update&add

inference-scaled-config/clip/http-add-on/httpscaledobject.yaml

@@ -7,7 +7,8 @@ spec:
   hosts:
      - annomidware.d-robotics.cc
   pathPrefixes:
-    - /v1/predict
+    - /classification/clip/v1/predict
+    - /classification/clip/v1/embedding/
   
   # 目标 Deployment 或 StatefulSet
   scaleTargetRef:

inference-scaled-config/groudingdino/http-add-on/httpscaledobject.yaml

@@ -0,0 +1,38 @@
+apiVersion: http.keda.sh/v1alpha1
+kind: HTTPScaledObject
+metadata:
+  name: groudingdino-http-scaler
+  namespace: bj1-dcloud
+spec:
+  hosts:
+     - annomidware.d-robotics.cc
+  pathPrefixes:
+    - /detection/groundingdino/v1/predict
+  
+  # 目标 Deployment 或 StatefulSet
+  scaleTargetRef:
+    name: groundingdino
+    kind: Deployment
+    apiVersion: apps/v1
+    service: groundingdino-svc  # 对应 HTTPRoute 中的后端服务
+    port: 80
+
+  # 副本数配置
+  replicas:
+    min: 0  # 支持缩容到 0（空闲时节省资源）
+    max: 1  # 最大副本数（根据实际需求调整）
+
+  # 扩缩容策略
+  scalingMetric:
+    # requestRate:
+    #   granularity: 1s
+    #   targetValue: 2  # 每秒 10 个请求触发扩容
+    #   window: 1m  # 1 分钟时间窗口
+    concurrency:
+      targetValue: 3
+    
+
+  # 扩缩容行为控制
+  scaledownPeriod: 300
+  targetPendingRequests: 8  # 等待处理的请求数阈值
+    

inference-scaled-config/segment/http-add-on/httpscaledobject.yaml

@@ -7,7 +7,7 @@ spec:
   hosts:
      - annomidware.d-robotics.cc
   pathPrefixes:
-    - /v1/predict
+    - /segmentation/sam/v1/predict
   
   # 目标 Deployment 或 StatefulSet
   scaleTargetRef:

inference-scaled-config/selective-igev/http-add-on/httpscaledobject.yaml

@@ -0,0 +1,38 @@
+apiVersion: http.keda.sh/v1alpha1
+kind: HTTPScaledObject
+metadata:
+  name: selective-igev-http-scaler
+  namespace: bj1-dcloud
+spec:
+  hosts:
+     - annomidware.d-robotics.cc
+  pathPrefixes:
+    - /stereo/selective-igev/v1/predict
+  
+  # 目标 Deployment 或 StatefulSet
+  scaleTargetRef:
+    name: selective-igev
+    kind: Deployment
+    apiVersion: apps/v1
+    service: selective-igev-svc  # 对应 HTTPRoute 中的后端服务
+    port: 80
+
+  # 副本数配置
+  replicas:
+    min: 0  # 支持缩容到 0（空闲时节省资源）
+    max: 1  # 最大副本数（根据实际需求调整）
+
+  # 扩缩容策略
+  scalingMetric:
+    # requestRate:
+    #   granularity: 1s
+    #   targetValue: 2  # 每秒 10 个请求触发扩容
+    #   window: 1m  # 1 分钟时间窗口
+    concurrency:
+      targetValue: 3
+    
+
+  # 扩缩容行为控制
+  scaledownPeriod: 300
+  targetPendingRequests: 8  # 等待处理的请求数阈值
+