MyDoggieJessie (7/8/2013)
I am wondering even stopping the distribution job some batch of transactions was still in progress can cause this or not .
This was my original thought. Perhaps we didn't wait long enough for everything to trickle through...
We can't create a new publication because of the snapshot issue. It would involve an entire snap of all the data and place us in a rather long maintenance window we can't really get approval for right now.
The only think I can think of is:
1. Redo it again, waiting much longer to ensure there are no pending rpl_commands, checking and then rechecking the distribution database to make 100% certain everything is replicated before taking the FULL backup of the subscription databases
2. If there are lingering commands, we need to find a way to push those to the new subscriber as well before turning the distribution agent back on
3. Dropping the troublesome article(s) from the existing publication, then re-adding them back in to ensure we minimize the impact on our OLTP system
Perfect. Please share us the results of your re work. I can't think of any reason why you miss of records. And please turn on the server side trace when you enable the replication to make sure you capture the all the events.
-- Babu