Thanks.

I have not had that issue, but do your containers and targets reach a healthy state before this occurs? If the problem is with your target group my guess would be that the health check runs too early, something you can adjust when creating your service, or you could try to give the container more resources. It could also be a problem with the health check URL, one typo is enough.

If it reaches a healthy state I would take a look in CloudWatch if any log statement could explain why it is recreating. And also chech the status on the stopped containers in the ECS dashboard.