From my own experience in a H.A. setup, the postgres repository is quite stable and I haven't seen it failover. But it could happen if and when the server that hosts the worker with the primary crashes for whatever reason. The cluster controller will recognize that the active repository has failed and will swap over to the passive. This repository is mission critical to the operation of any functionality with TS.
2 of 2 people found this helpful
If the active repository stops responding after a certain amount of time, then failover to the passive repository occurs. It can stop responding due to a network issue, or simply resource exhaustion on the host (e.g., excessive cpu utilization from another application). Check your Windows event logs, CPU and network monitoring if you have them. You can look in the Tableau Server clustercontroller log for "Active Repository DOWN" messages, which should tell you when the issue occurred. Looking at the pgsql logs can also be helpful in helping you determine the issue.
Thanks for your responses!