fix precommit errors

Author: threcc

tests/model_serving/model_server/llmd/conftest.py

@@ -375,10 +375,10 @@ def singlenode_precise_prefix_cache(
                 "name": "VLLM_ADDITIONAL_ARGS",
                 "value": (
                     f"--prefix-caching-hash-algo {PREFIX_CACHE_HASH_ALGO} --block-size {PREFIX_CACHE_BLOCK_SIZE} "
-                    "--kv_transfer_config '{\"kv_connector\":\"NixlConnector\",\"kv_role\":\"kv_both\"}' "
-                    "--kv-events-config '{\"enable_kv_cache_events\":true,\"publisher\":\"zmq\","
-                    "\"endpoint\":\"tcp://{{ ChildName .ObjectMeta.Name `-epp-service` }}:5557\","
-                    "\"topic\":\"kv@${POD_IP}@${MODEL_NAME}\"}'"
+                    '--kv_transfer_config \'{"kv_connector":"NixlConnector","kv_role":"kv_both"}\' '
+                    '--kv-events-config \'{"enable_kv_cache_events":true,"publisher":"zmq",'
+                    '"endpoint":"tcp://{{ ChildName .ObjectMeta.Name `-epp-service` }}:5557",'
+                    '"topic":"kv@${POD_IP}@${MODEL_NAME}"}\''
                 ),
             },
             {

tests/model_serving/model_server/llmd/constants.py

@@ -41,4 +41,4 @@
             ],
         }
     ],
-}
+}

tests/model_serving/model_server/llmd/test_singlenode_precise_prefix_cache.py

@@ -1,16 +1,3 @@
-import pytest
-from kubernetes.dynamic import DynamicClient
-from ocp_resources.llm_inference_service import LLMInferenceService
-
-from tests.model_serving.model_server.llmd.utils import (
-    get_llmd_router_scheduler_pod,
-    get_llmd_workload_pods,
-    verify_gateway_status,
-    verify_llm_service_status,
-    verify_singlenode_prefix_cache_routing,
-)
-from simple_logger.logger import get_logger
-
 """
 Test Single-Node Precise Prefix Caching.
 
@@ -24,10 +11,24 @@
 - Send multiple requests with shared prefixes and size greater than PREFIX_CACHE_BLOCK_SIZE
 """
 
+import pytest
+from kubernetes.dynamic import DynamicClient
+from ocp_resources.llm_inference_service import LLMInferenceService
+
+from tests.model_serving.model_server.llmd.utils import (
+    get_llmd_router_scheduler_pod,
+    get_llmd_workload_pods,
+    verify_gateway_status,
+    verify_llm_service_status,
+    verify_singlenode_prefix_cache_routing,
+)
+from simple_logger.logger import get_logger
+
 LOGGER = get_logger(name=__name__)
 
 pytestmark = [pytest.mark.llmd_gpu]
 
+
 @pytest.mark.parametrize(
     "unprivileged_model_namespace",
     [pytest.param({"name": "singlenode-prefix-cache-test"})],
@@ -75,12 +76,14 @@ def test_singlenode_precise_prefix_cache(
         assert verify_llm_service_status(singlenode_precise_prefix_cache), "LLMInferenceService should be ready"
 
         # Verify router-scheduler pod exists and is running
-        router_scheduler_pod = get_llmd_router_scheduler_pod(unprivileged_client, singlenode_precise_prefix_cache)
+        router_scheduler_pod = get_llmd_router_scheduler_pod(
+            client=unprivileged_client, llmisvc=singlenode_precise_prefix_cache
+        )
         assert router_scheduler_pod is not None, "Router-scheduler pod should exist"
         assert router_scheduler_pod.instance.status.phase == "Running", "Router-scheduler pod should be running"
 
         # Verify workload pods
-        workload_pods = get_llmd_workload_pods(unprivileged_client, singlenode_precise_prefix_cache)
+        workload_pods = get_llmd_workload_pods(client=unprivileged_client, llmisvc=singlenode_precise_prefix_cache)
         assert len(workload_pods) == 2, f"Expected 2 workload pods, found {len(workload_pods)}"
 
