Join GitHub today
GitHub is home to over 40 million developers working together to host and review code, manage projects, and build software together.Sign up
Improvement of sending event to Audit Service #1407
Changing sending method as follows.
Improvement of network communication frequency
@daisuke-yoshimoto ok I want to understand more about this. I do understand that events will be sent even if the state of the process engine is rolled back in the case of an issue. So basically, you are sending all the events when the transaction is committed, am I right? In principle, I do agree with that idea, but we need to be careful on how we implement this. In my opinion, we can solve this by making all the process interactions with external entities async by default. Meaning that we will assume that BPMN events and Sequence Flows will not fail so transactions will be small and will not cause big rollbacks. In the example of (start event) -> (service task) - > (user task) -> (end event) we will force the engine to deal with the first service task in an async way always, so the first transaction will only include the start event + scheduling the service task. Which means also that you are right in bundling several events together in the same message, because even that simple example has two events happening together in the same tx.
Does this make sense?