I understand that T-Rep includes a publisher, distributor and subscriber. Usually I find replication failing due to a Primary Key conflict or a row not found issue. I am planning to leverage the distribution databases and the publisher database to find the exact error of failure and then automatically fix the error by a SQL job (running every 15 mins or so)
Incase of a PK Issue the conflicting record be deleted and the new record be inserted and in case of a row not found issue the missing row be inserted from the publisher.
How feasible this solution would be? What tables in the distribution DB will be helpful with this autofix job?