The issue where APs failed over to the tertiary controller instead of the secondary controller during a controller failure is likely due to the secondary controller being at capacity. When a controller reaches its maximum capacity, it cannot accept any additional APs, causing them to failover to the next available controller in the high availability configuration.