iflytek/astron-agent · error · CustomException

KNOWLEDGE_REQUEST_ERROR

KNOWLEDGE_REQUEST_ERROR

Error message

Knowledge Pro node response status: {response.status}

What it means

After posting to the Knowledge Pro API, the node checks the HTTP status; any status other than 200 (httpx.codes.OK) raises a CustomException with KNOWLEDGE_REQUEST_ERROR whose cause includes the status code. It means the Knowledge Pro service rejected or failed the request at the HTTP level.

Solutions

  1. Check the Knowledge Pro service logs for the failure corresponding to the returned status
  2. Verify the node's Knowledge Pro API URL and credentials are correct and the service is reachable
  3. If 401, refresh/fix the auth token; if 5xx, retry after the service recovers or check gateway health
Defensive patterns

Strategy: try-catch

Validate before calling

import httpx
async def knowledge_pro_reachable(url: str, headers: dict) -> bool:
    try:
        r = await httpx.AsyncClient().get(url, headers=headers, timeout=5)
        return r.status_code < 500
    except httpx.HTTPError:
        return False

Try / catch

try:
    result = await node.async_execute(variable_pool, span)
except CustomException as e:
    if e.err_code == CodeEnum.KNOWLEDGE_REQUEST_ERROR:
        log.error("Knowledge Pro HTTP failure: %s", e.cause_error)  # contains status
        # retry with backoff for 5xx, alert for 4xx

Prevention

When it happens

Trigger: The async aiohttp/httpx POST to the Knowledge Pro retrieval API returns e.g. 401, 404, 500, or 502 instead of 200; auth token invalid, URL wrong, or the knowledge service crashing.

Common situations: Knowledge Pro service is down or restarting in the cluster; API base URL configured incorrectly in node settings; expired credentials rejected with 401; gateway returning 502/504 under load.

Understand the failure class

Background: "API error: {status}" and "HTTP 401/403/404/429/5xx" errors: non-2xx HTTP responses explained — this error's family across 27 libraries.

Related errors

AI-assisted analysis of iflytek/astron-agent@5e758547a8 (2026-09-12). Data as JSON: /api/errors/1968077c2dc49f27. Report an issue: GitHub.

Appendix: source

Thrown at core/workflow/engine/nodes/knowledge_pro/knowledge_pro_node.py:148

            # Get the Knowledge Pro API endpoint URL from environment or use default
            url = f"{os.getenv('KNOWLEDGE_PRO_BASE_URL')}/knowledge/v1/agent/achat"

            # Validate CBG RAG parameters before proceeding
            await self._check_cbg_rag_param()

            # Generate request payload for the Knowledge Pro API
            payload = self.gen_req_payload(query, span)
            await span.add_info_event_async(f"request body: {payload}")
            # Create HTTP session with appropriate timeout configuration
            async with aiohttp.ClientSession(
                timeout=ClientTimeout(
                    total=30 * 60, sock_connect=30, sock_read=interval_timeout
                )
            ) as session:
                # Send POST request to Knowledge Pro API
                async with session.post(url=url, json=payload) as response:
                    if response.status != httpx.codes.OK:
                        raise CustomException(
                            err_code=CodeEnum.KNOWLEDGE_REQUEST_ERROR,
                            cause_error=f"Knowledge Pro node response status: {response.status}",
                        )

                    content_list, knowledge_metadata, token_usage = (
                        await self._handle_response(
                            response, span, variable_pool, msg_or_end_node_deps
                        )
                    )

            # Prepare final outputs with combined content and metadata
            outputs = {"output": "".join(content_list), "result": knowledge_metadata}
        except asyncio.TimeoutError:
            # Handle timeout errors during API request
            log_err = CustomException(
                err_code=CodeEnum.KNOWLEDGE_REQUEST_ERROR,
                err_msg=f"Knowledge Pro node response timeout ({interval_timeout}s)",
            )

View on GitHub (pinned to 5e758547a8)