xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-03-20 01:21:19 +08:00
Code Issues Packages Projects Releases Wiki Activity

fix bad merge

Browse Source 
Signed-off-by: Nick Hill <nhill@redhat.com>
This commit is contained in:
Nick Hill 2025-07-23 12:00:41 +01:00
parent 1bd5f2f1ad
commit 8601a22d22
1 changed files with 7 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
vllm/v1/engine/core_client.py
View File

@ -900,6 +900,12 @@ class DPAsyncMPClient(AsyncMPClient):
return
assert self.stats_update_address is not None
assert len(self.engine_ranks_managed) > 0
# NOTE: running and waiting counts are all global from
# the Coordinator include all global EngineCores. This
# slice includes just the cores managed by this client.
count_slice = slice(self.engine_ranks_managed[0],
self.engine_ranks_managed[-1] + 1)
async def run_engine_stats_update_task():
with make_zmq_socket(self.ctx, self.stats_update_address,
@ -964,7 +970,7 @@ class DPAsyncMPClient(AsyncMPClient):
counts, wave, running = msgspec.msgpack.decode(buf)
self.current_wave = wave
self.engines_running = running
self.lb_engines = counts
self.lb_engines = counts[count_slice]
resources.stats_update_task = asyncio.create_task(
run_engine_stats_update_task())