Hi,
We have 4 DCs in DFSR mesh, and every midnight the backlog starts growing from zero to several thousands, over a period of about 3-5 hours, and then it starts clearing, eventually hitting zero by 9-10 AM. This repeats every night, and it has to be some internal DFSR process as there is nothing happening on the servers at that time. All servers experience the backlog, in all directions (to all other servers). Not all replication groups are involved, one or two random groups per incident.
I also have no idea how is this backlog suddenly created - if we assume everything is replicated ok, where is this coming from? There can't possibly be 600K changes every night, without us doing anything to cause changes to DFS folders. We have 19 replication groups and 7 DFS namespaces.
Thanks