Advanced Features

This guide covers the advanced features and capabilities of Aria that power users might find useful.

Custom Model Configuration

Aria uses Ollama to run AI models. You can customize the model behavior by adjusting parameters in your .env file:

OLLAMA_MODEL_ID=cogito:14b
OLLAMA_MODEL_TEMPARATURE=0.7
OLLAMA_MODEL_CONTEXT_LENGTH=8192

Supported Models

Aria works with any model supported by Ollama, including:

To use a different model, simply change the OLLAMA_MODEL_ID in your .env file.

API Integration

Aria provides a REST API that allows you to integrate it with other applications:

Authentication

curl -X POST http://localhost:8000/api/auth/token \
  -H "Content-Type: application/json" \
  -d '{"username": "your-username", "password": "your-password"}'

Chat Completion

curl -X POST http://localhost:8000/api/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_TOKEN" \
  -d '{
    "messages": [
      {"role": "user", "content": "Hello, how are you?"}
    ],
    "stream": true
  }'

Custom Plugins

Aria supports custom plugins that extend its functionality. Plugins are Python modules that implement specific interfaces:

from aria.plugins import AriaPlugin

class MyCustomPlugin(AriaPlugin):
    def __init__(self):
        super().__init__(
            name="my_custom_plugin",
            description="A custom plugin for Aria"
        )
    
    async def execute(self, query, context):
        # Your plugin logic here
        return {"result": "Custom plugin response"}

Place your plugin in the plugins directory and restart Aria to load it.

Advanced Search Configuration

SearXNG, the search engine used by Aria, can be customized by editing the searxng/settings.yml file:

search:
  safe_search: 0  # 0: None, 1: Moderate, 2: Strict
  autocomplete: 'google'
  default_lang: 'en'

Performance Tuning

For better performance on resource-constrained systems:

  1. Reduce the number of workers in searxng/uwsgi.ini: 
    workers = 2
threads = 2
  2. Use a smaller model by setting OLLAMA_MODEL_ID to a more compact model: 
    OLLAMA_MODEL_ID=tinyllama:1.1b
  3. Adjust Redis cache settings in your docker-compose.yml: 
    redis:
  command: redis-server --maxmemory 256mb --maxmemory-policy allkeys-lru

Backup and Restore

Backup

To backup your Aria data:

docker-compose stop
tar -czvf aria-backup.tar.gz data/
docker-compose start

Restore

To restore from a backup:

docker-compose stop
rm -rf data/
tar -xzvf aria-backup.tar.gz
docker-compose start

Troubleshooting

Logs

View logs for troubleshooting:

# All services
docker-compose logs

# Specific service
docker-compose logs aria

Debug Mode

Enable debug mode for more detailed logs:

DEBUG_MODE=true

Common Issues

  1. Connection refused to Ollama:
    • Ensure Ollama is running and accessible at the URL specified in OLLAMA_URL
    • Check network connectivity between the Aria container and Ollama
  2. High memory usage:
    • Use a smaller model
    • Reduce the context length with OLLAMA_MODEL_CONTEXT_LENGTH
  3. Slow responses:
    • Check CPU usage and consider using a more powerful machine
    • Reduce the number of concurrent users

Next Steps