Always Use Model Server (#1306)

deployment/kubernetes/env-configmap.yaml

@@ -43,9 +43,9 @@ data:
   ASYM_PASSAGE_PREFIX: ""
   ENABLE_RERANKING_REAL_TIME_FLOW: ""
   ENABLE_RERANKING_ASYNC_FLOW: ""
-  MODEL_SERVER_HOST: ""
+  MODEL_SERVER_HOST: "inference-model-server-service"
   MODEL_SERVER_PORT: ""
-  INDEXING_MODEL_SERVER_HOST: ""
+  INDEXING_MODEL_SERVER_HOST: "indexing-model-server-service"
   MIN_THREADS_ML_MODELS: ""
   # Indexing Configs
   NUM_INDEXING_WORKERS: ""

deployment/kubernetes/indexing_model_server-service-deployment.yaml

@@ -0,0 +1,59 @@
+apiVersion: v1
+kind: Service
+metadata:
+  name: indexing-model-server-service
+spec:
+  selector:
+    app: indexing-model-server
+  ports:
+    - name: indexing-model-server-port
+      protocol: TCP
+      port: 9000
+      targetPort: 9000
+  type: ClusterIP
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: indexing-model-server-deployment
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: indexing-model-server
+  template:
+    metadata:
+      labels:
+        app: indexing-model-server
+    spec:
+      containers:
+      - name: indexing-model-server
+        image: danswer/danswer-model-server:latest
+        imagePullPolicy: IfNotPresent
+        command: [ "uvicorn", "model_server.main:app", "--host", "0.0.0.0", "--port", "9000" ]
+        ports:
+        - containerPort: 9000
+        envFrom:
+        - configMapRef:
+            name: env-configmap
+        env:
+          - name: INDEXING_ONLY
+            value: "True"
+        volumeMounts:
+        - name: indexing-model-storage
+          mountPath: /root/.cache
+      volumes:
+      - name: indexing-model-storage
+        persistentVolumeClaim:
+          claimName: indexing-model-pvc
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: indexing-model-pvc
+spec:
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 3Gi

deployment/kubernetes/inference_model_server-service-deployment.yaml

@@ -0,0 +1,56 @@
+apiVersion: v1
+kind: Service
+metadata:
+  name: inference-model-server-service
+spec:
+  selector:
+    app: inference-model-server
+  ports:
+    - name: inference-model-server-port
+      protocol: TCP
+      port: 9000
+      targetPort: 9000
+  type: ClusterIP
+---
+apiVersion: apps/v1
+kind: Deployment
+metadata:
+  name: inference-model-server-deployment
+spec:
+  replicas: 1
+  selector:
+    matchLabels:
+      app: inference-model-server
+  template:
+    metadata:
+      labels:
+        app: inference-model-server
+    spec:
+      containers:
+      - name: inference-model-server
+        image: danswer/danswer-model-server:latest
+        imagePullPolicy: IfNotPresent
+        command: [ "uvicorn", "model_server.main:app", "--host", "0.0.0.0", "--port", "9000" ]
+        ports:
+        - containerPort: 9000
+        envFrom:
+        - configMapRef:
+            name: env-configmap
+        volumeMounts:
+        - name: inference-model-storage
+          mountPath: /root/.cache
+      volumes:
+      - name: inference-model-storage
+        persistentVolumeClaim:
+          claimName: inference-model-pvc
+---
+apiVersion: v1
+kind: PersistentVolumeClaim
+metadata:
+  name: inference-model-pvc
+spec:
+  accessModes:
+    - ReadWriteOnce
+  resources:
+    requests:
+      storage: 3Gi