If you recently recovered a primary database and replication is failing, one of these problems may have occurred:
A new generation ID for the primary database has not been set. See Chapter 19, “Replication System Recovery,” in the Replication Server Administration Guide.
If you loaded the primary and replicate databases from the same backup, make sure that the rs_lastcommit table for the replicate database has the correct entries. If the origin_qid in the rs_lastcommit table is incorrect (old or changed), the DSI may ignore certain transactions associated with that origin_qid. You should bcp out the rs_lastcommit table, load the data, and then bcp in the rs_lastcommit table. If DSI still ignores certain transactions, update the origin_qid value to “0” to force DSI to accept these transactions.
If you used dump and load to synchronize the primary and replicate databases because of a failure, be sure to increase the generation number in the primary database. If you do not set the generation number correctly, replication from this database may stop because Replication Server perceives that it has already processed the messages.
Also see Chapter 19, “Replication System Recovery,” in the Replication Server Administration Guide, which contains information for recovering primary databases.
