lfnovo/open-notebook · error · HTTPException

Ask operation failed: {str(e)}

Error message

Ask operation failed: {str(e)}

What it means

Catch-all 500 from the streaming ask endpoint for exceptions outside HTTPException/OpenNotebookError — e.g. LangGraph pipeline crashes, provider API failures, or streaming setup errors. Logged as 'Error in ask endpoint'.

Source

Thrown at api/routers/search.py:174

        return StreamingResponse(
            stream_ask_response(
                ask_request.question, strategy_model, answer_model, final_answer_model
            ),
            media_type="text/event-stream",
            headers={
                "Cache-Control": "no-cache",
                "Connection": "keep-alive",
                "X-Accel-Buffering": "no",
            },
        )

    except HTTPException:
        raise
    except OpenNotebookError:
        raise
    except Exception as e:
        logger.error(f"Error in ask endpoint: {str(e)}")
        raise HTTPException(status_code=500, detail=f"Ask operation failed: {str(e)}")


@router.post("/search/ask/simple", response_model=AskResponse)
async def ask_knowledge_base_simple(ask_request: AskRequest):
    """Ask the knowledge base a question and return a simple response (non-streaming)."""
    try:
        # Validate models exist
        strategy_model = await Model.get(ask_request.strategy_model)
        answer_model = await Model.get(ask_request.answer_model)
        final_answer_model = await Model.get(ask_request.final_answer_model)

        if not strategy_model:
            raise HTTPException(
                status_code=400,
                detail=f"Strategy model {ask_request.strategy_model} not found",
            )
        if not answer_model:
            raise HTTPException(

View on GitHub (pinned to a7de90d38a)

Solutions

1. Check logs for 'Error in ask endpoint' to get the real exception
2. Verify the LLM provider API key and connectivity for all three selected models
3. Test each model individually via a chat completion call
4. Update to matching versions if recently upgraded

for (const id of [strategy, answer, final]) {
  if (!models.some(m => m.id === id)) throw new Error(`Model ${id} missing`);
}

catch (e) {
  if (e.status === 500 && /Ask operation failed/.test(e.detail)) showProviderDiagnostic(e.detail);
}

When it happens

Trigger: POST /api/search/ask when graph invocation or StreamingResponse setup raises an unexpected exception: bad provider credentials at call time, LLM provider outage, malformed graph state.

Common situations: Expired/invalid LLM API key, provider rate limits surfacing as raw exceptions, version drift between graph definitions and router code.

Related errors

• No answer generated
• Error executing transformation: {str(e)}
• Error fetching chat sessions: {str(e)}
• Error creating chat session: {str(e)}
• Error fetching session: {str(e)}

AI-assisted analysis of lfnovo/open-notebook@a7de90d38a (2026-08-27). Data as JSON: /api/errors/9b8a4dc0b9f61648. Report an issue: GitHub.