xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2026-05-07 02:19:08 +08:00
Code Issues Packages Projects Releases Wiki Activity

add unlimited HWM

Browse Source 
Signed-off-by: clark <panf2333@gmail.com>
This commit is contained in:
clark 2025-03-11 16:17:36 +08:00
parent c0b1443345
commit 8355358fb3
1 changed files with 8 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

12
vllm/entrypoints/disagg_connector.py
View File

@ -42,12 +42,16 @@ async def log_stats(request_queues: dict[str, asyncio.Queue]):
# create async socket use ZMQ_DEALER # create async socket use ZMQ_DEALER
async def create_socket(url: str, async def create_socket(url: str,
zmqctx: zmq.asyncio.Context) -> zmq.asyncio.Socket: zmqctx: zmq.asyncio.Context) -> zmq.asyncio.Socket:
sock = zmqctx.socket(zmq.DEALER) socket = zmqctx.socket(zmq.DEALER)
identity = f"connector-{uuid.uuid4()}" identity = f"connector-{uuid.uuid4()}"
sock.setsockopt(zmq.IDENTITY, identity.encode()) # unlimited HWM
sock.connect(url) hwm_limit = 0
socket.setsockopt(zmq.IDENTITY, identity.encode())
socket.setsockopt(zmq.SNDHWM, hwm_limit)
socket.setsockopt(zmq.RCVHWM, hwm_limit)
socket.connect(url)
logger.info("%s started at %s", identity, url) logger.info("%s started at %s", identity, url)
return sock return socket
@asynccontextmanager @asynccontextmanager