From c35dcfdb30cc3839acb503f6dd739fce86a66e15 Mon Sep 17 00:00:00 2001 From: Yusong Gao Date: Fri, 11 Apr 2025 23:03:07 +0800 Subject: [PATCH] [PD] fix: skip warmup request in disaggregation mode to prevent crash on timeout (#5292) --- python/sglang/srt/entrypoints/http_server.py | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) diff --git a/python/sglang/srt/entrypoints/http_server.py b/python/sglang/srt/entrypoints/http_server.py index 1f93b475c..3f4feaf15 100644 --- a/python/sglang/srt/entrypoints/http_server.py +++ b/python/sglang/srt/entrypoints/http_server.py @@ -785,13 +785,17 @@ def _wait_and_warmup( json_data["sampling_params"]["max_new_tokens"] = 0 try: - res = requests.post( - url + request_name, - json=json_data, - headers=headers, - timeout=600, - ) - assert res.status_code == 200, f"{res}" + if server_args.disaggregation_mode == "null": + res = requests.post( + url + request_name, + json=json_data, + headers=headers, + timeout=600, + ) + assert res.status_code == 200, f"{res}" + else: + # Warmup request currently hangs in disaggregation mode, so we skip it. + logger.info("Skipping warmup request in disaggregation mode") except Exception: last_traceback = get_exception_traceback() if pipe_finish_writer is not None: