feat：Pr/add docker compose

.gitignore

@@ -33,3 +33,6 @@ servers/*/server.yaml
 
 # Python package metadata
 src/*.egg-info/
+
+# Docker volumes
+docker/volumes/

docker/.env.cpu.example

@@ -0,0 +1,22 @@
+# CPU-only compose vars
+UI_PORT=5050
+MILVUS_PORT=19530
+MILVUS_METRICS_PORT=9091
+MINIO_API_PORT=9000
+MINIO_CONSOLE_PORT=9001
+MINIO_ROOT_USER=minioadmin
+MINIO_ROOT_PASSWORD=minioadmin
+
+# ----------------------------
+# UI defaults for external LLM/Embedding providers
+# These values auto-fill in the UI settings panel.
+# ----------------------------
+# ULTRARAG_UI_DEFAULT_AI_PROVIDER options: openai | custom | azure | anthropic
+ULTRARAG_UI_DEFAULT_AI_PROVIDER=openai
+ULTRARAG_UI_DEFAULT_AI_BASE_URL=https://api.openai.com/v1
+ULTRARAG_UI_DEFAULT_AI_API_KEY=
+ULTRARAG_UI_DEFAULT_AI_MODEL=gpt-5-mini
+
+ULTRARAG_UI_DEFAULT_EMB_BASE_URL=https://api.openai.com/v1
+ULTRARAG_UI_DEFAULT_EMB_API_KEY=
+ULTRARAG_UI_DEFAULT_EMB_MODEL_NAME=text-embedding-3-small

docker/.env.example

@@ -0,0 +1,52 @@
+# ----------------------------
+# UltraRAG UI
+# ----------------------------
+UI_PORT=5050
+
+# ----------------------------
+# Milvus / MinIO ports (host)
+# ----------------------------
+MILVUS_PORT=19530
+MILVUS_METRICS_PORT=9091
+MINIO_API_PORT=9000
+MINIO_CONSOLE_PORT=9001
+MINIO_ROOT_USER=minioadmin
+MINIO_ROOT_PASSWORD=minioadmin
+
+# ----------------------------
+# LLM (vLLM OpenAI-compatible API)
+# ----------------------------
+# Example:
+# LLM_MODEL_PATH=Qwen/Qwen3-8B
+# LLM_MODEL_NAME=qwen3-8b
+LLM_MODEL_PATH=Qwen/Qwen3-8B
+LLM_MODEL_NAME=qwen3-8b
+LLM_DTYPE=auto
+LLM_MAX_MODEL_LEN=32768
+LLM_GPU_MEMORY_UTILIZATION=0.9
+LLM_API_PORT=8000
+
+# ----------------------------
+# Embedding (vLLM OpenAI-compatible API)
+# ----------------------------
+# Example:
+# EMB_MODEL_PATH=Qwen/Qwen3-Embedding-0.6B
+# EMB_MODEL_NAME=qwen-embedding
+EMB_MODEL_PATH=Qwen/Qwen3-Embedding-0.6B
+EMB_MODEL_NAME=qwen-embedding
+EMB_DTYPE=auto
+EMB_GPU_MEMORY_UTILIZATION=0.8
+EMB_API_PORT=8001
+
+# ----------------------------
+# UI defaults (auto-filled in settings panel)
+# ----------------------------
+# ULTRARAG_UI_DEFAULT_AI_PROVIDER options: openai | custom | azure | anthropic
+ULTRARAG_UI_DEFAULT_AI_PROVIDER=custom
+ULTRARAG_UI_DEFAULT_AI_BASE_URL=http://vllm-llm:8000/v1
+ULTRARAG_UI_DEFAULT_AI_API_KEY=dummy
+ULTRARAG_UI_DEFAULT_AI_MODEL=qwen3-8b
+
+ULTRARAG_UI_DEFAULT_EMB_BASE_URL=http://vllm-emb:8000/v1
+ULTRARAG_UI_DEFAULT_EMB_API_KEY=dummy
+ULTRARAG_UI_DEFAULT_EMB_MODEL_NAME=qwen-embedding

