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 2, 14:45 UTC
[BugFix] Fix hard-coded timeout for multi-API-server startup (#43768)
Signed-off-by: Vadim Gimpelson <vadim.gimpelson@gmail.com> Co-authored-by: Nick Hill <nickhill123@gmail.com>
Vadim Gimpelson committed
May 28, 2026 at 11:09 UTC
d6b48f928f922849db55751ceadbbb55a5a85260
1 file changed
+1
-1
vllm/v1/utils.py
+1
-1
@@ -241,7 +241,7 @@ class APIServerProcessManager:
241
242
def gather_actual_addresses(
243
self,
244
- timeout: float = 60.0,
244
+ timeout: float = envs.VLLM_ENGINE_READY_TIMEOUT_S,
245
) -> tuple[list[str], list[str]]:
246
"""Return (inputs, outputs) reported by each child, indexed by
247
``client_index``. Raises ``RuntimeError`` on timeout or premature