Ollama Integration

Integrate OpenRegister with Ollama to run local Large Language Models (LLMs) for AI-powered features like chat, agents, and RAG (Retrieval Augmented Generation).

Overview

Ollama allows you to run open-source LLMs locally on your machine, providing:

• Privacy: Your data never leaves your infrastructure
• Cost-effective: No API costs for inference
• Customization: Run any compatible model
• Offline capability: Works without internet connection

Technical Implementation

OpenRegister uses LLPhant's native Ollama support, which provides:

• Direct API Integration: Uses Ollama's native /api/ endpoints (not OpenAI-compatible layer)
• Better Performance: Optimized for Ollama's specific API structure
• Simpler Configuration: No need for API key workarounds
• Full Feature Support: Native support for chat, embeddings, and function calling

Prerequisites

• Nextcloud 28+ with OpenRegister installed
• Docker and Docker Compose
• GPU recommended (8GB+ VRAM) for optimal performance
• At least 16GB RAM for larger models

Quick Start

Step 1: Start Ollama Container

The Ollama container is included in the docker-compose configuration:

# Start all services including Ollama
docker-compose up -d

# Or specifically start Ollama
docker-compose up -d ollama

Step 2: Pull a Model

Pull one of the supported models:

# Llama 3.2 (8B) - Recommended, best balance
docker exec openregister-ollama ollama pull llama3.2

# Llama 3.2 (3B) - Lighter alternative
docker exec openregister-ollama ollama pull llama3.2:3b

# Mistral (7B) - Fast and efficient
docker exec openregister-ollama ollama pull mistral:7b

# Phi-3 Mini (3.8B) - Lightweight option
docker exec openregister-ollama ollama pull phi3:mini

# CodeLlama (7B) - Optimized for code
docker exec openregister-ollama ollama pull codellama:latest

Step 3: Pull Embedding Model (for RAG)

If using RAG features, pull an embedding model:

# Nomic Embed Text - Recommended for embeddings
docker exec openregister-ollama ollama pull nomic-embed-text:latest

# Alternative: all-minilm (smaller, faster)
docker exec openregister-ollama ollama pull all-minilm:latest

Step 4: Configure OpenRegister

1. Navigate to Settings → OpenRegister → LLM Configuration
2. Select Ollama as the chat provider
3. Configure the settings:

Integrated Setup (docker-compose.yml):

• Ollama URL: http://openregister-ollama:11434 ⚠️ NOT http://localhost:11434
• Chat Model: llama3.2:latest (use full name including tag)
• Embedding Model: nomic-embed-text:latest
• Why not localhost? OpenRegister runs inside Nextcloud container; use container name instead

Standalone Setup:

• Via Docker Network: http://standalone-ollama:11434 (after connecting networks)
• Via Host IP: http://YOUR_HOST_IP:11434 (e.g., http://192.168.1.100:11434)
• ❌ NOT: http://localhost:11434 (won't work from inside container)

Configuration Details

Ollama Service Configuration

The Ollama service is configured in docker-compose.yml:

ollama:
  image: ollama/ollama:latest
  container_name: openregister-ollama
  restart: always
  ports:
    - "11434:11434"
  volumes:
    - ollama:/root/.ollama
  environment:
    - OLLAMA_HOST=0.0.0.0
    - OLLAMA_NUM_PARALLEL=4
    - OLLAMA_KEEP_ALIVE=30m
  deploy:
    resources:
      limits:
        memory: 16G
      reservations:
        memory: 8G
        devices:
          - driver: nvidia
            count: all
            capabilities: [gpu]

Accessing Ollama

Important: Docker Container Communication

When configuring Ollama in OpenRegister, you MUST use the Docker service name, not localhost:

• ✅ Correct (from Nextcloud container): http://openregister-ollama:11434 or http://ollama:11434
• ❌ Wrong: http://localhost:11434 (this only works from your host machine, not from inside containers)

Access Points:

• From host machine (terminal, browser): http://localhost:11434
• From Nextcloud container (OpenRegister settings): http://openregister-ollama:11434
• From other Docker containers: http://openregister-ollama:11434

Why? Inside a Docker container, localhost refers to the container itself, not your host machine. Containers communicate with each other using service names defined in docker-compose.yml.

