refactor(docker): 将GPU配置分离到独立的docker-compose_gpu.yml文件

为了更清晰地管理GPU和非GPU环境，将GPU相关的资源配置从docker-compose.yml中分离出来，并创建了独立的docker-compose_gpu.yml文件。

docker/docker-compose.yml

@@ -72,12 +72,6 @@ services:
       - "host.docker.internal:host-gateway"
     networks:
       - ragflow
-    deploy:
-      resources:
-        reservations:
-          devices:
-            - driver: nvidia
-              capabilities: [gpu]
 
   # vllm 嵌入模型
   vllm-bge:
@@ -94,12 +88,6 @@ services:
     ]
     ports:
       - "8000:8000"
-    deploy:
-      resources:
-        reservations:
-          devices:
-            - driver: nvidia
-              capabilities: [gpu]
     networks:
       - ragflow
 
@@ -119,11 +107,5 @@ services:
     ]
     ports:
       - "8001:8000"
-    deploy:
-      resources:
-        reservations:
-          devices:
-            - driver: nvidia
-              capabilities: [gpu]
     networks:
       - ragflow

docker/docker-compose_gpu.yml

@@ -0,0 +1,129 @@
+include:
+  - ./docker-compose-base.yml
+
+services:
+  ragflow:
+    depends_on:
+      mysql:
+        condition: service_healthy
+    image: ${RAGFLOW_IMAGE}
+    container_name: ragflowplus-server
+    ports:
+      - ${SVR_HTTP_PORT}:9380
+      - 80:80
+      - 443:443
+    volumes:
+      - ./ragflow-logs:/ragflow/logs
+      - ./nginx/ragflow.conf:/etc/nginx/conf.d/ragflow.conf
+      - ./nginx/proxy.conf:/etc/nginx/proxy.conf
+      - ./nginx/nginx.conf:/etc/nginx/nginx.conf
+    env_file: .env
+    environment:
+      - TZ=${TIMEZONE}
+      - HF_ENDPOINT=${HF_ENDPOINT}
+      - MACOS=${MACOS}
+    networks:
+      - ragflow
+    restart: on-failure
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
+
+  # 后台信息管理系统前端
+  management-frontend:
+    container_name: ragflowplus-management-frontend
+    image: zstar1003/ragflowplus-management-web:v0.2.0
+    build:
+      context: .
+      dockerfile: Dockerfile
+      target: frontend
+    ports:
+      - "8888:80"
+    depends_on:
+      - management-backend
+    environment:
+      - API_BASE_URL=/api 
+    networks:
+      - ragflow
+      
+  # 后台信息管理系统后端
+  management-backend:
+    container_name: ragflowplus-management-backend
+    image: zstar1003/ragflowplus-management-server:v0.2.0
+    build:
+      context: .
+      dockerfile: Dockerfile
+      target: backend
+    ports:
+      - "5000:5000"
+    volumes:
+      - ./magic-pdf.json:/root/magic-pdf.json
+    depends_on: 
+      mysql:
+        condition: service_healthy
+      es01:
+        condition: service_healthy 
+    environment:
+      - FLASK_ENV=development
+      - CORS_ALLOWED_ORIGINS=http://management-frontend
+      - MANAGEMENT_ADMIN_USERNAME=${MANAGEMENT_ADMIN_USERNAME:-admin}
+      - MANAGEMENT_ADMIN_PASSWORD=${MANAGEMENT_ADMIN_PASSWORD:-12345678}
+      - MANAGEMENT_JWT_SECRET=${MANAGEMENT_JWT_SECRET:-12345678}
+    extra_hosts:
+      - "host.docker.internal:host-gateway"
+    networks:
+      - ragflow
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              capabilities: [gpu]
+
+  # vllm 嵌入模型
+  vllm-bge:
+    container_name: vllm-bge
+    image: vllm/vllm-openai:latest
+    ipc: host
+    volumes:
+      - ./models/bge-m3:/models
+    command: [
+      "--model", "/models",
+      "--served-model-name", "bge-m3",
+      "--dtype", "float16",
+      "--gpu-memory-utilization", "0.9",
+    ]
+    ports:
+      - "8000:8000"
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              capabilities: [gpu]
+    networks:
+      - ragflow
+
+  # vllm 语言模型
+  vllm-deepseek:
+    container_name: vllm-deepseek
+    image: vllm/vllm-openai:latest
+    ipc: host
+    volumes:
+      - ./models/DeepSeek-R1-1.5B:/models
+    command: [
+      "--model", "/models",
+      "--served-model-name", "deepseek-r1",
+      "--dtype", "float16",
+      "--tensor-parallel-size", "1",
+      "--max-model-len", "4096"
+    ]
+    ports:
+      - "8001:8000"
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              capabilities: [gpu]
+    networks:
+      - ragflow

management/server/services/knowledgebases/service.py

@@ -909,6 +909,7 @@ class KnowledgebaseService:
 
         conn = None
         cursor = None
+        # TUDO: 修改查询逻辑
         try:
             conn = cls._get_db_connection()
             cursor = conn.cursor(dictionary=True) # 使用字典游标方便访问列名
@@ -916,6 +917,7 @@ class KnowledgebaseService:
                 SELECT llm_name, api_key, api_base
                 FROM tenant_llm
                 WHERE tenant_id = %s
+                AND model type = 'embedding'
                 LIMIT 1
             """
             cursor.execute(query, (tenant_id,))