I have a Peer to Peer replication topology set up between two SQL Server 2008 R2 servers. I was doing some testing on fault tolerance and conflict handling. I have conflict detection enabled, and continue relpication after conflict detected enabled. I took one of the servers off the network, and updated the same row on both servers.
For simplicity, let's say i updated columns A and B on the first node, and A, B, and C on the second node. On the first node, column C was null before and after the updates. On the second node, column C was null before the update, but not null after.
I reconnected the network, and replication picked back up. The changes from the first node (higher peer originator id) were made at the second node. However, the changes to column C were still present at the second node, but NOT at the first node.
Is this the expected behavior? I am thinking the transaction at the first node to update A and B overrode the transaction from the second node, and therefore column C was touched at node 2, and still null at node 1. Is there a setting that I a missing that will allow column C to be set to null at node 2?