58 lines
1.4 KiB
Python
58 lines
1.4 KiB
Python
# SPDX-License-Identifier: Apache-2.0
|
|
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
|
|
|
|
from fastapi import FastAPI
|
|
|
|
import vllm.envs as envs
|
|
from vllm.logger import init_logger
|
|
|
|
logger = init_logger(__name__)
|
|
|
|
|
|
def register_vllm_serve_api_routers(app: FastAPI):
|
|
if envs.VLLM_SERVER_DEV_MODE:
|
|
logger.warning(
|
|
"SECURITY WARNING: Development endpoints are enabled! "
|
|
"This should NOT be used in production!"
|
|
)
|
|
|
|
from vllm.entrypoints.serve.lora.api_router import (
|
|
attach_router as attach_lora_router,
|
|
)
|
|
|
|
attach_lora_router(app)
|
|
|
|
from vllm.entrypoints.serve.profile.api_router import (
|
|
attach_router as attach_profile_router,
|
|
)
|
|
|
|
attach_profile_router(app)
|
|
|
|
from vllm.entrypoints.serve.sleep.api_router import (
|
|
attach_router as attach_sleep_router,
|
|
)
|
|
|
|
attach_sleep_router(app)
|
|
|
|
from vllm.entrypoints.serve.rpc.api_router import (
|
|
attach_router as attach_rpc_router,
|
|
)
|
|
|
|
attach_rpc_router(app)
|
|
|
|
from vllm.entrypoints.serve.cache.api_router import (
|
|
attach_router as attach_cache_router,
|
|
)
|
|
|
|
attach_cache_router(app)
|
|
|
|
from vllm.entrypoints.serve.tokenize.api_router import (
|
|
attach_router as attach_tokenize_router,
|
|
)
|
|
|
|
attach_tokenize_router(app)
|
|
|
|
from .instrumentator import register_instrumentator_api_routers
|
|
|
|
register_instrumentator_api_routers(app)
|