ScrapeGraphAI/Scrapegraph-ai · error · Timeout

Response took longer than {timeout} seconds

Error message

Response took longer than {timeout} seconds

What it means

GenerateAnswerNode.invoke_with_timeout calls chain.invoke() and afterwards compares elapsed wall time against the timeout; if the call took longer than the limit, langchain_core.errors.Timeout is raised (and logged) so the graph can branch on RetryError/timeout handling.

Source

Thrown at scrapegraphai/nodes/generate_answer_node.py:83

            if node_config.get("schema", None) is None:
                self.llm_model.format = "json"
            else:
                self.llm_model.format = self.node_config["schema"].model_json_schema()

        self.verbose = node_config.get("verbose", False)
        self.force = node_config.get("force", False)
        self.script_creator = node_config.get("script_creator", False)
        self.is_md_scraper = node_config.get("is_md_scraper", False)
        self.additional_info = node_config.get("additional_info")
        self.timeout = node_config.get("timeout", 480)

    def invoke_with_timeout(self, chain, inputs, timeout):
        """Helper method to invoke chain with timeout"""
        try:
            start_time = time.time()
            response = chain.invoke(inputs)
            if time.time() - start_time > timeout:
                raise Timeout(f"Response took longer than {timeout} seconds")
            return response
        except Timeout as e:
            self.logger.error(f"Timeout error: {str(e)}")
            raise
        except Exception as e:
            self.logger.error(f"Error during chain execution: {str(e)}")
            raise

    def process(self, state: dict) -> dict:
        """Process the input state and generate an answer."""
        user_prompt = state.get("user_prompt")
        # Check for content in different possible state keys
        content = (
            state.get("relevant_chunks")
            or state.get("parsed_doc")
            or state.get("doc")
            or state.get("content")
        )

View on GitHub (pinned to 532dfffbf6)

Solutions

1. Increase the timeout in graph config: {'llm': {...}, 'timeout': 300} or the node/model timeout setting
2. Reduce input size (truncate docs) or lower max_output_tokens so responses finish faster
3. Implement the node's retry/error-handling path or catch Timeout upstream and retry with backoff

Example fix

# before
graph_config = {'llm': {'model': 'openai/gpt-4o'}, 'timeout': 60}
# after
graph_config = {'llm': {'model': 'openai/gpt-4o'}, 'timeout': 300}

# pre-flight: choose timeout proportional to expected output size
est_tokens = len(docs_text) // 4
timeout = max(120, est_tokens // 100)  # rough heuristic

from langchain_core.errors import Timeout
for attempt in range(2):
    try:
        result = node.invoke_with_timeout(chain, inputs, timeout)
        break
    except Timeout:
        if attempt == 1:
            raise
        timeout *= 2

When it happens

Trigger: A slow LLM (long generation, rate-limit retries inside the client, overloaded provider) making chain.invoke exceed the configured timeout (e.g. 90s default) set via config 'timeout'.

Common situations: Large contexts/long outputs with small timeouts; provider throttling; intermittent network latency; timeout too aggressive for reasoning models.

Understand the failure class

• Timeouts: ETIMEDOUT, deadlines, and hung requests — what actually expires when a request times out.

Related errors

• Could not determine model name from llm_model. Please specif
• PDF parsing exceeded timeout of {self.timeout} seconds
• Model not supported
• model_tokens not specified
• Provider {llm_params["model_provider"]} is not supported.

AI-assisted analysis of ScrapeGraphAI/Scrapegraph-ai@532dfffbf6 (2026-08-28). Data as JSON: /api/errors/37231215ca251f0d. Report an issue: GitHub.