Admin4 is hugh, it's killing you, your servers, and network bandwidth during replication to other domain servers. It's loaded with tons of data generated daily at breakneck rates, especially if your environment is medium->large. Frequently a request becomes corrupt causing problems with replication, etc. As Regis would say, "It's out of control !!!"
If you have a hub-and-spoke replication topology, then you can use this selective replication formula on admin4.nsf replicas (use only in replication "spoke" server replicas; not in replication "hub" nor in "AdminP" server replicas) ...
This will only allow data into the replica that is pertinent to that server's adminp activities putting a dramatic end to it's prior days of diskspace and network bandwidth gobbling activities allowing you to spend more time on baby-sitting other Notes server problems.
NOTE: After a new server setup or recreation of admin4.nsf one needs to remember to do the above as this setting is not automatically replicated from another replica nor inherited from admin4.ntf.