Replication and fail over questions

I’m not sure if this is more of an admin question or design question. I’m a designer and don’t really do anything on the Admin side. Basically we have a cluster with two servers Mail1 and Mail2. So occasionally mail1 will go down and users fail over to mail2. When Mail1 is down and a user tries to open a database from their favorites it will find the database on mail2 and open it.

The problem is when Mail1 comes back online, their mail database reverts back to mail1 but the favorite bookmark will continue to open up the database on mail2. It’s like the bookmark itself changes. Is there a way to make it not fail over for a specific database? Or would the only way to prevent this be to not replicate the database to mail2? The problem is that our database generates unique numbers and when someone starts using it on mail2 we end up getting duplicates. Thanks for any info.

Subject: Replication and fail over questions

We’re moving to a clustered environment later this year so I’m no expert but my understanding is that clusters use load balancing to decide which replica to open so you can’t force a particular replica to open. However, replication within a cluster is almost instantaneous so there should be few duplicates unless the database is very heavily used. At what point do you generate the unique number, opening a document, saving it or some other place?