What's the best way to handle HA failover testing without disrupting users?

We're planning our annual DR test for our FortiGate HA pair. Active-Passive setup, running 7.2. Production environment.

Goal: Prove failover works without kicking everyone off. Need to test the actual failover trigger, not just link failures. Past attempts caused brief outages.

What's the cleanest method? I've heard of using `execute ha manage` to shift sessions, but documentation is vague on the impact. Is there a way to force the standby unit to become active *and* seamlessly take over existing connections? Or is a brief blip unavoidable?

Looking for specific CLI steps and the expected user impact. If a brief outage is mandatory, I need to know the duration to plan the change window.