Improve multi-node stability (#1171)

2024-08-20 22:35:05 -07:00
parent cd10654e7e
commit bea2bb9eea
11 changed files with 94 additions and 76 deletions
--- a/python/sglang/srt/managers/controller_multi.py
+++ b/python/sglang/srt/managers/controller_multi.py
@@ -212,6 +212,4 @@ def start_controller_process(
    except Exception:
        logger.error("Exception in ControllerMulti:\n" + get_exception_traceback())
    finally:
-        for w in controller.workers:
-            os.kill(w.proc.pid, 9)
        kill_parent_process()