Misc fix for min_p_sampling, --cuda-graph-bs (#2761)

python/sglang/srt/managers/data_parallel_controller.py

@@ -20,6 +20,7 @@ import threading
 from enum import Enum, auto
 
 import psutil
+import setproctitle
 import zmq
 
 from sglang.srt.managers.io_struct import (
@@ -230,6 +231,7 @@ def run_data_parallel_controller_process(
     port_args: PortArgs,
     pipe_writer,
 ):
+    setproctitle.setproctitle("sglang::data_parallel_controller")
     configure_logger(server_args)
     parent_process = psutil.Process().parent()
 

python/sglang/srt/managers/scheduler.py

@@ -1516,8 +1516,9 @@ class Scheduler:
         return success, message
 
     def update_weights_from_distributed(
-        self, recv_req: UpdateWeightsFromDistributedReqInput
-    ):
+        self,
+        recv_req: UpdateWeightsFromDistributedReqInput,
+    ) -> Tuple[bool, str]:
         """Update the online model parameter."""
         success, message = self.tp_worker.update_weights_from_distributed(recv_req)
         if success: