xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-03-29 08:21:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

updated

Browse Source 
Signed-off-by: Robert Shaw <robshaw@redhat.com>
This commit is contained in:
Robert Shaw 2025-07-20 22:19:37 +00:00
parent 9f7d3217dd
commit 91608889a4
2 changed files with 4 additions and 6 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/v1/engine/core.py
View File

@ -422,9 +422,8 @@ class EngineCoreProc(EngineCore):
addresses.frontend_stats_publish_address)
# Only publish request queue stats to coordinator for "internal"
# LB mode.
self.publish_dp_lb_stats = (
self.has_coordinator
and not vllm_config.parallel_config.data_parallel_external_lb)
self.publish_dp_lb_stats = (self.has_coordinator and True)
# and not vllm_config.parallel_config.data_parallel_external_lb)
self._init_data_parallel(vllm_config)

5
vllm/v1/engine/core_client.py
View File

@ -903,8 +903,7 @@ class DPAsyncMPClient(AsyncMPClient):
assert len(self.engine_ranks_managed) > 1
start_idx = self.engine_ranks_managed[0]
end_idx = self.engine_ranks_managed[-1]
logger.info(f"=============== {start_idx=}, {end_idx=}")
end_idx = self.engine_ranks_managed[-1] + 1
async def run_engine_stats_update_task():
with make_zmq_socket(self.ctx, self.stats_update_address,
@ -973,7 +972,7 @@ class DPAsyncMPClient(AsyncMPClient):
# 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=}")
# logger.info(f"{counts=}")
self.lb_engines = counts[start_idx:end_idx]
resources.stats_update_task = asyncio.create_task(