reduce scaledown time to 3 minutes to try to reduce costs of unused containers (#2205)

rafel-roboflow · web-flow · commit 5df179592637 · 2026-04-08T08:25:31.000Z
diff --git a/inference/core/env.py b/inference/core/env.py
@@ -859,7 +859,7 @@
 )
 # seconds
 WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW = int(
-    os.getenv("WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW", "15")
+    os.getenv("WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW", "3")
 )
 WEBRTC_MODAL_IMAGE_NAME = os.getenv(
     "WEBRTC_MODAL_IMAGE_NAME", "roboflow/roboflow-inference-server-gpu"

Original file line number	Diff line number	Diff line change
`@@ -859,7 +859,7 @@`
`859`	`859`	`)`
`860`	`860`	`# seconds`
`861`	`861`	`WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW = int(`
`862`		`- os.getenv("WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW", "15")`
	`862`	`+ os.getenv("WEBRTC_MODAL_FUNCTION_SCALEDOWN_WINDOW", "3")`
`863`	`863`	`)`
`864`	`864`	`WEBRTC_MODAL_IMAGE_NAME = os.getenv(`
`865`	`865`	`"WEBRTC_MODAL_IMAGE_NAME", "roboflow/roboflow-inference-server-gpu"`