I have two ACS servers 1113 that will replicate, then FAIL replication later with shared secret mismatch. I tracked it down to the self profiles on each server using the loopback IP instead of the ethernet IP. Once that happens, the shared secret reverts to secret_value, the default secret, and cannot be changed due to the loopback address. If I try to add a new host using the IP, I get fail because host already exists. I set the remote servers secret to match the default, and was able to get a replication to succeed, but now I get server not responding, and when I looked at the slave, it no longer has the master listed, but a second self with its address, and I am unable to add the master back in.
For replication I disabled distribution table, as I think that caused the first problem, but now I am not so sure. It is driving me nuts, first it works, then it doesn't, then I get it working again, then it FAILs again.