AI_DATA/doris-mcp-server
6
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix the way to start server

Browse Source
This commit is contained in:
HuangHaijun
2025-06-19 13:03:17 +08:00
parent e859fbb778
commit 159399bd38
3 changed files with 53 additions and 270 deletions
Download Patch File Download Diff File Expand all files Collapse all files

301
examples/dify/dify_demo.md
View File

@@ -24,84 +24,60 @@ This document demonstrates how to integrate and use `doris-mcp-server` in Dify t
## Table of Contents
- [Prerequisites](#prerequisites)
- [Installation & Configuration](#installation--configuration)
- [Starting the MCP Server](#starting-the-mcp-server)
- [Ngrok Tunnel (Optional)](#ngrok-tunnel-optional)
- [Installing & Configuring the Plugin in Dify](#installing--configuring-the-plugin-in-dify)
- [Creating a Dify App](#creating-a-dify-app)
- [Adding MCP Tools](#adding-mcp-tools)
- [Example Calls](#example-calls)
- [Prerequisites](https://www.google.com/search?q=%23prerequisites)
- [Installation & Configuration](https://www.google.com/search?q=%23installation--configuration)
- [Starting the MCP Server](https://www.google.com/search?q=%23starting-the-mcp-server)
- [Ngrok Tunnel (Optional)](https://www.google.com/search?q=%23ngrok-tunnel-optional)
- [Installing & Configuring the Plugin in Dify](https://www.google.com/search?q=%23installing--configuring-the-plugin-in-dify)
- [Creating a Dify App](https://www.google.com/search?q=%23creating-a-dify-app)
- [Adding MCP Tools](https://www.google.com/search?q=%23adding-mcp-tools)
- [Example Calls](https://www.google.com/search?q=%23example-calls)
---
-----
## Prerequisites
### Prerequisites
1. Clone the repository:
```bash
git clone https://github.com/apache/doris-mcp-server.git
cd doris-mcp-server
```
2. Install the latest Python package and copy the environment template:
```bash
pip install mcp-doris-server
cp .env.example .env
vim .env
```
Fill in your Doris FE connection settings in `.env`:
```env
DORIS_HOST=localhost
DORIS_PORT=9030
DORIS_USER=root
DORIS_PASSWORD=
DORIS_DATABASE=information_schema
DORIS_FE_HTTP_PORT=8030
# Optional: custom BE host list
DORIS_BE_HOSTS=
```
First, install `mcp-doris-server`:
```bash
pip install mcp-doris-server
```
## Starting the MCP Server
Run the startup script:
```bash
./start_server.sh
# Full configuration with database connection
doris-mcp-server \
--transport http \
--host 0.0.0.0 \
--port 3000 \
--db-host 127.0.0.1 \
--db-port 9030 \
--db-user root \
--db-password your_password
```
If successful, you will see logs like this:
If successful, you'll see logs similar to this:
![Server start logs](../images/dify_start_server.png)
---
-----
## Ngrok Tunnel (Optional)
If your Dify deployment requires a public endpoint, use [ngrok](https://ngrok.com):
If your Dify deployment requires a publicly accessible endpoint, you can use the **ngrok** tool. Ngrok is a third-party service that securely exposes local servers to the internet.
```bash
ngrok http http://localhost:3000
```
Use the generated public URL in subsequent steps.
![Ngrok tunnel](../images/dify_ngrok.png)
---
-----
## Installing & Configuring the Plugin in Dify
1. In the Dify console, go to **Plugin Marketplace**, search for and install **MCP‑SSE / StreamableHTTP**:
1. In the Dify console, go to **Plugin Marketplace**, search for, and install **MCP‑SSE / StreamableHTTP**:
![Install plugin](../images/dify_install_plugin.png)
2. After installation, click **Configure** and set the URL to your public or local address (e.g., `https://abc123.ngrok.io/mcp`):
2. After installation, click **Configure** and set the URL to your public or local address. For example, if you're using `ngrok`, this should be the public URL `ngrok` provides, in the format `https://<your-domain>/mcp`. If Dify can directly access your local server, use `http://localhost:3000/mcp`.
```json
{
@@ -111,27 +87,23 @@ Use the generated public URL in subsequent steps.
}
}
```
![Configure plugin](../images/dify_config_mcp.png)
3. Click **Save**. If configured correctly, you will see a green **Authorized** indicator:
3. Click **Save**. If configured correctly, you'll see a green **Authorized** indicator:
![Authorized](../images/dify_authorized.png)
---
-----
## Creating a Dify App
1. In the Dify console, click **New App** → **Blank App**.
![Create app](../images/dify_create_app.png)
2. Select **Agent** as the template, set **App Name** (e.g., `Doris ChatBI`).
2. Select **Agent** as the template and set the **App Name** (e.g., `Doris ChatBI`).
![Agent setup](../images/dify_agent_setup.png)
---
-----
## Instructions & Tool Configuration
@@ -142,18 +114,15 @@ Paste the following into the **Instruction** field:
```
<instruction>
Use MCP tools to complete tasks as much as possible. Carefully read the annotations, method names, and parameter descriptions of each tool. Please follow these steps:
1. Analyze the user's question and match the most appropriate tool.
2. Use tool names and parameters exactly as defined; do not invent new ones.
3. Pass parameters in the required JSON format.
4. When calling tools, use:
{"mcp_sse_call_tool": {"tool_name": "<tool_name>", "arguments": "{}"}}
5. Output plain text only—no XML tags.
<input>
User question: user_query
</input>
<output>
Return tool results or a final answer, including analysis.
</output>
@@ -162,209 +131,23 @@ Return tool results or a final answer, including analysis.
### Adding MCP Tools
In the **Tools** pane, click **Add** twice to add two entries both named `mcp_sse` (they will inherit transport and URL from the plugin):
