Hello,
This is being solved right now. We are trying to keep the error rates to zero and are implementing internal retry mechanism to avoid the users to do so.

Thanks for your comprehension.