How is it possible to perform an instance refresh within an autoscaling group without any downtime?
Running two ec2 instances with 3 public facing services running 2 tasks each.
The services have an order placement strategy set to spread instanceId, which from my understanding should split the containers between the two instances. When running an instance refresh I am setting Minimum healthy percentage to 50% with the expectation that while one instance is still operational the healthy instance will serve traffic correctly. However checking my load balancer target groups not every service is split properly which then causes at least one to go offline during the refresh.
Not sure what other information I can provide but can pull more configuration details from the services & tasks.
