We use transactional replication to replicate data from table1 on server1 to table2 on server2. However, one of my colleagues accidentally inserted some data into table2. In order to keep the two tables synced up, should we delete the rows inserted? If so, do we need to reset identity?
Delete the rows to not cause any conflicts later. If you still need the data insert them on the publisher after you deleted them on the subscriber.
You do not need to worry about the identity seed. Replication inserts its own values and does not rely on the identity-auto-value-generation. As this is a table on the subscriber you should not insert data ever again, so to you it should not matter what the identity seed is either.