How would you describe a Kubernetes operator if we have a master database and two replicas, and what happens if the main database fails? Briefly.

A Kubernetes operator is a controller that extends Kubernetes functionality by automating the management of complex applications and resources.

In the case of a master database with two replicas, the operator monitors the database's state. If the primary database fails, the operator detects this (for example, through readiness/liveness probes or StatefulSet status) and automatically switches the role to one of the replicas, making it the new primary database.

Thus, the operator ensures automatic recovery and maintenance of the desired cluster state without manual intervention.