frozenleaves/vllm
1
0
Fork 0
mirror of https://github.com/vllm-project/vllm.git synced 2025-10-20 14:53:52 +08:00
Code Issues Packages Projects Releases Wiki Activity

[V1] Collect env var for usage stats (#12115)

Browse Source
This commit is contained in:
Simon Mo
2025-01-18 19:07:15 -08:00
committed by GitHub
parent 32eb0da808
commit 7a8a48d51e
1 changed files with 18 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
vllm/usage/usage_lib.py
View File

@ -27,6 +27,17 @@ _USAGE_STATS_SERVER = envs.VLLM_USAGE_STATS_SERVER
_GLOBAL_RUNTIME_DATA: Dict[str, Union[str, int, bool]] = {}
_USAGE_ENV_VARS_TO_COLLECT = [
"VLLM_USE_MODELSCOPE",
"VLLM_USE_TRITON_FLASH_ATTN",
"VLLM_ATTENTION_BACKEND",
"VLLM_USE_FLASHINFER_SAMPLER",
"VLLM_PP_LAYER_PARTITION",
"VLLM_USE_TRITON_AWQ",
"VLLM_USE_V1",
"VLLM_ENABLE_V1_MULTIPROCESSING",
]
def set_runtime_usage_data(key: str, value: Union[str, int, bool]) -> None:
"""Set global usage data that will be sent with every usage heartbeat."""
@ -122,6 +133,7 @@ class UsageMessage:
self.gpu_count: Optional[int] = None
self.gpu_type: Optional[str] = None
self.gpu_memory_per_device: Optional[int] = None
self.env_var_json: Optional[str] = None
# vLLM Information
self.model_architecture: Optional[str] = None
@ -176,6 +188,12 @@ class UsageMessage:
self.vllm_version = VLLM_VERSION
self.model_architecture = model_architecture
# Environment variables
self.env_var_json = json.dumps({
env_var: getattr(envs, env_var)
for env_var in _USAGE_ENV_VARS_TO_COLLECT
})
# Metadata
self.log_time = _get_current_timestamp_ns()
self.source = envs.VLLM_USAGE_SOURCE