You should monitor the failure notifications in any case. Sometimes replication failures are only transient, because of high load or similar. A subsequent run might work again then the status will again be OK, but it's still good to know about the failure.