Skip to content

ADMIN_OLLAMA_MANAGEMENT.md

Latest commit

 

History

History
369 lines (298 loc) · 7.68 KB

ADMIN_OLLAMA_MANAGEMENT.md

File metadata and controls

369 lines (298 loc) · 7.68 KB

Admin Dashboard - Ollama Management

Overview

The admin dashboard now includes a comprehensive Ollama management interface, allowing you to manage the AI assistant directly from the web interface without SSH access.

Access

  1. Navigate to: https://politiquera.com/admin
  2. Sign in with Google (superadmin account)
  3. Click the 🧠 AI Assistant tab

Features

1. Service Status Monitor

Real-time status display:

  • ✓ Ollama service running/stopped
  • Version information
  • API availability check
  • Number of installed models
  • One-click refresh button

If Ollama is not installed:

  • Shows installation prompt
  • 📦 Install Ollama button
  • Automated installation process
  • Progress tracking in action log

2. Update Management

Automatic update checker:

  • Checks for new Ollama versions
  • Compares current vs latest version
  • Visual indicator when updates available

One-click updates:

  • ⬆️ Update Now button
  • Downloads and installs latest version
  • Automatically restarts Ollama service
  • Shows update progress in action log
  • Requires superadmin role

Update process:

  1. Click ⬆️ Update Now
  2. Confirm update dialog
  3. System downloads latest version
  4. Ollama service restarts automatically
  5. Status refreshes to show new version

3. Model Management

View installed models:

  • Model name and version
  • File size (GB/MB)
  • Last modified date
  • Recommended model badge (Llama 3.2 3B)

Pull new models:

  1. Click ➕ Pull New Model
  2. Enter model name (e.g., llama3.2:3b-instruct, llama3.2:1b)
  3. System downloads model
  4. Progress shown in action log
  5. Model appears in list when complete

Available models:

  • llama3.2:3b-instruct (Recommended, 2GB)
  • llama3.2:1b (Lightweight, 1GB)
  • llama3.2:8b-instruct (High quality, 5GB)
  • phi-3:mini (Alternative, 1.5GB)
  • Any model from Ollama registry

Test models:

  • Click 🧪 Test button on any model
  • Runs sample SQL generation query
  • Shows response time (ms)
  • Displays sample output
  • Verifies model is working correctly

Delete models:

  • Click 🗑️ Delete button
  • Confirmation dialog
  • Permanently removes model
  • Frees up disk space
  • Requires superadmin role

4. Performance Statistics

Metrics displayed:

  • Total queries processed
  • Average response time
  • Success rate percentage
  • Memory usage

Future enhancements:

  • Query history
  • Performance graphs
  • Usage trends
  • Error rate tracking

5. Action Log

Real-time operation tracking:

  • Timestamped entries
  • All operations logged
  • Success/failure indicators
  • Detailed error messages
  • Auto-scrolls to latest entry

Logged operations:

  • Ollama installation
  • Version updates
  • Model pulls
  • Model deletions
  • Model tests
  • Status checks

API Endpoints

All endpoints require authentication. Some require superadmin role.

GET /api/admin/ollama/status

Auth: Required
Returns: Service status, version, models list

{
  "ollama_installed": true,
  "ollama_version": "0.1.17",
  "api_available": true,
  "models_count": 1,
  "models": [
    {
      "name": "llama3.2:3b-instruct",
      "size": 2147483648,
      "modified_at": "2024-03-04T12:00:00Z"
    }
  ]
}

GET /api/admin/ollama/check-updates

Auth: Required
Returns: Update availability status

{
  "update_available": true,
  "current_version": "0.1.16",
  "latest_version": "0.1.17"
}

