The Qlik Sense Repository Service (QRS) behaves differently depending on if it is deployed on the central node or on a rim node.
Central node
Within a multi-node site, one instance of the Qlik Sense Repository Service (QRS) runs on each node. The QRS running on the central node is considered to be the master. The master QRS has direct access to the central repository database, whereas the other QRSs only have access to a local repository database on the node where they are running. The master QRS synchronizes the central repository database and the local repository databases.
When the master QRS starts, it connects to the central repository database. If no database exists, the master QRS builds the database and populates it with initial data. In a default Qlik Sense installation, the repository database is a specific instance of PostgreSQL that runs its own database cluster specifically for the repository.
Rim nodes
When the Qlik Sense Repository Service (QRS) on a rim node starts, it connects to the local repository database on the node. If no local repository database exists, the QRS waits until it communicates with the central node.
In a default Qlik Sense installation, the repository database is a specific instance of PostgreSQL that runs its own database cluster specifically for the repository.