What process decides when a Repository failover is required?

The Cluster Controller process is responsible for monitoring the health of Tableau Server components in a high-availability configuration. It continuously checks the status of the active Repository (PostgreSQL database). If the Cluster Controller detects that the active Repository has failed or is unresponsive, it initiates the failover process, promoting the passive Repository to become the new active one. This ensures minimal disruption to the server's operation by maintaining the availability of critical metadata.

B. Coordination Service: The Coordination Service (based on Apache ZooKeeper) maintains the server's state and topology, ensuring consensus in the cluster, but it does not actively monitor and initiate the failover decision.

C. Gateway: The Gateway process is a web server that handles and routes client requests to other Tableau Server processes. It is not involved in monitoring internal component health or managing failover.

D. Backgrounder: The Backgrounder process runs asynchronous server tasks, such as extract refreshes and subscriptions. It has no role in the high-availability monitoring or failover mechanisms of the server.

---

1. Official Salesforce/Tableau Documentation: Tableau Server Help, "Tableau Server Processes".

Section: Cluster Controller

Content: "The Cluster Controller is responsible for monitoring the various components, detecting failures, and running failover when needed." This statement directly attributes the failover decision to the Cluster Controller.

2. Official Salesforce/Tableau Documentation: Tableau Server Help, "Repository Failover".

Section: How failover works

Content: "When the active repository fails, the passive repository automatically becomes the active repository. The failover is handled by the Cluster Controller..." This explicitly confirms the Cluster Controller's role in managing Repository failover.

3. Official Salesforce/Tableau Documentation: Tableau Server Help, "Tableau Server Coordination Service".

Section: Overview

Content: "The Coordination Service is the 'single source of truth' for the server’s topology, configuration, and state." This highlights its role in state management, which supports failover, but distinguishes it from the Cluster Controller's role of initiating the failover.