[Install]: Failed to send message

[d-shehu]

Installation Method

Docker (single container - v1-latest-single)

What is the issue?

Neat program and great to see someone building an alternative to Google's NotebookLM. Thanks.

Unfortunately, I've tried to configure it using llama.cpp using OpenAI compatible API as well as Ollama without success. In either case when trying to chat with the notebook it fails with toast notification.

 _Failed to send message_

In the logs, it says my model is "not language model". The notebook only contains a single URL to a wikipedia page which I could see was parsed by Open Notebook. For the simplest case I used Ollama models shown below for the embeddings and chat/transformation.

Confirmed my Ollama and Llama.cpp URLs are working & accessible via curl, OpenWebUI. Followed guides below. Thanks.

Open Notebook Health:

{"status":"healthy"

Ollama:

ollama list

NAME                        ID              SIZE      MODIFIED      
mxbai-embed-large:latest    468836162de7    669 MB    7 minutes ago    
gemma3:12b                  f4031aab637d    8.1 GB    2 months ago     

curl $OLLAMA_URL:11434 
Ollama is running

Guides:

1. https://github.com/lfnovo/open-notebook/blob/main/docs/features/ai-models.md
2. https://github.com/lfnovo/open-notebook/blob/main/docs/features/ai-models.md

Logs

2025-12-26 21:16:59.575 | ERROR    | api.routers.chat:execute_chat:384 - Error executing chat: Model is not a LanguageModel: None

Docker Compose Configuration

Testing with docker run before I deploy with Docker compose.


docker run --name=usr_llm_notebook \
        -p 26000:8502 \
        -p 26055:5055 \
        -e API_URL=http://${HOSTNAME}.${ENVIRONMENT}.${DOMAIN}:26055 \
        -e OPENAI_COMPATIBLE_BASE_URL=https://${LLAMACPP_URL}/v1 \
        -e OLLAMA_API_BASE=http://${OLLAMA_URL}:11434 \
        -e SURREAL_URL=ws://localhost:8000/rpc \
        -e SURREAL_USER=root \
        -e SURREAL_PASSWORD=root \
        -e SURREAL_NAMESPACE=open_notebook \
        -e SURREAL_DATABASE=production \
        -v ./app:/app/data \
        -v ./user:/mydata \
        lfnovo/open_notebook:v1-latest-single

Environment File

N/A

System Information

Ubuntu 24.04
Arch: x86 64

Additional Context

No response

Pre-submission Checklist

• I tried the Installation Assistant ChatGPT
• I read the relevant documentation (Installation Guide or Ollama Guide)
• I searched existing issues to see if this was already reported
• I redacted all sensitive information (API keys, passwords, etc.)

[d-shehu]

What's interesting is with the same configuration if I use "Ask and Search" using "Ask" mode, it does trigger a request to Ollama and also the OpenAI API (Llama.cpp). The outcomes are as follows:

Ollama:
After churning through Ollama for a couple of minutes, "Ask and Search" stays stuck on "Processing your question". But from the output of the open notebook container it clears seems idle.

OpenAI API (Llama.cpp):
Received request and formulated a strategy but then stopped there. I used the same model for all 3 (strategy, answer and final answer).

Btw, I've been using Docker hub to fetch the images. I tried all 4 versions which don't exactly match with Github but seem to be 1.2.x. Version 1.2.1 also triggered an exception though I could see it connecting to Llama.cpp. The other 3 ran without exception but exhibited the faulty behavior I described.

Docker hub:

https://hub.docker.com/r/lfnovo/open_notebook

Github:

https://github.com/lfnovo/open-notebook/releases

v1.2.1 exception:

  File "/app/.venv/lib/python3.12/site-packages/httpcore/_backends/anyio.py", line 113, in connect_tcp                                                                                                                                                                         
    with map_exceptions(exc_map):                                                                                                                                                                                                                                              
         │              └ {<class 'TimeoutError'>: <class 'httpcore.ConnectTimeout'>, <class 'OSError'>: <class 'httpcore.ConnectError'>, <class 'anyio...                                                                                                                     
         └ <function map_exceptions at 0x7fa2a1fce980>                                                                                                                                                                                                                         
  File "/usr/local/lib/python3.12/contextlib.py", line 158, in __exit__                                                                                                                                                                                                        
    self.gen.throw(value)                                                                                                                                                                                                                                                      
    │    │   │     └ OSError('All connection attempts failed')                                                                                                                                                                                                                 
    │    │   └ <method 'throw' of 'generator' objects>                                                                                                                                                                                                                         
    │    └ <generator object map_exceptions at 0x7fa2a033ba60>                                                                                                                                                                                                                 
    └ <contextlib._GeneratorContextManager object at 0x7fa2a03224b0>                                                                                                                                                                                                           
  File "/app/.venv/lib/python3.12/site-packages/httpcore/_exceptions.py", line 14, in map_exceptions
    raise to_exc(exc) from exc
          └ <class 'httpcore.ConnectError'>

httpcore.ConnectError: All connection attempts failed

[jgoldfed]

I ran into something similar. I was running Ollama directly on my host machine, running Open-Notebook in Docker. Ensure your OLLAMA_API_BASE=http://host.docker.internal:11434, since that's the way an application in Docker can communicate with localhost. It fixed my issue.

[d-shehu]

Thanks for the suggestion but I'd like to run the notebook and ollama, llama.cpp on different machines for performance reasons.

The "Failed to send message" triggering every time and Open Notebook not connecting to either Ollama or Llama.cpp seems like a clear bug. Sadly, the Dock log doesn't output any useful info with 1.2.2 - 1.2.4.

[ita333]

Identical problem , same message "Failed to send message".
I tried Gemini2 and 3 and Ollama->Gemma3-4b .
Embeddeing is processing ok.
Chat Failed to send message.
Version Open Notebook 1.2.4

[bernhardcygan]

Same symptom here.

Ollama as a service on local machine
Using podman with lfnovo/open_notebook:v1-latest-single image from docker.io, version 1.2.4
Chat yields "Failed to send message", no matter which model I use
Ask yields nothing
Search finds keywords

The only meaningful information I could find in the Open Notebook log is

[open_notebook] | 2026-01-05 12:26:51.753 | ERROR    | api.routers.chat:execute_chat:384 - Error executing chat: [Errno 111] Connection refused
[open_notebook] | INFO:     10.89.0.2:50082 - "POST /api/chat/execute HTTP/1.1" 500 Internal Server Error

ollama runs as a service on the same machine and is configured with
Environment="OLLAMA_HOST=0.0.0.0:11434"

There is no difference wheter I enter OLLAMA_API_BASE as http://hostcontainers.internal:11434, http://localhost:11434, http://host.docker.internal:11434 or http://:11434 in compose.yml