skip tests conditionally + linting

Signed-off-by: threcc <trecchiu@redhat.com>

Author: threcc

tests/model_serving/model_server/llmd/test_llmd_auth.py

@@ -33,7 +33,10 @@ def test_llmisvc_authorized(self, llmisvc_auth_pair):
 
         for entry in [entry_a, entry_b]:
             status, body = send_chat_completions(
-                llmisvc=entry.service, prompt=prompt, token=entry.token, insecure=False
+                llmisvc=entry.service,
+                prompt=prompt,
+                token=entry.token,
+                insecure=False,
             )
             assert status == 200, f"Authorized request failed with {status}: {body}"
             completion = parse_completion_text(response_body=body)

tests/model_serving/model_server/llmd/test_llmd_connection_cpu.py

@@ -21,7 +21,7 @@
     ],
     indirect=True,
 )
-@pytest.mark.usefixtures("valid_aws_config")
+@pytest.mark.usefixtures("valid_aws_config", "skip_if_disconnected")
 class TestLlmdConnectionCpu:
     """Deploy TinyLlama on CPU via S3 and HuggingFace and verify chat completions."""
 

tests/model_serving/model_server/llmd/test_llmd_connection_gpu.py

@@ -21,7 +21,7 @@
     ],
     indirect=True,
 )
-@pytest.mark.usefixtures("valid_aws_config", "skip_if_no_gpu_available")
+@pytest.mark.usefixtures("valid_aws_config", "skip_if_no_gpu_available", "skip_if_disconnected")
 class TestLlmdConnectionGpu:
     """Deploy Qwen on GPU via S3 and HuggingFace and verify chat completions."""
 

tests/model_serving/model_server/llmd/test_llmd_no_scheduler.py

@@ -26,7 +26,7 @@ def router_config(cls):
     [({"name": NAMESPACE}, S3GpuNoSchedulerConfig)],
     indirect=True,
 )
-@pytest.mark.usefixtures("valid_aws_config", "skip_if_no_gpu_available")
+@pytest.mark.usefixtures("valid_aws_config", "skip_if_no_gpu_available", "skip_if_disconnected")
 class TestLlmdNoScheduler:
     """Deploy Qwen on GPU with the scheduler disabled and verify chat completions."""
 

tests/model_serving/model_server/llmd/test_llmd_prefill_decode.py

@@ -18,7 +18,7 @@
     [({"name": NAMESPACE}, PrefillDecodeConfig)],
     indirect=True,
 )
-@pytest.mark.usefixtures("valid_aws_config", "skip_if_no_gpu_available")
+@pytest.mark.usefixtures("valid_aws_config", "skip_if_no_gpu_available", "skip_if_disconnected")
 class TestLlmdPrefillDecode:
     """Deploy Qwen on GPU with prefill-decode disaggregation and verify chat completions."""
 

tests/model_serving/model_server/llmd/test_llmd_singlenode_estimated_prefix_cache.py

@@ -30,7 +30,7 @@
     [({"name": NAMESPACE}, EstimatedPrefixCacheConfig)],
     indirect=True,
 )
-@pytest.mark.usefixtures("valid_aws_config", "skip_if_less_than_2_gpus")
+@pytest.mark.usefixtures("valid_aws_config", "skip_if_less_than_2_gpus", "skip_if_disconnected")
 class TestSingleNodeEstimatedPrefixCache:
     """Deploy Qwen on GPU with 2 replicas and estimated prefix cache routing,
     then verify cache hits via Prometheus metrics.
@@ -58,7 +58,10 @@ def test_singlenode_estimated_prefix_cache(
         assert len(workload_pods) == 2, f"Expected 2 workload pods, found {len(workload_pods)}"
 
         successful = send_prefix_cache_requests(
-            llmisvc=llmisvc, prompt=PREFIX_CACHE_PROMPT, token=llmisvc_token, count=NUM_REQUESTS
+            llmisvc=llmisvc,
+            prompt=PREFIX_CACHE_PROMPT,
+            token=llmisvc_token,
+            count=NUM_REQUESTS,
         )
         assert successful == NUM_REQUESTS, f"Expected all {NUM_REQUESTS} requests to succeed, got {successful}"
 

tests/model_serving/model_server/llmd/test_llmd_singlenode_precise_prefix_cache.py

@@ -31,7 +31,7 @@
     [({"name": NAMESPACE}, PrecisePrefixCacheConfig)],
     indirect=True,
 )
-@pytest.mark.usefixtures("valid_aws_config", "skip_if_less_than_2_gpus")
+@pytest.mark.usefixtures("valid_aws_config", "skip_if_less_than_2_gpus", "skip_if_disconnected")
 class TestSingleNodePrecisePrefixCache:
     """Deploy Qwen on GPU with 2 replicas and precise prefix cache routing,
     then verify cache hits via Prometheus metrics.
@@ -60,7 +60,10 @@ def test_singlenode_precise_prefix_cache(
         assert len(workload_pods) == 2, f"Expected 2 workload pods, found {len(workload_pods)}"
 
         successful = send_prefix_cache_requests(
-            llmisvc=llmisvc, prompt=PREFIX_CACHE_PROMPT, token=llmisvc_token, count=NUM_REQUESTS
+            llmisvc=llmisvc,
+            prompt=PREFIX_CACHE_PROMPT,
+            token=llmisvc_token,
+            count=NUM_REQUESTS,
         )
         assert successful == NUM_REQUESTS, f"Expected all {NUM_REQUESTS} requests to succeed, got {successful}"