Trying holmesgpt with local ollama3.1 fails w. KeyError: 'name'

[suukit]

Hi,
I wanted to give this a try and installed ollama locally. I am able to use the ollama API on http://localhost:11434/api/generate with curl.
I evaluated export OLLAMA_API_BASE=http://localhost:11434, installed holmes using brew as described in the readme and startet
holmes ask --model ollama/llama3.1:latest "what issues do I have in my cluster" -v
resulting in

APIConnectionError: litellm.APIConnectionError: 'name'
Traceback (most recent call last):
File "litellm/main.py", line 2422, in completion
File "litellm/llms/ollama.py", line 293, in get_ollama_response
KeyError: 'name'

Any help how to debug this? Already tryied -v which gives no further helpful output. Changed the OLLAMA_API_BASE to http://localhost:11434/ZZ and then receive and 404 as expected, so I assume OLLAMA_API_BASE is working. Also switched to ollama/llama2 model with same result.

Thank you in advance
Max

[juliendf]

Also have the same error. Did you find a solution or workaround ?

Thanks

[arikalon1]

Hey @juliendf

I think it's this issue on litellm
BerriAI/litellm#3912 (reference)
ollama/ollama#4398

Which version of Holmes are you using?