta(
+            name=name,
+            namespace=namespace,
+            labels={constants.RUNTIME_FRAMEWORK_LABEL: name},
+        ),
+        spec=models.TrainerV1alpha1TrainingRuntimeSpec(
+            mlPolicy=models.TrainerV1alpha1MLPolicy(
+                torch=models.TrainerV1alpha1TorchMLPolicySource(
+                    numProcPerNode=models.IoK8sApimachineryPkgUtilIntstrIntOrString(2)
+                ),
+                numNodes=2,
+            ),
+            template=models.TrainerV1alpha1JobSetTemplateSpec(
+                metadata=models.IoK8sApimachineryPkgApisMetaV1ObjectMeta(
+                    name=name,
+                    namespace=namespace,
+                ),
+                spec=models.JobsetV1alpha2JobSetSpec(replicatedJobs=[get_replicated_job()]),
+            ),
+        ),
+    )
+
+
 def get_replicated_job() -> models.JobsetV1alpha2ReplicatedJob:
     return models.JobsetV1alpha2ReplicatedJob(
         name="node",

kubeflow/trainer/constants/constants.py

@@ -33,7 +33,7 @@
 CLUSTER_TRAINING_RUNTIME_PLURAL = "clustertrainingruntimes"
 
 # The Kind name for the TrainingRuntime.
-TRAINING_RUNTIME_KIND = "ClusterTrainingRuntime"
+TRAINING_RUNTIME_KIND = "TrainingRuntime"
 
 # The plural for the ClusterTrainingRuntime.
 TRAINING_RUNTIME_PLURAL = "trainingruntimes"