Model Naming

Important: Ollama models must be referenced with their full name including version tags:

• ✅ Correct: mistral:7b, llama3.2:latest, phi3:mini
• ❌ Wrong: mistral, llama3.2, phi3 (without tags)

The OpenRegister UI dropdown shows full model names with tags to ensure compatibility.

Recommended Models

For Chat & Agents

Model	Size	RAM Required	Speed	Quality	Use Case
Llama 3.2 (8B)	8B	16GB	⚡⚡	⭐⭐⭐⭐	General purpose, best balance
Llama 3.2 (3B)	3B	8GB	⚡⚡⚡	⭐⭐⭐	Fast, lightweight
Mistral (7B)	7B	16GB	⚡⚡⚡	⭐⭐⭐⭐	Fast and efficient
Phi-3 Mini	3.8B	8GB	⚡⚡⚡	⭐⭐⭐	Very fast, good quality
CodeLlama	7B	16GB	⚡⚡	⭐⭐⭐⭐	Code generation, technical

For Embeddings (RAG)

Model	Size	Dimensions	Speed	Quality	Use Case
nomic-embed-text	137M	768	⚡⚡⚡	⭐⭐⭐⭐	General purpose, recommended
all-minilm	22M	384	⚡⚡⚡⚡	⭐⭐⭐	Fast, smaller vectors

Use Cases

1. AI Chat

Enable conversational AI in OpenRegister:

// Configured via Settings → LLM Configuration
// Select Ollama as provider
// Choose chat model (e.g., llama3.2:latest)

2. RAG (Retrieval Augmented Generation)

Answer questions using your data:

1. Configure embedding model: nomic-embed-text:latest
2. Vectorize your objects and files
3. Ask questions - AI retrieves relevant context and answers

3. Function Calling

Use Ollama with OpenRegister's function calling capabilities:

• Search objects
• Create objects
• Update objects
• Query registers

Troubleshooting

Ollama Container Won't Start

# Check logs
docker logs openregister-ollama

# Common issues:
# 1. Port 11434 already in use
sudo lsof -i :11434

# 2. Insufficient memory
docker stats openregister-ollama

# 3. GPU not available
docker exec openregister-ollama nvidia-smi

Model Not Found

# List available models
docker exec openregister-ollama ollama list

# Pull missing model
docker exec openregister-ollama ollama pull llama3.2:latest

# Verify model name includes tag
docker exec openregister-ollama ollama show llama3.2:latest

Connection Errors from OpenRegister

Problem: OpenRegister can't connect to Ollama.

Solutions:

1. Verify Ollama URL uses container name: http://openregister-ollama:11434
2. Check containers are on same Docker network: docker network ls
3. Test connection from Nextcloud container:

   docker exec <nextcloud-container> curl http://openregister-ollama:11434/api/tags

Slow Performance

Solutions:

1. Use GPU acceleration (10-100x faster)
2. Choose smaller model (3B instead of 8B)
3. Increase OLLAMA_KEEP_ALIVE to keep models loaded
4. Use quantization (smaller model variants)

Performance Optimization

GPU Acceleration

For best performance, use GPU:

deploy:
  resources:
    devices:
      - driver: nvidia
        count: all
        capabilities: [gpu]

Performance Gain: 10-100x faster inference with GPU

Model Loading

Keep models loaded in memory:

environment:
  - OLLAMA_KEEP_ALIVE=30m  # Keep models loaded for 30 minutes

Parallel Requests

Handle multiple requests:

environment:
  - OLLAMA_NUM_PARALLEL=4  # Process 4 requests simultaneously

API Usage

Direct API Calls

Test Ollama directly:

# List models
curl http://localhost:11434/api/tags

# Chat completion
curl http://localhost:11434/api/chat -d '{
  "model": "llama3.2:latest",
  "messages": [
    {"role": "user", "content": "Hello!"}
  ]
}'

# Generate embeddings
curl http://localhost:11434/api/embeddings -d '{
  "model": "nomic-embed-text:latest",
  "prompt": "Your text here"
}'

Further Reading

• Ollama Official Documentation
• LLPhant Ollama Integration
• RAG Implementation
• AI Chat Features

Support

For issues specific to:

• Ollama setup: Check Ollama Documentation
• OpenRegister integration: OpenRegister GitHub issues
• Model selection: See recommended models above