Really like your OON architecture. One suggestion with OpenFlow, though. I would suggest that you change your docker-compose.yml (and others) to include installation of NodeRED. I’ve changed mine so that it installs but it’s just a suggestion.
Second suggestion: add a LLM model capability. Many of the processes that I’m working with now are calling for a LLM capability (I’m playing with AutoGPT and SuperAPI but have used gpt4all and Gorilla since there’s a need for API connectivity).
Love to chat sometime just to pick your brain on a few things …
If you are admin or customer admin, you can check “auto start” on any agent, to make openflow ensure it’s running ( doing startup and once an hour ) and since NodeRED is just an agent, that is also covered there.
you can run python agent’s. And most LLM frameworks are based of python. I use langchain in many project to interface to either openai or open source models from agents, but you can install any AI framework you want. If you need cuda access you can always install the nodeagent on a server/client with access to gpu’s
Just an aside, but my suggestion regarding NodeRED was about the INSTALLATION of NodeRED, not the startup.Your docker-compose yaml file has MongoDB, RabbitMQ, Traefik, and OpenFlow installing but not NodeRED. Unless the intent was to just focus on the builds associated with just OpenFlow.
you start/install nodered from within openflow.
In the early versions, if you wanted a nodered instance to start after a reboot, you would need to put it inside the docker compose file, but that posed a security risk.
Now openflow handles it, by you checking the “auto start” checkmark (or if using resource manager, assigning a plan )
Okay, I must be going blind. I don’t see any location in OpenFlow to start/install NodeRED. Seriously. Here’s a screen shot of the default installation - could you point out where the start/install area is?
