fix labels

tests/model_serving/model_server/private_endpoint/conftest.py

@@ -12,7 +12,7 @@
 from tests.model_serving.model_server.private_endpoint.utils import create_sidecar_pod
 from tests.model_serving.model_server.utils import create_isvc
 from utilities.constants import KServeDeploymentType, ModelFormat, ModelStoragePath
-from utilities.infra import create_ns, create_storage_config_secret
+from utilities.infra import create_ns
 
 LOGGER = get_logger(name=__name__)
 
@@ -28,7 +28,6 @@ def endpoint_isvc(
     admin_client: DynamicClient,
     serving_runtime_from_template: ServingRuntime,
     models_endpoint_s3_secret: Secret,
-    storage_config_secret: Secret,
 ) -> Generator[InferenceService, None, None]:
     with create_isvc(
         client=admin_client,
@@ -44,29 +43,6 @@ def endpoint_isvc(
         yield isvc
 
 
-@pytest.fixture(scope="class")
-def storage_config_secret(
-    admin_client: DynamicClient,
-    models_endpoint_s3_secret: Secret,
-    aws_access_key_id: str,
-    aws_secret_access_key: str,
-    models_s3_bucket_name: str,
-    models_s3_bucket_region: str,
-    models_s3_bucket_endpoint: str,
-) -> Generator[Secret, None, None]:
-    with create_storage_config_secret(
-        admin_client=admin_client,
-        endpoint_secret_name=models_endpoint_s3_secret.name,
-        namespace=models_endpoint_s3_secret.namespace,
-        aws_access_key=aws_access_key_id,
-        aws_secret_access_key=aws_secret_access_key,
-        aws_s3_bucket=models_s3_bucket_name,
-        aws_s3_region=models_s3_bucket_region,
-        aws_s3_endpoint=models_s3_bucket_endpoint,
-    ) as storage_config:
-        yield storage_config
-
-
 @pytest.fixture()
 def endpoint_pod_with_istio_sidecar(
     admin_client: DynamicClient, model_namespace: Namespace

tests/model_serving/model_server/utils.py

@@ -9,7 +9,7 @@
 from simple_logger.logger import get_logger
 from timeout_sampler import TimeoutSampler
 
-from utilities.constants import KServeDeploymentType
+from utilities.constants import KServeDeploymentType, KubernetesAnnotations
 from utilities.exceptions import FailedPodsError, InferenceResponseError, InvalidStorageArgumentError
 from utilities.inference_utils import UserInference
 from utilities.infra import (
@@ -110,7 +110,7 @@ def create_isvc(
     if volumes:
         predictor_dict["volumes"] = volumes
 
-    annotations = {"serving.kserve.io/deploymentMode": deployment_mode}
+    annotations = {KubernetesAnnotations.DEPLOYMENT_MODE: deployment_mode}
 
     if deployment_mode == KServeDeploymentType.SERVERLESS:
         annotations.update({

utilities/constants.py

@@ -112,6 +112,7 @@ class KubernetesAnnotations:
     INSTANCE: str = f"{Resource.ApiGroup.APP_KUBERNETES_IO}/instance"
     PART_OF: str = f"{Resource.ApiGroup.APP_KUBERNETES_IO}/part-of"
     CREATED_BY: str = f"{Resource.ApiGroup.APP_KUBERNETES_IO}/created-by"
+    DEPLOYMENT_MODE: str = "serving.kserve.io/deploymentMode"
 
 
 class StorageClassName:

utilities/general.py

@@ -1,11 +1,15 @@
+from __future__ import annotations
+
 import base64
 from typing import Dict
 
 from kubernetes.dynamic import DynamicClient
+from ocp_resources.inference_service import InferenceService
 from ocp_resources.pod import Pod
 from simple_logger.logger import get_logger
 
 import utilities.infra
+from utilities.constants import KServeDeploymentType, KubernetesAnnotations, MODELMESH_SERVING
 
 LOGGER = get_logger(name=__name__)
 
@@ -104,3 +108,15 @@ def download_model_data(
         pod.wait_for_status(status=Pod.Status.SUCCEEDED, timeout=25 * 60)
 
     return model_path
+
+
+def create_isvc_label_selector_str(isvc: InferenceService) -> str:
+    if isvc.instance.metadata.annotations.get(KubernetesAnnotations.DEPLOYMENT_MODE) in (
+        KServeDeploymentType.SERVERLESS,
+        KServeDeploymentType.RAW_DEPLOYMENT,
+    ):
+        label_selector = f"{isvc.ApiGroup.SERVING_KSERVE_IO}/inferenceservice={isvc.name}"
+
+    else:
+        label_selector = f"modelmesh-service={MODELMESH_SERVING}"
+    return label_selector

utilities/inference_utils.py

@@ -17,7 +17,6 @@
 from utilities.certificates_utils import get_ca_bundle
 from utilities.constants import (
     KServeDeploymentType,
-    MODELMESH_SERVING,
     ModelInferenceRuntime,
     Protocols,
     HTTPRequest,
@@ -235,7 +234,7 @@ def run_inference(
 
