My minimum healthy percentage is 100, and maximum is 200. After deployment, it takes 30 seconds to spin up all the new instances (2 in my testing) to the “Running” status. After that, Fargate waits 4 whole minutes before changing the status of the old instances from “Running” to “Deactivating”. Why can’t it do that right away? What’s the wait? Can it be cut?

I have a Fargate cluster setup to handle my API. It keeps dropping out, and I’m not sure what is causing it. When I say dropping out, I mean that when I first deploy the image, it works fine. But after maybe an hour of no use, it stops working entirely.