Investigate solutions to vacuuming our database prior to failover
During a recent failover production#637 (closed), patroni-04 was elected leader. Upon becoming the leader, we received an alert that we've exceeded our thresholds for Dead Tuples.
This node started off with high percentages of dead tuples.
Use this issue to investigate and resolve the following questions:
- Why do secondaries not have the same data or analyze table information for dead tuples as the primary?
- Should this alert be considered high priority?
- If this is a problem what can be done prior to failovers to ensure potentially elected secondaries do not have a high percentage of dead tuples?