Amazon EC2 Auto Scaling group should cover multiple Availability Zones

How to fix it

• Interrupts service: Auto Scaling then rebalances, replacing some instances with ones in the new zone, which get no load balancer traffic until that zone is enabled on the load balancer.

In the console

EC2 → Auto Scaling Groups → the group → Details → Network

Subnets

Add a subnet in a second Availability Zone, keeping the current ones

With the AWS CLI

• Overwrites existing settings: The --vpc-zone-identifier list replaces the group's subnets, so list the subnets it uses now as well as the new one.

aws autoscaling update-auto-scaling-group --auto-scaling-group-name <group-name> --vpc-zone-identifier <subnet-id-zone-a>,<subnet-id-zone-b> --region <region>

With Terraform

resource "aws_autoscaling_group" "this" {
  name                = var.asg_name
  min_size            = var.min_size
  max_size            = var.max_size
  vpc_zone_identifier = [var.subnet_id_zone_a, var.subnet_id_zone_b]

  launch_template {
    id      = aws_launch_template.this.id
    version = "$Default"
  }
}