docker/README.md

@@ -0,0 +1,117 @@
+# UltraRAG One-Click Compose
+
+This folder provides two compose setups:
+
+- `docker-compose.yml` (GPU): UI + Milvus + vLLM(LLM) + vLLM(Embedding)
+- `docker-compose.cpu.yml` (CPU-only): UI + Milvus (no vLLM)
+
+GPU setup includes:
+
+- UltraRAG UI
+- Milvus (standalone with etcd + minio)
+- vLLM OpenAI-compatible LLM endpoint
+- vLLM OpenAI-compatible Embedding endpoint
+
+## 1) Prepare env
+
+```bash
+cd /Users/rqq/UltraRAG/docker
+cp .env.example .env
+```
+
+Edit `.env` at least for these fields:
+
+- `LLM_MODEL_PATH`
+- `LLM_MODEL_NAME`
+- `EMB_MODEL_PATH`
+- `EMB_MODEL_NAME`
+- Optional UI auto-fill defaults:
+  - `ULTRARAG_UI_DEFAULT_AI_BASE_URL`
+  - `ULTRARAG_UI_DEFAULT_AI_MODEL`
+  - `ULTRARAG_UI_DEFAULT_EMB_BASE_URL`
+  - `ULTRARAG_UI_DEFAULT_EMB_MODEL_NAME`
+
+## 2) Start all services
+
+```bash
+docker compose up -d --build
+```
+
+Open UI:
+
+- [http://localhost:5050](http://localhost:5050)
+
+## 3) Recommended UI settings
+
+Inside UltraRAG UI:
+
+- Knowledge Base -> Milvus URI should be `tcp://milvus-standalone:19530`
+  - Already pre-mounted from `docker/config/kb_config.json`.
+- Embedding config:
+  - `Base URL`: `http://vllm-emb:8000/v1`
+  - `Model Name`: same as `EMB_MODEL_NAME` in `.env`
+  - `API Key`: optional (for local vLLM you can use any non-empty value if required by UI)
+- AI assistant / generation config:
+  - `Base URL`: `http://vllm-llm:8000/v1`
+  - `Model`: same as `LLM_MODEL_NAME` in `.env`
+  - `API Key`: optional (for local vLLM you can use any non-empty value)
+
+## 4) Stop
+
+```bash
+docker compose down
+```
+
+To remove volumes too:
+
+```bash
+docker compose down -v
+```
+
+## Notes
+
+- `vllm-llm` and `vllm-emb` require NVIDIA GPU + working Docker GPU runtime.
+- First startup may take a long time due to model/image downloads.
+- Persistent data is under `docker/volumes/`.
+
+---
+
+## CPU-only setup (no vLLM)
+
+If your machine has no NVIDIA GPU, use this flow.
+
+### 1) Prepare env
+
+```bash
+cd /Users/rqq/UltraRAG/docker
+cp .env.cpu.example .env
+```
+
+### 2) Start CPU stack
+
+```bash
+docker compose -f docker-compose.cpu.yml up -d --build
+```
+
+Open UI:
+
+- [http://localhost:5050](http://localhost:5050)
+
+### 3) UI settings for CPU mode
+
+- Knowledge Base -> Milvus URI:
+  - `tcp://milvus-standalone:19530`
+  - Already pre-mounted from `docker/config/kb_config.json`.
+- Embedding and Generation models:
+  - Use external OpenAI-compatible API (OpenAI/Azure/other hosted endpoint).
+  - You can prefill UI via `.env`:
+    - `ULTRARAG_UI_DEFAULT_AI_*`
+    - `ULTRARAG_UI_DEFAULT_EMB_*`
+  - In UI you can still manually change provider `Base URL`, `Model`, `API Key`.
+  - CPU compose does not start local vLLM endpoints.
+
+### 4) Stop CPU stack
+
+```bash
+docker compose -f docker-compose.cpu.yml down
+```

docker/config/kb_config.json

@@ -0,0 +1,21 @@
+{
+  "milvus": {
+    "uri": "tcp://milvus-standalone:19530",
+    "token": "",
+    "id_field_name": "id",
+    "vector_field_name": "vector",
+    "text_field_name": "contents",
+    "id_max_length": 64,
+    "text_max_length": 60000,
+    "metric_type": "IP",
+    "index_params": {
+      "index_type": "AUTOINDEX",
+      "metric_type": "IP"
+    },
+    "search_params": {
+      "metric_type": "IP",
+      "params": {}
+    },
+    "index_chunk_size": 1000
+  }
+}

docker/docker-compose.cpu.yml

@@ -0,0 +1,84 @@
+services:
+  ultrarag-ui:
+    build:
+      context: ..
+      dockerfile: Dockerfile.base-cpu
+    container_name: ultrarag-ui-cpu
+    restart: unless-stopped
+    env_file:
+      - .env
+    environment:
+      PYTHONUNBUFFERED: "1"
+    command:
+      - ultrarag
+      - show
+      - ui
+      - --admin
+      - --port
+      - "5050"
+      - --host
+      - 0.0.0.0
+    ports:
+      - "${UI_PORT:-5050}:5050"
+    volumes:
+      - ../data:/ultrarag/data
+      - ../output:/ultrarag/output
+      - ../logs:/ultrarag/logs
+      - ./config/kb_config.json:/ultrarag/data/knowledge_base/kb_config.json
+    depends_on:
+      - milvus-standalone
+    networks:
+      - ultrarag-net
+
+  etcd:
+    container_name: ultrarag-etcd-cpu
+    image: quay.io/coreos/etcd:v3.5.12
+    restart: unless-stopped
+    command: >
+      etcd
+      --advertise-client-urls=http://etcd:2379
+      --listen-client-urls=http://0.0.0.0:2379
+      --data-dir=/etcd
+    volumes:
+      - ./volumes/etcd:/etcd
+    networks:
+      - ultrarag-net
+
+  minio:
+    container_name: ultrarag-minio-cpu
+    image: minio/minio:RELEASE.2024-12-18T13-15-44Z
+    restart: unless-stopped
+    environment:
+      MINIO_ROOT_USER: "${MINIO_ROOT_USER:-minioadmin}"
+      MINIO_ROOT_PASSWORD: "${MINIO_ROOT_PASSWORD:-minioadmin}"
+    command: minio server /minio_data --console-address ":9001"
+    ports:
+      - "${MINIO_API_PORT:-9000}:9000"
+      - "${MINIO_CONSOLE_PORT:-9001}:9001"
+    volumes:
+      - ./volumes/minio:/minio_data
+    networks:
+      - ultrarag-net
+
+  milvus-standalone:
+    container_name: ultrarag-milvus-cpu
+    image: milvusdb/milvus:v2.5.4
+    restart: unless-stopped
+    command: ["milvus", "run", "standalone"]
+    environment:
+      ETCD_ENDPOINTS: etcd:2379
+      MINIO_ADDRESS: minio:9000
+    ports:
+      - "${MILVUS_PORT:-19530}:19530"
+      - "${MILVUS_METRICS_PORT:-9091}:9091"
+    volumes:
+      - ./volumes/milvus:/var/lib/milvus
+    depends_on:
+      - etcd
+      - minio
+    networks:
+      - ultrarag-net
+
+networks:
+  ultrarag-net:
+    name: ultrarag-net