fix: update timeout

Author: rnetser

tests/model_serving/model_server/inference_service_configuration/conftest.py

@@ -3,13 +3,15 @@
 import pytest
 from kubernetes.dynamic import DynamicClient
 from ocp_resources.inference_service import InferenceService
+from ocp_resources.pod import Pod
 
 from tests.model_serving.model_server.inference_service_configuration.constants import (
     ISVC_ENV_VARS,
 )
 from tests.model_serving.model_server.inference_service_configuration.utils import (
     update_inference_service,
 )
+from utilities.infra import get_pods_by_isvc_label
 
 
 @pytest.fixture(scope="class")
@@ -29,3 +31,31 @@ def removed_isvc_env_vars(
         isvc_updated_dict={"spec": {"predictor": {"model": {"env": isvc_predictor_spec_model_env}}}},
     ):
         yield ovms_kserve_inference_service
+
+
+@pytest.fixture
+def isvc_pods(
+    admin_client: DynamicClient, ovms_kserve_inference_service: InferenceService
+) -> Generator[list[Pod], Any, Any]:
+    yield get_pods_by_isvc_label(client=admin_client, isvc=ovms_kserve_inference_service)
+
+
+@pytest.fixture(scope="class")
+def patched_isvc_replicas(
+    request: pytest.FixtureRequest,
+    admin_client: DynamicClient,
+    ovms_kserve_inference_service: InferenceService,
+) -> Generator[InferenceService, Any, Any]:
+    with update_inference_service(
+        client=admin_client,
+        isvc=ovms_kserve_inference_service,
+        isvc_updated_dict={
+            "spec": {
+                "predictor": {
+                    "maxReplicas": request.param["max-replicas"],
+                    "minReplicas": request.param["min-replicas"],
+                }
+            }
+        },
+    ):
+        yield ovms_kserve_inference_service

tests/model_serving/model_server/inference_service_configuration/test_isvc_env_vars_updates.py

@@ -83,6 +83,7 @@ def test_serverless_with_isvc_env_vars(self, ovms_kserve_inference_service):
         """Test adding environment variables to the inference service"""
         verify_env_vars_in_isvc_pods(isvc=ovms_kserve_inference_service, env_vars=ISVC_ENV_VARS, vars_exist=True)
 
+    @pytest.mark.slow
     def test_serverless_remove_isvc_env_vars(self, removed_isvc_env_vars):
         """Test removing environment variables from the inference service"""
         verify_env_vars_in_isvc_pods(isvc=removed_isvc_env_vars, env_vars=ISVC_ENV_VARS, vars_exist=False)

tests/model_serving/model_server/inference_service_configuration/test_isvc_replicas_update.py

