ds_zagres_ai/DEPLOYMENT.md

Deployment Instructions

This document provides instructions for deploying the chatbot application with Ollama and OpenWebUI integration.

Prerequisites

• Python 3.8 or higher
• pip
• virtualenv or venv
• Access to OpenWebUI at http://104.225.217.215:8080

Deployment Steps

1. Clone the repository

   git clone <repository-url>
   cd <repository-directory>
   

2. Create and activate a virtual environment

   python -m venv venv
   source venv/bin/activate  # On Windows: venv\Scripts\activate
   

3. Install dependencies

   pip install -r requirements.txt
   pip install python-dotenv langchain-text-splitters
   

4. Create a .env file

   Copy the .env.example file to .env and update the values:

   cp ai_service/.env.example ai_service/.env
   # Edit the .env file with appropriate values
   

   Make sure to include the OpenWebUI configuration:

   # OpenWebUI configuration
   OPENWEBUI_URL=http://104.225.217.215:8080
   OPENWEBUI_API_KEY=GdCU4ieYDqHsLfH2
   
   # Ollama configuration
   OLLAMA_API_URL=http://104.225.217.215:8080/ollama
   DEFAULT_MODEL=llama3.1
   

5. Run the deployment script

   For local deployment:

   python -m ai_service.run
   

   For server deployment:

   ./ai_service/deploy.sh
   

   This will start the application on port 5251 using uvicorn with nohup.

   For remote deployment from your local machine:

   ./remote_deploy.sh 157.157.221.29 user 22 /home/user/ds_zagres_ai
   

6. Verify the application is running

   curl http://localhost:5251/api/health
   

   You should see a response like:

   {
     "status": "healthy"
   }
   

Managing the Deployed Application

• View logs

  tail -f app.log
  

• Stop the application

  ps aux | grep uvicorn  # Find the process ID (PID)
  kill <PID>             # Replace <PID> with the actual process ID
  

• Restart the application

  ./deploy.sh
  

API Endpoints

• GET /health - Health check endpoint
• POST /chats - Create a new chat
• POST /chats/{chat_id}/messages - Send a message to the chatbot
• GET /chats/{chat_id} - Get chat history

Ollama and OpenWebUI Integration

The chatbot now uses Ollama models via OpenWebUI. The following models are available:

• gemma3: Google Gemma 3 model
• llama3.3: Meta Llama 3 70B model
• llama3.1: Meta Llama 3 8B model
• mistral: Mistral AI model
• deepseek: DeepSeek model

Document Training

To use RAG with your documents:

1. Go to the OpenWebUI interface at http://104.225.217.215:8080/
2. Navigate to the Knowledge section
3. Upload your documents
4. OpenWebUI will automatically process them for RAG

When using the chatbot API, set use_rag=True in your chat requests to enable RAG.