rootdn and updatedn really really really need to be different. We've
had this discussion about 18 months ago (maybe more). I believe it is
even spelled out in the admin guide.
So, I believe the slave is getting confused and not knowing if you are
trying to do an replication-update or if the rootdn has just connected
to it to do a force update (via ldapmodify say) and is therefore
assuming the rootdn and rejecting the update. I expect you will solve
your problem if you change the updatedn to be different from the rootdn.