xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-09-06 08:57:00 +08:00
Code Issues Packages Projects Releases Wiki Activity

stash

Browse Source 
Signed-off-by: Robert Shaw <robshaw@redhat.com>
This commit is contained in:
Robert Shaw 2025-07-20 21:23:02 +00:00
parent 85cd2da6b4
commit 9f7d3217dd
1 changed files with 4 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
vllm/v1/engine/core_client.py
View File

@ -904,6 +904,7 @@ class DPAsyncMPClient(AsyncMPClient):
assert len(self.engine_ranks_managed) > 1 assert len(self.engine_ranks_managed) > 1
start_idx = self.engine_ranks_managed[0] start_idx = self.engine_ranks_managed[0]
end_idx = self.engine_ranks_managed[-1] end_idx = self.engine_ranks_managed[-1]
logger.info(f"=============== {start_idx=}, {end_idx=}")
async def run_engine_stats_update_task(): async def run_engine_stats_update_task():
with make_zmq_socket(self.ctx, self.stats_update_address, with make_zmq_socket(self.ctx, self.stats_update_address,
@ -970,6 +971,9 @@ class DPAsyncMPClient(AsyncMPClient):
self.engines_running = running self.engines_running = running
# NOTE: counts include all global Cores. Slice # NOTE: counts include all global Cores. Slice
# to get get the Core's managed by this client. # to get get the Core's managed by this client.
# TODO(rob): update this so we only pull in
# the counts that we are load balancing across.
logger.info(f"{counts=}")
self.lb_engines = counts[start_idx:end_idx] self.lb_engines = counts[start_idx:end_idx]
resources.stats_update_task = asyncio.create_task( resources.stats_update_task = asyncio.create_task(