[Bugfix] Fix inner_dp_world initialization order for multi-node TP (#35892)

Signed-off-by: Yongye Zhu <zyy1102000@gmail.com>
Signed-off-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Co-authored-by: Nick Hill <nickhill123@gmail.com>
Co-authored-by: Nick Hill <nhill@redhat.com>

vllm/v1/executor/multiproc_executor.py

@@ -608,7 +608,6 @@ class WorkerProc:
         )
 
         # Load model
-        self._init_message_queues(input_shm_handle, vllm_config)
         is_eep_new_worker = envs.VLLM_ELASTIC_EP_SCALE_UP_LAUNCH
         if not is_eep_new_worker:
             self.worker.init_device()
@@ -618,6 +617,10 @@ class WorkerProc:
             )
             self.worker.load_model()
 
+        # Initialize message queues after init_device() since multi-node setups
+        # (nnodes_within_dp > 1) require distributed groups to be initialized
+        self._init_message_queues(input_shm_handle, vllm_config)
+
         # Enable environment variable cache (e.g. assume no more
         # environment variable overrides after this point)
         enable_envs_cache()