About repository clusters

A repository cluster is a cluster of repository services that uses the master/slave principle.

One service is the master, which is the active repository service in a Collibra Data Intelligence Cloud environment. This means that all data is stored in and retrieved from this service. The master is mandatory in a cluster.

The other repository services are slaves. The slaves are regularly synchronized with the master to become exact copies of the master repository. If the master is no longer available, you can replace it with one of the slaves. The slaves follow the principle of warm stand-by or log shipping.

With warm standby servers, the master server operates in continuous archive mode, the slave servers operate in continuous recovery mode. The recovery mode consists of reading the write-ahead log files, WAL files, of the master server. It has very little impact on the master server's performance. For more information about warm standby, see the PostgreSQL documentation.

The repository services that you add to a cluster have to run on nodes that don't have the hostname localhost.