Hello everyone! I am super excited to share a challenge we had in Japan! We have an Architect flow for our 800 agents that uses a Data Action to look up the customer’s “VIP Status” in our local CRM. If they are a VIP, we route them to a special queue. However, last week our CRM server crashed, and the Data Action started timing out after 15 seconds. Instead of taking the “Failure” path in the Architect flow and routing them to the standard queue, the flow seemed to freeze, and many customers disconnected. How do we properly handle a hard timeout on a Data Action so the flow immediately moves to a safe fallback queue without leaving the caller in silence?
This drives me crazy. I maintain 50 flows and this is a huge design flaw in Architect. The “Failure” path on a Data Action block in Architect only triggers if the API actually returns a well-formed error response (like a 404 or 500 HTTP code). If the connection completely hangs and the action hits the 15-second hard timeout, Architect sometimes throws a “System Error” instead of taking the normal failure path! You must wrap your Data Action inside a dedicated “Task” block, and configure the overall “Error Handling” for that entire Task to jump to a “Disconnect” or “Transfer” action.
I verify The point above is correct. During my load tests, if I intentionally freeze the backend database, the Data Actions do not fail gracefully. They hit the timeout limit and the Architect engine terminates the flow execution entirely, resulting in a dropped call. You must configure the global Error Event handling at the Task or Flow level to catch these hard timeout exceptions.