All,
Looking for some clarification regarding DFS functionality. I'm currently in the process of disaster recovery on a DFS system I inherited. It was never set up correctly to begin with (DFS was configured on an existing data share, then another
server added). Replication was never truly working, in that data placed on Server_A never actually got transferred to Server_B. The admin who I inherited this from used a combination of robocopy scripts and forced namespace usage to ensure that
people actually got served the data they needed... basically a mess.
Anyhow, I did some digging, found that replication was never started, and initiated a replication through the following command:
Dfsradmin Membership Set /RGName:RG1 /RFName:DATA /MemName:DFSMEMBER /IsPrimary:True
Much to my surprise, it immediately started staging and 'replicating' data. Server_b stated there were gobs of pre-existing data (expected, a ton of existing/expired robocopied crap there) which it moved to pre-existing, promptly overflowed, and got
rid of (also expected, there's about 8TB I'm working with). As time went on, I got constant event log hits on Server_A stating there was a file conflict, used algorithm to determine 'correct' one, as expected... sort of. See, I expected the algorithm
to determine that the primary server (Server_A) was in fact the holder of the primary copy, and just copy that file to the secondary server. Much to my surprise when I came in this morning with a DFS drive half empty, and about 60% of my files were just
*gone* from both Server_A and Server_B. So, now I'm doing a restoral of the everything.
So I guess I'm looking for clarification on if it's even possible to take Server_A with pre-existing data, set up a replication group, with a second member, and have it just send the data over? If Server_B is a mirror of Server_A, and you set as primary,
is it supposed to delete everything from both systems? That seems exceptionally counter-intuitive.
Thanks!