         # Test prefix cache routing (includes assertions for routing affinity)

tests/model_serving/model_server/llmd/utils.py

@@ -265,17 +265,17 @@ def count_chat_completions_requests_in_pod(pod: Pod) -> int:
         logs = pod.log(container="main", since_seconds=120)
 
         # Match: "POST /v1/chat/completions HTTP/1.1" 200
-        pattern = r'POST /v1/chat/completions HTTP/1.1.*200'
+        pattern = r"POST /v1/chat/completions HTTP/1.1.*200"
         matches = re.findall(pattern, logs)
 
         LOGGER.info(f"Pod {pod.name}: Found {len(matches)} requests matching pattern")
 
         # Debug: Show sample log lines if no matches found
         if len(matches) == 0:
-            log_lines = logs.split('\n')
+            log_lines = logs.split("\n")
             LOGGER.info(f"Pod {pod.name}: Total log lines: {len(log_lines)}")
             # Show lines containing "POST" or "completions"
-            relevant_lines = [line for line in log_lines if 'POST' in line or 'completion' in line.lower()]
+            relevant_lines = [line for line in log_lines if "POST" in line or "completion" in line.lower()]
             if relevant_lines:
                 LOGGER.info(f"Pod {pod.name}: Sample relevant lines (first 5):")
                 for line in relevant_lines[:5]:
@@ -305,11 +305,11 @@ def get_pod_that_handled_request(
     Returns:
         Pod name that handled the request, or None if not found
     """
-    time.sleep(5)
+    time.sleep(seconds=5)
 
     current_counts = {}
     for pod in workload_pods:
-        current_counts[pod.name] = count_chat_completions_requests_in_pod(pod)
+        current_counts[pod.name] = count_chat_completions_requests_in_pod(pod=pod)
 
     for pod in workload_pods:
         baseline = baseline_counts.get(pod.name, 0)
@@ -347,7 +347,7 @@ def verify_singlenode_prefix_cache_routing(
     baseline_counts = {}
 
     for pod in workload_pods:
-        baseline_counts[pod.name] = count_chat_completions_requests_in_pod(pod)
+        baseline_counts[pod.name] = count_chat_completions_requests_in_pod(pod=pod)
 
     # Phase 1: Repeated prompts (full cache hit)
     LOGGER.info("Phase 1: Testing repeated prompts")
@@ -363,7 +363,7 @@ def verify_singlenode_prefix_cache_routing(
         inference_config = {
             "default_query_model": {
                 "query_input": repeated_prompt,
-                "query_output": r'.*',
+                "query_output": r".*",
                 "use_regex": True,
             },
             "chat_completions": TINYLLAMA_INFERENCE_CONFIG["chat_completions"],
@@ -382,7 +382,12 @@ def verify_singlenode_prefix_cache_routing(
             authorized_user=True,
         )
 
-        handling_pod = get_pod_that_handled_request(workload_pods, repeated_prompt, timestamp_before, baseline_counts)
+        handling_pod = get_pod_that_handled_request(
+            workload_pods=workload_pods,
+            query=repeated_prompt,
+            timestamp_before=timestamp_before,
+            baseline_counts=baseline_counts,
+        )
         phase1_pods.append(handling_pod)
         if handling_pod:
             baseline_counts[handling_pod] = baseline_counts.get(handling_pod, 0) + 1
@@ -411,7 +416,7 @@ def verify_singlenode_prefix_cache_routing(
         inference_config = {
             "default_query_model": {
                 "query_input": prompt,
-                "query_output": r'.*',
+                "query_output": r".*",
                 "use_regex": True,
             },
             "chat_completions": TINYLLAMA_INFERENCE_CONFIG["chat_completions"],
@@ -430,7 +435,12 @@ def verify_singlenode_prefix_cache_routing(
             authorized_user=True,
         )
 
-        handling_pod = get_pod_that_handled_request(workload_pods, prompt, timestamp_before, baseline_counts)
+        handling_pod = get_pod_that_handled_request(
+            workload_pods=workload_pods,
+            query=prompt,
+            timestamp_before=timestamp_before,
+            baseline_counts=baseline_counts,
+        )
         phase2_pods.append(handling_pod)
         if handling_pod:
             baseline_counts[handling_pod] = baseline_counts.get(handling_pod, 0) + 1