I was looking at the sample code here: https://docs.aws.amazon.com/qldb/latest/developerguide/getting-started.python.step-5.html, and noticed that the three functions get_document_id_by_gov_id, is_secondary_owner_for_vehicle and add_secondary_owner_for_vin are executed separately in three driver.execute_lambda calls. So if there are two concurrent requests that are trying to add a secondary owner, would this trigger a serialization conflict for one of the requests?
The reason I'm asking is that I initially thought we would have to run all three functions within the same execute_lambda call in order for the serialization conflict to happen since each execute_lambda call uses one session, which in turn uses one transaction. If they are run in three execute_lambda calls, then they would be spread out into three transactions, and QLDB wouldn't be able to detect a conflict. But it seems like my assumption is not true, and the only benefit of batching up the function calls would just be better performance?
Got an answer from a QLDB specialist so going to answer my own question: the operations should have been wrapped in a single transaction so my original assumption was actually true. They are going to update the code sample to reflect this.