Currently, the documentation requires transformers to simply ack and logs events they are unable to process.
This is probably not a good idea because:
Aside from the logs there is no indication that the transformer isn't working correctly
Once the issue is solved, we'd have to manually recreate the events that failed from log messages
We should discuss introducing deadletter mechanisms for failed transformer events.
While we're at it, introducing a standardized requeue limit across transformers could be a good idea
Currently, the documentation requires transformers to simply ack and logs events they are unable to process.
This is probably not a good idea because:
We should discuss introducing deadletter mechanisms for failed transformer events. While we're at it, introducing a standardized requeue limit across transformers could be a good idea