EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[PD] Optimize transfer queue forward logic for dummy rank (#6922)

Browse Source 
Signed-off-by: Shangming Cai <caishangming@linux.alibaba.com>
This commit is contained in:
shangmingc
2025-06-07 09:26:14 +08:00
committed by GitHub
parent 60fdad7cf3
commit 132dad874d
1 changed files with 6 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
python/sglang/srt/disaggregation/mooncake/conn.py
View File

@@ -562,6 +562,12 @@ class MooncakeKVManager(BaseKVManager):
) )
return return
if bootstrap_room not in self.transfer_infos:
# This means that the current rank is a dummy rank for this request,
# and it has already been marked as success, so there is no need to
# add further chunks into the transfer queue.
return
# NOTE(shangming): sharding according to the dst_infos to make sure # NOTE(shangming): sharding according to the dst_infos to make sure
# requests with the same dst_sessions will be added into the same # requests with the same dst_sessions will be added into the same
# queue, which enables early abort with failed sessions. # queue, which enables early abort with failed sessions.
@@ -578,7 +584,6 @@ class MooncakeKVManager(BaseKVManager):
prefill_aux_index=aux_index, prefill_aux_index=aux_index,
) )
) )
self.update_status(bootstrap_room, KVPoll.WaitingForInput)
def check_status(self, bootstrap_room: int): def check_status(self, bootstrap_room: int):
return self.request_status[bootstrap_room] return self.request_status[bootstrap_room]