It was possible that a one-second outage that caused immediate rejection
of network connections would cause all of the query retry attempts to fail
within a very short period of time. By preventing rapid reconnections,
query_retries is more effective as an error filtering mechanism.