[Log] Optimize duplicate startup log (#33944)

Signed-off-by: yewentao256 <zhyanwentao@126.com>

vllm/compilation/backends.py

@@ -263,11 +263,12 @@ class CompilerManager:
                 now = time.time()
                 elapsed = now - compilation_start_time
                 compilation_config.compilation_time += elapsed
-                logger.info(
+                logger.info_once(
                     "Directly load the compiled graph(s) for compile range %s "
                     "from the cache, took %.3f s",
                     str(compile_range),
                     elapsed,
+                    scope="local",
                 )
             return compiled_graph
 

vllm/utils/deep_gemm.py

@@ -91,14 +91,16 @@ def is_deep_gemm_e8m0_used() -> bool:
     _lazy_init()
 
     if _fp8_gemm_nt_impl is None:
-        logger.info_once("DeepGEMM E8M0 disabled: _fp8_gemm_nt_impl not found")
+        logger.info_once(
+            "DeepGEMM E8M0 disabled: _fp8_gemm_nt_impl not found", scope="local"
+        )
         return False
 
     if envs.VLLM_USE_DEEP_GEMM_E8M0:
-        logger.info_once("DeepGEMM E8M0 enabled on current platform.")
+        logger.info_once("DeepGEMM E8M0 enabled on current platform.", scope="local")
         return True
 
-    logger.info_once("DeepGEMM E8M0 disabled on current configuration.")
+    logger.info_once("DeepGEMM E8M0 disabled on current configuration.", scope="local")
     return False
 
 

vllm/v1/worker/gpu_worker.py

@@ -119,9 +119,6 @@ class Worker(WorkerBase):
 
         self.use_v2_model_runner = envs.VLLM_USE_V2_MODEL_RUNNER
 
-        if self.use_v2_model_runner:
-            logger.info_once("Using V2 Model Runner", scope="global")
-
     def sleep(self, level: int = 1) -> None:
         from vllm.device_allocator.cumem import CuMemAllocator
 
@@ -240,6 +237,9 @@ class Worker(WorkerBase):
                 current_platform.dist_backend,
             )
 
+            if self.use_v2_model_runner:
+                logger.info_once("Using V2 Model Runner", scope="local")
+
             # Set random seed.
             set_random_seed(self.model_config.seed)