chore: Improve KVBM test handling and update gitignore (#3987)

.gitignore

@@ -38,6 +38,10 @@ CMakeCache.txt
 *pytest_report.md
 *pytest_report.xml
 
+# Test artifacts and output directories
+*.shakespeare.txt
+test_determinism_*/
+
 **/__pycache__
 **/venv
 **/.venv

docs/kvbm/trtllm-setup.md

@@ -55,7 +55,7 @@ export DYN_KVBM_CPU_CACHE_GB=4
 export DYN_KVBM_DISK_CACHE_GB=8
 
 # [Experimental] Option 3: Disk cache only (GPU -> Disk direct offloading, bypassing CPU)
-# NOTE: this option is only experimental and it might give out the best performance.
+# NOTE: this option is only experimental and it might not give out the best performance.
 # NOTE: disk offload filtering is not support when using this option.
 export DYN_KVBM_DISK_CACHE_GB=8
 

docs/kvbm/vllm-setup.md

@@ -69,7 +69,7 @@ cd $DYNAMO_HOME/components/backends/vllm
 > export DYN_KVBM_DISK_CACHE_GB=8
 >
 > # [Experimental] Option 3: Disk cache only (GPU -> Disk direct offloading, bypassing CPU)
-> # NOTE: this option is only experimental and it might give out the best performance.
+> # NOTE: this option is only experimental and it might not give out the best performance.
 > # NOTE: disk offload filtering is not support when using this option.
 > export DYN_KVBM_DISK_CACHE_GB=8
 > ```

tests/kvbm/test_determinism_disagg.py

@@ -409,7 +409,7 @@ def llm_server(request, runtime_services):
     if importlib.util.find_spec("vllm") is not None:
         server_type = ServerType.vllm
     else:
-        raise Exception("vllm module is not available in the current environment.")
+        pytest.skip("vllm module is not available in the current environment.")
 
     server_manager = LLMServerManager(
         port=port,