[model-gateway] support customizing Prometheus duration buckets (#14716)

This commit is contained in:
fzyzcjy
2025-12-10 08:30:03 -08:00
committed by GitHub
parent 83e35a7c29
commit 617e9b3bf8
4 changed files with 37 additions and 4 deletions
@@ -54,6 +54,7 @@ class RouterArgs:
# Prometheus configuration
prometheus_port: Optional[int] = None
prometheus_host: Optional[str] = None
prometheus_duration_buckets: Optional[List[float]] = None
# Request ID headers configuration
request_id_headers: Optional[List[str]] = None
# Request timeout in seconds
@@ -342,6 +343,12 @@ class RouterArgs:
default="0.0.0.0",
help="Host address to bind the Prometheus metrics server. Supports IPv4, IPv6 (e.g., ::, ::1), or 0.0.0.0 for all interfaces",
)
parser.add_argument(
f"--{prefix}prometheus-duration-buckets",
type=float,
nargs="+",
help="Buckets for Prometheus duration metrics",
)
parser.add_argument(
f"--{prefix}request-id-headers",
type=str,