Thank you very much for your post. We had similar problem in production server that was not been able to find the root cause for quite a while. We did so many tests with P2p replication. Finally I gave up. I never thought about the triggers though. Good catch.