I have a replication setup with 1 publisher and 3 push subscribers. I need to add a new push subscriber. I cannot afford downtime for the publisher and the existing subscribers.

My plan is to do the following:

  1. take a backup of the publisher. Say that this backup finished at 6 pm.
  2. restore the backup to the new subscriber server
  3. use the latest snapshot which was taken 5 days ago as the snapshot to be used to set up the new subscriber
  4. set up the new subscriber and do not use the snapshot to initialize it. This means that the tables will not be dropped, recreated, nor populated from the snapshot. The snapshot is only used for the metadata definitions contained in it.
  5. start replication to the new subscriber. Say that it was 6:10 pm by the time the new subscriber started synchronizing.

How will any differences between the publisher and the new subscriber be resolved for data that changed during the time the subscriber was restored from the backup of the publisher taken at 6 pm and the time when synchronization of the new subscriber started at 6:10 pm?

a. Will changes that occurred on the publisher between 6 pm and 6:10 pm be pushed to the new subscriber?

b. Since this is merge replication, will any changes made on the publisher between 6 pm and 6:10 pm be affected by the new subscriber itself? I want to make sure I avoid any legitimate changes that occurred on the publisher between 6 pm and 6:10 pm being changed back on the publisher due to the fact that the new subscriber is not aware of the change that occurred on the publisher.

Am I overthinking this?

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.