In the **Tools** pane, click **Add** twice to add two entries, both named `mcp_sse` (they will inherit the transport and URL from the plugin):
![Add tools](../images/dify_add_tools.png)
---
-----
## Example Calls
### List Tables in Database
* **User**: What tables are in the database?
* **Result**: Dify will call the MCP tool to run `SHOW TABLES` and return the list.
* **User**: What tables are in the database?
* **Result**: Dify will call the MCP tool to run `SHOW TABLES` and return the list.
![Query tables](../images/dify_query_tabels.png)
### Sales Trend Over Ten Years
* **User**: What has been the sales trend over the past ten years in the ssb database, and which year had the fastest growth?
* **Result**: The tool will execute the SQL, calculate growth rates, and return a trend chart or table.
* **User**: What has been the sales trend over the past ten years in the ssb database, and which year had the fastest growth?
* **Result**: The tool will execute the SQL, calculate growth rates, and return data.
![Sales trend](../images/dify_sale_trend.png)
````md
<!--
Licensed to the Apache Software Foundation (ASF) under one
or more contributor license agreements. See the NOTICE file
... (license omitted for brevity)
-->
# Dify Example: Integrating Doris MCP Server
This document demonstrates how to integrate and use the `doris-mcp-server` in Dify to perform Doris SQL calls via MCP.
## Table of Contents
- [Prerequisites](#prerequisites)
- [Installation & Configuration](#installation--configuration)
- [Starting the MCP Server](#starting-the-mcp-server)
- [Ngrok Tunnel (Optional)](#ngrok-tunnel-optional)
- [Installing & Configuring the Plugin](#installing--configuring-the-plugin)
- [Creating a Dify App](#creating-a-dify-app)
- [Instructions & Tool Configuration](#instructions--tool-configuration)
- [Example Calls](#example-calls)
---
## Prerequisites
1. Clone the repository:
```bash
git clone https://github.com/apache/doris-mcp-server.git
cd doris-mcp-server
````
2. Install the Python package and prepare the environment file:
```bash
pip install mcp-doris-server
cp .env.example .env
vim .env
```
3. In `.env`, set your Doris front-end connection:
```env
DORIS_HOST=localhost
DORIS_PORT=9030
DORIS_USER=root
DORIS_PASSWORD=
DORIS_DATABASE=information_schema
DORIS_FE_HTTP_PORT=8030
# Optional: custom BE host list
DORIS_BE_HOSTS=
```
---
## Installation & Configuration
1. Install the MCP server package:
```bash
pip install mcp-doris-server
```
2. Copy and edit the environment template as shown above.
---
## Starting the MCP Server
Run the startup script:
```bash
./start_server.sh
```
If successful, you will see logs like this:
![Server start logs](../images/dify_start_server.png)
---
## Ngrok Tunnel (Optional)
If your Dify deployment requires a public endpoint, use [ngrok](https://ngrok.com/):
```bash
ngrok http http://localhost:3000
```
Use the generated URL in the plugin configuration.
![Ngrok tunnel](../images/dify_ngrok.png)
---
## Installing & Configuring the Plugin
1. In the Dify console, go to **Plugin Marketplace**, search for **MCP‑SSE / StreamableHTTP**, and install it.
![Install plugin](../images/dify_install_plugin.png)
2. Click **Configure**, and set the URL (e.g., `https://<your-domain>/mcp`):
```json
{
"doris_mcp_server": {
"transport": "streamable_http",
"url": "https://<your-domain>/mcp"
}
}
```
![Configure plugin](../images/dify_config_mcp.png)
3. Click **Save**. A green **Authorized** indicator should appear:
![Authorized](../images/dify_authorized.png)
---
## Creating a Dify App
1. In the Dify console, click **New App** → **Blank App**.
![Create app](../images/dify_create_app.png)
2. Select **Agent** as the template, set **App Name** (e.g., `Doris ChatBI`).
![Agent setup](../images/dify_agent_setup.png)
---
## Instructions & Tool Configuration
### Instruction Block
Paste the following into the **Instruction** field:
```
<instruction>
Use MCP tools to complete tasks as much as possible. Carefully read the annotations, method names, and parameter descriptions of each tool. Please follow these steps:
1. Analyze the user's question and match the most appropriate tool.
2. Use tool names and parameters exactly as defined; do not invent new ones.
3. Pass parameters in the required JSON format.
4. When calling tools, use:
{"mcp_sse_call_tool": {"tool_name": "<tool_name>", "arguments": "{}"}}
5. Output plain text only—no XML tags.
<input>
User question: user_query
</input>
<output>
Return tool results or a final answer, including analysis.
</output>
</instruction>
```
### Adding MCP Tools
In the **Tools** pane, click **Add** twice to add two entries both named `mcp_sse` (they will inherit transport and URL from the plugin):
![Add tools](../images/dify_add_tools.png)
---
## Example Calls
### List Tables in Database
* **User**: What tables are in the database?
* **Result**: Dify calls the MCP tool to run `SHOW TABLES` and returns the list.
![Query tables](../images/dify_query_tabels.png)
### Sales Trend Over Ten Years
* **User**: What has been the sales trend in the `ssb` database over the past ten years, and which year saw the fastest growth?
* **Result**: The tool executes SQL, computes growth rates, and returns a chart or table.
![Sales trend](../images/dify_sale_trend.png)

BIN
examples/images/dify_ngrok.png
View File

Binary file not shown.
Side by Side

Before

Width:  |  Height:  |  Size: 63 KiB

BIN
examples/images/dify_start_server.png
View File

Binary file not shown.
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 150 KiB

After

Width:  |  Height:  |  Size: 73 KiB