A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 14, 05:30 UTC
[P/D] MultiConnector supports shutdown (#24425)
Signed-off-by: chaunceyjiang <chaunceyjiang@gmail.com>
Chauncey committed
Sep 10, 2025 at 12:24 UTC
309d7aa401703a051301dbad471a6543021a760f
1 file changed
+12
vllm/distributed/kv_transfer/kv_connector/v1/multi_connector.py
+12
@@ -88,6 +88,18 @@ class MultiConnector(KVConnectorBase_V1):
88
for c in self._connectors:
89
c.clear_connector_metadata()
90
91
+ def shutdown(self):
92
+ exception: Optional[Exception] = None
93
+ for c in self._connectors:
94
+ try:
95
+ c.shutdown()
96
+ except Exception as e:
97
+ logger.exception("Exception during connector %s shutdown.",
98
+ c.__class__.__name__)
99
+ exception = e
100
+ if exception:
101
+ raise exception
102
+
103
# ==============================
104
# Worker-side methods
105
# ==============================