healthCheckGracePeriodSeconds

The period of time, in seconds, that the Amazon ECS service scheduler ignores unhealthy Elastic Load Balancing, VPC Lattice, and container health checks after a task has first started. If you do not specify a health check grace period value, the default value of 0 is used. If you do not use any of the health checks, then healthCheckGracePeriodSeconds is unused.

If your service has more running tasks than desired, unhealthy tasks in the grace period might be stopped to reach the desired count.