Dirty plan for aws_lb_target_group_attachment during in-place deployment

Terraform Providers AWS
1

We have a load balancer set up with several ec2 instances behind it. During a deploy, instances will be removed from the target group and updated before being placed back into the target group.

If terraform plan is run while this is happening, it notices a diff in the target group attachment. This makes sense but it’s a bit of a nuisance, since it makes you think the plan is dirty even though the instance will be added back shortly when the deploy completes. What is the best way to tell terraform to ignore this diff? Code sample:

resource "aws_alb_target_group" "flask_nginx" {
  name                 = "web-nginx"
  port                 = 80
  protocol             = "HTTP"
  vpc_id               = var.vpc_id
  deregistration_delay = 5
  health_check {
    interval            = 8
    path                = "/health"
    port                = 80
    protocol            = "http"
    timeout             = 7
    healthy_threshold   = 2
    unhealthy_threshold = 5
    matcher             = "200"
  }
  target_type = "instance"
}

resource "aws_alb_target_group_attachment" "flask_nginx" {
  target_group_arn = aws_alb_target_group.flask_nginx.arn
  target_id        = aws_instance.flask[count.index].id
  count            = var.instances

  lifecycle {
    ignore_changes = [target_id]
  }
}