I dosabled today the automatic resubmission of failed requests. I need to implement something more sophisticated.
Right now, it is pretty crude and basically resubmits the whole request. Any already taken data is ignored, so potentially you might end up with way more data than originally requested.