[Misc] Getting and passing ray runtime_env to workers (#22040)

Signed-off-by: Rui Qiao <ruisearch42@gmail.com>

vllm/engine/arg_utils.py

@@ -36,6 +36,7 @@ from vllm.config import (BlockSize, CacheConfig, CacheDType, CompilationConfig,
 from vllm.logger import init_logger
 from vllm.platforms import CpuArchEnum, current_platform
 from vllm.plugins import load_general_plugins
+from vllm.ray.lazy_utils import is_ray_initialized
 from vllm.reasoning import ReasoningParserManager
 from vllm.test_utils import MODEL_WEIGHTS_S3_BUCKET, MODELS_ON_S3
 from vllm.transformers_utils.utils import check_gguf_file
@@ -1099,6 +1100,15 @@ class EngineArgs:
             kv_sharing_fast_prefill=self.kv_sharing_fast_prefill,
         )
 
+        ray_runtime_env = None
+        if is_ray_initialized():
+            # Ray Serve LLM calls `create_engine_config` in the context
+            # of a Ray task, therefore we check is_ray_initialized()
+            # as opposed to is_in_ray_actor().
+            import ray
+            ray_runtime_env = ray.get_runtime_context().runtime_env
+            logger.info("Using ray runtime env: %s", ray_runtime_env)
+
         # Get the current placement group if Ray is initialized and
         # we are in a Ray actor. If so, then the placement group will be
         # passed to spawned processes.
@@ -1211,6 +1221,7 @@ class EngineArgs:
             max_parallel_loading_workers=self.max_parallel_loading_workers,
             disable_custom_all_reduce=self.disable_custom_all_reduce,
             ray_workers_use_nsight=self.ray_workers_use_nsight,
+            ray_runtime_env=ray_runtime_env,
             placement_group=placement_group,
             distributed_executor_backend=self.distributed_executor_backend,
             worker_cls=self.worker_cls,