POST /api/admin/ollama/update

Auth: Superadmin required
Returns: Update result

{
  "success": true,
  "output": "Ollama updated successfully..."
}

POST /api/admin/ollama/install

Auth: Superadmin required
Returns: Installation result

{
  "success": true,
  "output": "Ollama installed successfully..."
}

POST /api/admin/ollama/pull-model

Auth: Required
Body: {"model": "llama3.2:3b-instruct"}
Returns: Pull result

{
  "success": true,
  "model": "llama3.2:3b-instruct"
}

POST /api/admin/ollama/delete-model

Auth: Superadmin required
Body: {"model": "llama3.2:3b-instruct"}
Returns: Deletion result

{
  "success": true,
  "model": "llama3.2:3b-instruct"
}

POST /api/admin/ollama/test-model

Auth: Required
Body: {"model": "llama3.2:3b-instruct"}
Returns: Test result

{
  "success": true,
  "model": "llama3.2:3b-instruct",
  "response_time": 2341,
  "sample_output": "SELECT * FROM voters WHERE congressional_district = 'TX-15'..."
}

GET /api/admin/ollama/stats

Auth: Required
Returns: Usage statistics

{
  "total_queries": 1247,
  "avg_response_time": "2.3s",
  "success_rate": "98.5%",
  "memory_usage": "2.1 GB"
}

Security

Role-Based Access Control

Authenticated Users:

  • View status
  • Check for updates
  • Pull new models
  • Test models
  • View statistics

Superadmin Only:

  • Install Ollama
  • Update Ollama
  • Delete models

Audit Trail

All operations are logged with:

  • Timestamp
  • User who performed action
  • Operation type
  • Success/failure status
  • Error details (if failed)

Troubleshooting

"Ollama Not Installed"

  1. Click 📦 Install Ollama button
  2. Wait for installation to complete
  3. Check action log for progress
  4. Refresh status after installation

"API Not Available"

  1. Check if Ollama service is running: 
    systemctl status ollama
  2. Restart service if needed: 
    systemctl restart ollama
  3. Refresh status in admin dashboard

Update Fails

  1. Check action log for error details
  2. Try manual update via SSH: 
    curl -fsSL https://ollama.com/install.sh | sh
systemctl restart ollama
  3. Refresh admin dashboard

Model Pull Fails

  1. Check internet connectivity
  2. Verify model name is correct
  3. Check disk space (models are 1-5GB)
  4. Try again or use different model

Model Test Fails

  1. Check if model is fully downloaded
  2. Verify Ollama service is running
  3. Try restarting Ollama service
  4. Check action log for error details

Best Practices

Model Selection

  • Start with: llama3.2:3b-instruct (recommended, 2GB)
  • For low memory: llama3.2:1b (1GB)
  • For better quality: llama3.2:8b-instruct (5GB, slower)

Updates

  • Check for updates weekly
  • Update during low-traffic periods
  • Test AI search after updates
  • Keep at least one model installed

Disk Space

  • Each model: 1-5GB
  • Keep 10GB free for operations
  • Delete unused models
  • Monitor disk usage

Performance

  • Test models after installation
  • Monitor response times
  • Check success rates
  • Restart service if slow

Maintenance Schedule

Daily

  • Check service status
  • Review action log for errors

Weekly

  • Check for Ollama updates
  • Review performance stats
  • Test models if issues reported

Monthly

  • Clean up unused models
  • Review disk space usage
  • Update to latest version

Future Enhancements

Planned Features

  • Automatic update scheduling
  • Model performance comparison
  • Query history viewer
  • Usage analytics dashboard
  • Email alerts for failures
  • Backup/restore configurations
  • Multi-model A/B testing
  • Custom model fine-tuning

Integration Ideas

  • Slack notifications for updates
  • Prometheus metrics export
  • Grafana dashboard integration
  • Automated health checks
  • Load balancing multiple models

Support

Getting Help

  1. Check action log for error details
  2. Review troubleshooting section
  3. Check Ollama documentation: https://ollama.com/docs
  4. Contact system administrator

Reporting Issues

Include:

  • Screenshot of error
  • Action log entries
  • Service status details
  • Steps to reproduce

Admin Dashboard Ollama Management - Making AI assistant management accessible and easy!