Merged in dev/ttang2/recorder_scaling_across_AZs_21.08.23 (pull request

elastic#210)

recorder scaling across AZs

* recorder scaling across AZs

* Merged master into dev/ttang2/recorder_scaling_across_AZs_21.08.23
* remove placement

* Merge branch 'dev/ttang2/recorder_scaling_across_AZs_21.08.23' of bitbucket.org:engageli/devops into dev/ttang2/recorder_scaling_across_AZs_21.08.23


Approved-by: Can Yildiz

aws/ams-cluster-v1-tf/main.tf

@@ -955,7 +955,7 @@ module "recorder_instances" {
 
   # In order to compatible with the current subnet plan, use the existing one.
   # recorder_subnet_ids = module.network.recorder_subnet_ids
-  recorder_subnet_ids = aws_subnet.subnets.*.id
+   recorder_subnet_ids = module.network.recorder_subnet_ids
 
   recorder_key_name   = "tf-${var.stack_name}-${var.recorder_key_name}"
   ssh_public_key_file = var.ssh_public_key_file

aws/ams-cluster-v1-tf/tfmodules/recorder_instances/recorder_instances.tf

@@ -41,11 +41,6 @@ resource "aws_launch_template" "recorder" {
     }
   }
 
-  placement {
-    # recorder instance is created in the first AZ, which should be enough.
-    availability_zone = data.aws_availability_zones.available.names[0]
-  }
-
   tag_specifications {
     resource_type = "instance"
     tags = {
@@ -77,7 +72,7 @@ resource "aws_autoscaling_group" "recorder" {
   # TODO: availability_zones or vpc_zone_identifiers? How many?
   # availability_zones = data.aws_availability_zones.available
   # recorder asg is created in the subnet of the first AZ, which should be enough.
-  vpc_zone_identifier = [var.recorder_subnet_ids[0]]
+  vpc_zone_identifier = var.recorder_subnet_ids
   default_cooldown    = var.recorder_asg_default_cooldown
   launch_template {
     id      = aws_launch_template.recorder.id