ollama-fix-instructions.md

Fix for Ollama Installation

Problem

Backend integration tests fail when Ollama cannot download llama3.2 model due to network issues.

Solution

Replace the Install Ollama section in .github/workflows/ci.yml (lines ~286-307) with:

# Install Ollama for AI model testing
- name: Install Ollama
  run: |
    echo "🤖 Installing Ollama with retry logic..."
    curl -fsSL https://ollama.com/install.sh | sh
    
    # Install and start Ollama service
    ollama serve &
    
    # Wait for Ollama to start
    sleep 15
    
    # Pull model with retry logic
    echo "📥 Pulling llama3.2 model with retry logic..."
    MAX_RETRIES=3
    RETRY_DELAY=30
    MODEL_PULLED=false
    
    for i in ; do
      echo "Attempt  of  to pull llama3.2..."
      
      # Use timeout and background process with longer timeout (20 minutes)
      timeout 1200 ollama pull llama3.2 &&
      {
        echo "✅ Model pull successful!"
        MODEL_PULLED=true
        break
      } ||
      {
        echo "❌ Model pull failed (attempt )"
        if [  -eq  ]; then
          echo "🚨 All retry attempts failed"
          echo "⚠️ Continuing without llama3.2 model - tests will skip model-dependent features"
          break
        fi
        echo "⏳ Waiting  seconds before retry..."
        sleep 
      }
    done
    
    # Verify installation
    echo "Final Ollama status:"
    ollama list || echo "⚠️ Cannot list models - model may not be available"
    
    # Export variable for tests to check model availability
    if [ "" = "true" ]; then
      echo "MODEL_AVAILABLE=true" >> 
    else
      echo "MODEL_AVAILABLE=false" >> 
    fi

Key Improvements

1. Retry Logic: 3 attempts with 30-second delays
2. Extended Timeout: 20 minutes per attempt (vs default timeout)
3. Graceful Failure: Continue workflow even if model download fails
4. Environment Variable: Set MODEL_AVAILABLE for tests to check
5. Better Logging: Clear status messages for debugging

Result

Tests will run to completion even if Ollama model download fails, fixing the CI issue.