Hi everyone,
We have two servers that have SQL SERVER 2014 and there is Peer to Peer replication set up between them. There are 15 publications on each server and all of them on each server are using the same server as local distributor. In one of these publications we have one large article (more than 2 billion records ) . On all publications, 'Allow peer to peer conflict detection' and also 'Continue replication after conflictid detection' are set to true .One of developers deploy and run a job that massively start to delete rows from that large article independently on both servers. This job rans for 8-9 hours and created a massive backlog in transaction log(around 100GB) This caused all publication on each server show a very high latency and large back log in transaction log. The Error log is filled with lots of below error on both servers.
A conflict of type 'Delete-Delete' was detected at peer 81 between peer 181 (incoming), transaction ...).
After two days the transaction log(on both server ) become empty and all other publications caught up, but still that particular publication that delete operation has happened, shows there is latency for 2 days and also still the error is being flooded by 'A conflict of type 'Delete-Delete' was detected at peer 81 between peer 181 (incoming), transaction ...).' error message.
I was thinking after trying to run delete statement on each server , if there is a conflict sql server will not try to run it again. So I am not sure why still the errorlog is being filled with above error message. Will restarting Distribution agent help?
How can I eliminate that latency on that particular publication? Is there any way to fix the issue without resetting up the peer to peer?
A.G