Move mem_fraction_static adjustment for multimodal models to server_args.py & Fix session control & Other cleanups (#7748)

This commit is contained in:
Lianmin Zheng
2025-07-04 16:33:33 -07:00
committed by GitHub
parent 975a5ec69c
commit 14229ccf8f
16 changed files with 339 additions and 137 deletions

View File

@@ -1100,7 +1100,7 @@ class Scheduler(
recv_req.session_params is not None
and recv_req.session_params.id is not None
):
req.finished_reason = FINISH_ABORT(
req.set_finish_with_abort(
f"Invalid request: session id {recv_req.session_params.id} does not exist"
)
self._add_request_to_queue(req)