Tell me, I noticed such a problem in samba when I perform a join to windows ad and allowed the join to take about 20 minutes, during this time a number of works occur on the windows side, for example, the user's attributes change, etc., that is, active changes, then after starting the samba service, when we have already joined as DC, we begin to have the following problem in the log ./source4/rpc_server/drsuapi/getncchanges.c: DsGetNCChanges 2nd replication on DN DC=my,DC=test,DC=br older highwatermark and I see how the rpc(0) process unloads the core by 100% and this continues constantly, as I understand it, there is a comparison of hwm for objects, but the question is why is this comparison and a complete reset of the replication state necessary if at least for one object they do not match, isn't this redundant?