         # For internal inference, we need to use port forwarding to the service
         if not self.visibility_exposed:
-            svc = self.get_isvc_service()
+            svc = get_services_by_isvc_label(client=self.inference_service.client, isvc=self.inference_service)[0]
             port = self.get_target_port(svc=svc)
             cmd = cmd.replace("localhost", f"localhost:{port}")
 
@@ -281,23 +280,6 @@ def run_inference(
         except JSONDecodeError:
             return {"output": out}
 
-    def get_isvc_service(self) -> Service:
-        if self.deployment_mode == KServeDeploymentType.MODEL_MESH:
-            if svc := list(
-                Service.get(
-                    dyn_client=self.inference_service.client,
-                    name=MODELMESH_SERVING,
-                    namespace=self.inference_service.namespace,
-                )
-            ):
-                svc = svc[0]
-            else:
-                raise ValueError(f"Service {MODELMESH_SERVING} not found")
-
-        else:
-            svc = get_services_by_isvc_label(client=self.inference_service.client, isvc=self.inference_service)[0]
-        return svc
-
     def get_target_port(self, svc: Service) -> int:
         if self.protocol in Protocols.ALL_SUPPORTED_PROTOCOLS:
             svc_protocol = "TCP"

utilities/infra.py

@@ -25,9 +25,8 @@
 from pytest_testconfig import config as py_config
 from simple_logger.logger import get_logger
 
-from utilities.constants import KServeDeploymentType, MODELMESH_SERVING
 import utilities.general
-
+from utilities.general import create_isvc_label_selector_str
 
 LOGGER = get_logger(name=__name__)
 TIMEOUT_2MIN = 2 * 60
@@ -69,15 +68,7 @@ def wait_for_inference_deployment_replicas(
     client: DynamicClient, isvc: InferenceService, deployment_mode: str, expected_num_deployments: int = 1
 ) -> List[Deployment]:
     ns = isvc.namespace
-
-    if deployment_mode in (
-        KServeDeploymentType.SERVERLESS,
-        KServeDeploymentType.RAW_DEPLOYMENT,
-    ):
-        label_selector = f"{isvc.ApiGroup.SERVING_KSERVE_IO}/inferenceservice={isvc.name}"
-
-    else:
-        label_selector = f"modelmesh-service={MODELMESH_SERVING}"
+    label_selector = create_isvc_label_selector_str(isvc=isvc)
 
     deployments = list(
         Deployment.get(
@@ -131,37 +122,6 @@ def s3_endpoint_secret(
         yield secret
 
 
-@contextmanager
-def create_storage_config_secret(
-    admin_client: DynamicClient,
-    endpoint_secret_name: str,
-    namespace: str,
-    aws_access_key: str,
-    aws_secret_access_key: str,
-    aws_s3_bucket: str,
-    aws_s3_region: str,
-    aws_s3_endpoint: str,
-) -> Generator[Secret, None, None]:
-    secret = {
-        "access_key_id": aws_access_key,
-        "bucket": aws_s3_bucket,
-        "default_bucket": aws_s3_bucket,
-        "endpoint_url": aws_s3_endpoint,
-        "region": aws_s3_region,
-        "secret_access_key": aws_secret_access_key,
-        "type": "s3",
-    }
-    data = {endpoint_secret_name: utilities.general.b64_encoded_string(string_to_encode=json.dumps(secret))}
-    with Secret(
-        client=admin_client,
-        namespace=namespace,
-        data_dict=data,
-        wait_for_resource=True,
-        name="storage-config",
-    ) as storage_config:
-        yield storage_config
-
-
 @contextmanager
 def create_isvc_view_role(
     client: DynamicClient,
@@ -261,12 +221,14 @@ def get_services_by_isvc_label(client: DynamicClient, isvc: InferenceService) ->
     Raises:
         ResourceNotFoundError: if no pods are found.
     """
+    label_selector = create_isvc_label_selector_str(isvc=isvc)
+
     if svcs := [
         svc
         for svc in Service.get(
             dyn_client=client,
             namespace=isvc.namespace,
-            label_selector=f"{isvc.ApiGroup.SERVING_KSERVE_IO}/inferenceservice={isvc.name}",
+            label_selector=label_selector,
         )
     ]:
         return svcs
@@ -286,12 +248,13 @@ def get_pods_by_isvc_label(client: DynamicClient, isvc: InferenceService) -> Lis
     Raises:
         ResourceNotFoundError: if no pods are found.
     """
+    label_selector = create_isvc_label_selector_str(isvc=isvc)
     if pods := [
         pod
         for pod in Pod.get(
             dyn_client=client,
             namespace=isvc.namespace,
-            label_selector=f"{isvc.ApiGroup.SERVING_KSERVE_IO}/inferenceservice={isvc.name}",
+            label_selector=label_selector,
         )
     ]:
         return pods