@@ -0,0 +1,76 @@
+import pytest
+
+from tests.model_serving.model_server.inference_service_configuration.constants import (
+    BASE_ISVC_CONFIG,
+    RUNTIME_CONFIG,
+)
+from tests.model_serving.model_server.inference_service_configuration.utils import (
+    wait_for_new_running_inference_pods,
+)
+from tests.model_serving.model_server.utils import verify_inference_response
+from utilities.constants import (
+    KServeDeploymentType,
+    Protocols,
+)
+from utilities.inference_utils import Inference
+from utilities.manifests.onnx import ONNX_INFERENCE_CONFIG
+
+pytestmark = [pytest.mark.sanity, pytest.mark.usefixtures("valid_aws_config")]
+
+
+@pytest.mark.rawdeployment
+@pytest.mark.parametrize(
+    "model_namespace, openvino_kserve_serving_runtime, ovms_kserve_inference_service",
+    [
+        pytest.param(
+            {"name": "raw-isvc-replicas"},
+            RUNTIME_CONFIG,
+            {
+                **BASE_ISVC_CONFIG,
+                "min-replicas": 2,
+                "max-replicas": 4,
+                "deployment-mode": KServeDeploymentType.RAW_DEPLOYMENT,
+            },
+        )
+    ],
+    indirect=True,
+)
+class TestRawISVCReplicasUpdates:
+    @pytest.mark.dependency(name="test_raw_increase_isvc_replicas")
+    def test_raw_increase_isvc_replicas(self, isvc_pods, patched_isvc_replicas):
+        """Test replicas increase"""
+        wait_for_new_running_inference_pods(isvc=patched_isvc_replicas, orig_pods=isvc_pods, expected_num_pods=2)
+
+    @pytest.mark.dependency(depends=["test_raw_increase_isvc_replicas"])
+    def test_raw_increase_isvc_replicas_inference(self, ovms_kserve_inference_service):
+        """Verify inference after replicas increase"""
+        verify_inference_response(
+            inference_service=ovms_kserve_inference_service,
+            inference_config=ONNX_INFERENCE_CONFIG,
+            inference_type=Inference.INFER,
+            protocol=Protocols.HTTP,
+            use_default_query=True,
+        )
+
+    @pytest.mark.parametrize(
+        "patched_isvc_replicas",
+        [
+            pytest.param({"min-replicas": 1, "max-replicas": 1}),
+        ],
+        indirect=True,
+    )
+    @pytest.mark.dependency(name="test_raw_decrease_isvc_replicas")
+    def test_raw_decrease_isvc_replicas(self, isvc_pods, patched_isvc_replicas):
+        """Test replicas decrease"""
+        wait_for_new_running_inference_pods(isvc=patched_isvc_replicas, orig_pods=isvc_pods, expected_num_pods=2)
+
+    @pytest.mark.dependency(depends=["test_raw_decrease_isvc_replicas"])
+    def test_raw_decrease_isvc_replicas_inference(self, ovms_kserve_inference_service):
+        """Verify inference after replicas decrease"""
+        verify_inference_response(
+            inference_service=ovms_kserve_inference_service,
+            inference_config=ONNX_INFERENCE_CONFIG,
+            inference_type=Inference.INFER,
+            protocol=Protocols.HTTP,
+            use_default_query=True,
+        )

tests/model_serving/model_server/inference_service_configuration/utils.py

@@ -1,3 +1,5 @@
+from __future__ import annotations
+
 from contextlib import contextmanager
 from typing import Any, Generator
 
@@ -70,13 +72,17 @@ def verify_env_vars_in_isvc_pods(isvc: InferenceService, env_vars: list[dict[str
         )
 
 
-def wait_for_new_running_inference_pods(isvc: InferenceService, orig_pods: list[Pod]) -> None:
+def wait_for_new_running_inference_pods(
+    isvc: InferenceService, orig_pods: list[Pod], expected_num_pods: int | None = None
+) -> None:
     """
     Wait for the inference pod to be replaced.
 
     Args:
         isvc (InferenceService): InferenceService object.
         orig_pods (list): List of Pod objects.
+        expected_num_pods (int): Number of pods expected to be running. I
+            f not provided, the number of pods is expected to be len(orig_pods)
 
     Raises:
         TimeoutError: If the pods are not replaced.
@@ -85,15 +91,17 @@ def wait_for_new_running_inference_pods(isvc: InferenceService, orig_pods: list[
     LOGGER.info("Waiting for pods to be replaced")
     oring_pods_names = [pod.name for pod in orig_pods]
 
+    expected_num_pods = expected_num_pods or len(orig_pods)
+
     try:
         for pods in TimeoutSampler(
-            wait_timeout=Timeout.TIMEOUT_4MIN,
+            wait_timeout=Timeout.TIMEOUT_10MIN,
             sleep=5,
             func=get_pods_by_isvc_label,
             client=isvc.client,
             isvc=isvc,
         ):
-            if pods and len(pods) == len(orig_pods):
+            if pods and len(pods) == expected_num_pods:
                 if all(pod.name not in oring_pods_names and pod.status == pod.Status.RUNNING for pod in pods):
                     return