Skip to content

[Feature]: AI feature : Remove nomad_ollama as a requirement for AI Assistant install. #662

Description

@Draco-Lunaris

Feature Category

AI Assistant Enhancement

Problem Statement

If you have a complex GPU config or require customization requirements on the host that Nomad is to be installed on, the AI assistant install requires the ollama docker container to be installed. This conflicts with other LLM hosting software running on the host including customized ollama configs, vLLM, llama.cpp, etc.

The AI assistant install will fail if there is anything already running on the default ollama port.
The AI assistant will have random issues and will cause random issues for other LLM services by trying to utilize the same GPU's at the same time.

Proposed Solution

Possible, solution. Remove the ollama requirement for the AI Assistant install. If the ollama service is optional, then it won't be present to conflict for ports or GPU resources.

Alternative Solutions

Run the nomad_ollama container as default CPU only, and on a non-standard port to avoid conflicting with services that may already be in place.

It is possible to run Nomad on a totally different host just fine. IF you have additional hardware, and are willing to was the resources for it.

Use Case

This would allow more complex LLM configuration and deployments on the same host nomad is installed on. Allowing for more functionality outside of nomad on the same hardware.

Who would benefit from this feature?

All Users

How important is this feature to you?

High - Would significantly improve my experience

Implementation Ideas (Optional)

No response

Examples or References

Example, I utilize a single host machine to run Openclaw, Agent Zero, Nomad, vLLM, and ollama. All my GPU's and storage/memory are all on a single platform and run inside docker containers.

This allows me to fully utilize the hardware I have. Additional hardware would require more cost, more electricity, more cooling, etc.

Would you be willing to help implement this?

Yes - I can write the code

Additional Context

I am willing to help with a solution. I'm fully capable of coding and solution engineering. My GitHub experience is limited however. I have room to learn how to properly work with it to contribute correctly.

Pre-submission Checklist

  • I have searched for existing feature requests that might be similar
  • This feature aligns with N.O.M.A.D.'s mission of offline-first knowledge and education
  • I understand that feature requests are not guaranteed to be implemented

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions