Support Multiple Ollama Instances

[igordayen]

Multi-Ollama Instance Support

Please refer to issue: #886

Summary

Enables applications to configure and use multiple Ollama instances with explicit node-aware
model selection while maintaining full backward compatibility.

Configuration Options

Legacy Single Instance (Unchanged)

# Existing configuration continues to work unchanged
spring.ai.ollama.base-url=http://localhost:11434

Multi-Node Configuration

# Configure multiple Ollama instances
spring.ai.ollama.nodes[0].name=main
spring.ai.ollama.nodes[0].base-url=http://localhost:11434
spring.ai.ollama.nodes[1].name=gpu-server
spring.ai.ollama.nodes[1].base-url=http://localhost:11435

Hybrid Configuration

# Use both default instance and additional nodes
spring.ai.ollama.base-url=http://localhost:11434
spring.ai.ollama.nodes[0].name=main
spring.ai.ollama.nodes[0].base-url=http://localhost:11434
spring.ai.ollama.nodes[1].name=gpu-server  
spring.ai.ollama.nodes[1].base-url=http://localhost:11435

Application Model Selection

Backward Compatible (Legacy)

# Uses any available instance (default behavior)
embabel.models.default-llm=gemma3:latest

Node-Specific Selection

# Target specific Ollama instances by node name
embabel.models.default-llm=main-qwen3:latest
embabel.models.llms.cheapest=main-qwen3:latest  
embabel.models.llms.best=gpu-server-llama3:latest

Multi-LLM Agent Configuration

examples:
  book-writer:
    researcher-llm:
      model: main-gemma3:latest        # Fast model on CPU node
    writer-llm:
      model: gpu-server-llama3:latest  # Powerful model on GPU node

[igordayen]

addition of Unit tests for Olama Configuration:

[sonarqubecloud]

Quality Gate passed

Issues
2 New issues
0 Accepted issues

Measures
0 Security Hotspots
86.6% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud