A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[CI/Build] split true unit tests to Entrypoints Unit Tests (#24418)
Signed-off-by: Ye (Charlotte) Qi <yeq@meta.com>
Ye (Charlotte) Qi committed
Sep 10, 2025 at 06:16 UTC
492196ed0e17c7129ccc3ac4c82eda80bccda82e
2 files changed
+17
-6
.buildkite/test-pipeline.yaml
+16
-5
@@ -102,7 +102,18 @@ steps:
102
commands:
103
- pytest -v -s core
104
105
-- label: Entrypoints Test (LLM) # 30min
105
+- label: Entrypoints Unit Tests # 5min
106
+ timeout_in_minutes: 10
107
+ working_dir: "/vllm-workspace/tests"
108
+ fast_check: true
109
+ source_file_dependencies:
110
+ - vllm/entrypoints
111
+ - tests/entrypoints/
112
+ commands:
113
+ - pytest -v -s entrypoints/openai/tool_parsers
114
+ - pytest -v -s entrypoints/ --ignore=entrypoints/llm --ignore=entrypoints/openai --ignore=entrypoints/offline_mode --ignore=entrypoints/test_chat_utils.py
115
+
116
+- label: Entrypoints Integration Test (LLM) # 30min
117
timeout_in_minutes: 40
118
mirror_hardwares: [amdexperimental]
119
working_dir: "/vllm-workspace/tests"
@@ -119,7 +130,7 @@ steps:
130
- pytest -v -s entrypoints/llm/test_generate.py # it needs a clean process
131
- VLLM_USE_V1=0 pytest -v -s entrypoints/offline_mode # Needs to avoid interference with other tests
132
122
-- label: Entrypoints Test (API Server) # 100min
133
+- label: Entrypoints Integration Test (API Server) # 100min
134
timeout_in_minutes: 130
135
mirror_hardwares: [amdexperimental]
136
working_dir: "/vllm-workspace/tests"
@@ -132,7 +143,7 @@ steps:
143
commands:
144
- export VLLM_WORKER_MULTIPROC_METHOD=spawn
145
- PYTHONPATH=/vllm-workspace pytest -v -s entrypoints/openai/test_collective_rpc.py # PYTHONPATH is needed to import custom Worker extension
135
- - pytest -v -s entrypoints/openai --ignore=entrypoints/openai/test_chat_with_tool_reasoning.py --ignore=entrypoints/openai/test_oot_registration.py --ignore=entrypoints/openai/test_tensorizer_entrypoint.py --ignore=entrypoints/openai/correctness/ --ignore=entrypoints/openai/test_collective_rpc.py
146
+ - pytest -v -s entrypoints/openai --ignore=entrypoints/openai/test_chat_with_tool_reasoning.py --ignore=entrypoints/openai/test_oot_registration.py --ignore=entrypoints/openai/test_tensorizer_entrypoint.py --ignore=entrypoints/openai/correctness/ --ignore=entrypoints/openai/test_collective_rpc.py --ignore=entrypoints/openai/tool_parsers/
147
- pytest -v -s entrypoints/test_chat_utils.py
148
149
- label: Distributed Tests (4 GPUs) # 35min
@@ -823,7 +834,7 @@ steps:
834
# begin io_processor plugins test, all the code in between uses the prithvi_io_processor plugin
835
- pip install -e ./plugins/prithvi_io_processor_plugin
836
- pytest -v -s plugins_tests/test_io_processor_plugins.py
826
- - pip uninstall prithvi_io_processor_plugin -y
837
+ - pip uninstall prithvi_io_processor_plugin -y
838
# end io_processor plugins test
839
# other tests continue here:
840
- pytest -v -s plugins_tests/test_scheduler_plugins.py
@@ -871,7 +882,7 @@ steps:
882
timeout_in_minutes: 45
883
mirror_hardwares: [amdexperimental]
884
working_dir: "/vllm-workspace/tests"
874
- num_gpus: 2
885
+ num_gpus: 2
886
optional: true
887
source_file_dependencies:
888
- vllm/
tests/entrypoints/test_api_server_process_manager.py
+1
-1
@@ -95,7 +95,7 @@ def test_api_server_process_manager_init(api_server_args, with_stats_update):
95
assert not proc.is_alive()
96
97
98
-@patch("vllm.entrypoints.cli.serve.run_api_server_worker",
98
+@patch("vllm.entrypoints.cli.serve.run_api_server_worker_proc",
99
mock_run_api_server_worker)
100
def test_wait_for_completion_or_failure(api_server_args):
101
"""Test that wait_for_completion_or_failure works with failures."""