Hi Fausto,
Great question.
From my experience, I tend to handle errors close to where they occur rather than using one centralized error task.
For Data Actions, I use the Success, Failure, and Timeout paths. On the Success path, I still validate that the expected values have been returned before continuing. If a lookup fails or times out, I follow an appropriate fallback path so the customer journey can continue where possible.
In our implementations, we also use variables to validate the flow state before allowing the next step (for example, checking that a lookup succeeded before enabling the next action).
I'd be interested to hear how others structure error handling in larger Architect solutions.
------------------------------
Phaneendra
Technical Solutions Consultant
------------------------------