When there is a failure in communication between the Replication server on the primary machine and the Replication agent on the secondary machine, all database activity is failed over to the secondary machine. If failover occurs, the Replication-enabled target database is available as a hot standby. A hot standby database is a database that is updated and ready to go immediately.
You can have up to two target databases per source database. Only one of the target databases can be set to transition automatically.
You can identify only one target database as critical because you do not want users to be updating two different databases after automatic transition has occurred. If you have a second target database, that database is not transitioned unless you manually transition it. This provides you with options for your hot standby choice. You can find details about automatic transition and choosing one target database as critical in Configuring the OpenEdge Replication property files.
If you choose the target database that automatically transitions to a normal OpenEdge database as your hot standby, it is possible that transactions could be lost when the failure occurs. If you are running in synchronous mode and the target database transitions to a normal OpenEdge database, a maximum of one transaction per client is lost. If you are running in asynchronous mode, some number of transactions can be lost per client, depending on how far behind the OpenEdge Replication server was when the connection was lost. It also depends on how far behind your source database and server are.
The advantage to automatic transitioning is that your users will have full access to an alternate database as a hot standby during a failure condition. If you cannot allow for the chance that a transaction is lost during failure processing, the target database set for transition will synchronize with your source database once the connection is re-established, so long as you do not transition it. In this scenario there is no risk to transactions being lost.
For OpenEdge Replication to continue normal operations after a failure and the re-establishment of a connection, the OpenEdge Replication server and the OpenEdge Replication agent must perform synchronization.
The sections that follow describe synchronization and its role in communication failure recovery, including transition.
