Fallback Model

[filipeaaoliveira]

Hello,

First of all, thank you so much for taking the time to do this, it's wonderful.

I was trying on my own, and everything did work, but I had to connect the Fallback model to the Ollama Chat Model just to test it.
I understand the goal of it in case this model fails, but found no mention of this in the documentation. It's pretty straightforward, but should it be added?
I can take care of that if needed. Didn't want to open an issue as it may be like this by design.

Thank you!

[jeanpaul]

Hi @filipeaaoliveira Can you try and upgrading your n8n dockers? There was a bug a while back that required agents to connect a fall-back model. However, that has been changed so that it's only optional (and you have to toggle an option in the agent to do so). Please let us know if upgrading n8n doesn't work.

[filipeaaoliveira]

Hello, good morning, @jeanpaul

Thank you for answering this!

I just cloned the repository yesterday and I'm running the latest version, created 6 days ago. I deleted my images and pulled again, and got the same result.

[jeanpaul]

That's strange. I also just pulled the images and I didn't have that issue with fall-back option. I see the following:

A few things to check:

• If you insert a new agent node, do you have the same?
• Does this happen with the workflow that's included in the template?
• What's the n8n version that's reported in "Help > About n8n"?

Thank you for reporting and helping me debug this!

[jeanpaul]

Okay, I looked into it some more, and I see that indeed there's an issue with the older versions of the Basic LLM Chain. I updated the demo workflow, but that might not help with your current workflows that are already imported. You can remove the Basic LLM Chain from canvas, and insert the newest version, that should work properly. Thanks again for looking into this!