By default, our retry policy is forever. Due to a conversion bug, the amount of retries elapse causing unexpected behaviour after network disruption.