I have 2 metrics for my Instances attached to my Load balancer and Auto Scaling Group, one to scale instances up based on CPU Utilization and one to scale down.
My scaling up one works fine whereas the scaling down Instances when under 40% CPU Util seems to work by the alarm being "In Alarm" and it having <40% CPU, it removes 1 Instance as it should but after that it doesn't react taking more than 1 Instance down and I have left it like this for up to 7minutes with just the 1 instance being taken down.
Any idea why it might be doing this?