My issue is not being unable to handle message. For one, I know what the error is, it is a disconnect + timeout, 'cos I intentionally caused it in the test.
The more severe issue is that it will take a long long time for it to even deliver you those failed_objects.
as I mentioned, I have no problem with this if the disconnect is only intermittent, and I did observed it before in a realistic run, and capable of inspecting those failed_objects, and do the appropriate retry.
I am not sure if I have made the issue clear enough.