Abnormality observed during Updation of Workers in the kubify cluster

[pyogesh2]

I have setup up the cluster using Kubify. Initially, the worker type of the VM's were m4.large y default and the Master nodes were 3, Worker nodes were 10. We want the vm types to be m4.4xlarge. I updated the file- https://github.com/gardener/kubify/blob/master/modules/vms/versions.tf with the vm type as m4.4xlarge and increased the worker nodes to 12. Executed terraform init variant, terraform plan variant and terraform apply varaint, After that checked, 2 nodes of master and 2 nodes of worker were created with m4.4xlarge and the remaining 1 master node and 10 nodes were still of the type m4.large.

~ module.instance.module.worker.module.vms.aws_instance.nodes[4]
      instance_type:                             "m4.large" => "m4.4xlarge"

  + module.instance.module.worker.module.vms.aws_instance.nodes[6]
      id:                                        <computed>
      ami:                                       "ami-b5742acf"
      arn:                                       <computed>
      associate_public_ip_address:               <computed>
      availability_zone:                         "us-east-1b"
      cpu_core_count:                            <computed>
      cpu_threads_per_core:                      <computed>
      disable_api_termination:                   "false"
      ebs_block_device.#:                        <computed>
      ephemeral_block_device.#:                  <computed>
      get_password_data:                         "false"
      host_id:                                   <computed>
      iam_instance_profile:                      "perfgardener-eval-worker"
      instance_state:                            <computed>
      instance_type:                             "m4.large"
      ipv6_address_count:                        <computed>
      ipv6_addresses.#:                          <computed>
      key_name:                                  "perfgardener-eval"
      network_interface.#:                       <computed>
      network_interface_id:                      <computed>
      password_data:                             <computed>
      placement_group:                           <computed>

Ideally, all the master and worker nodes should of type m4.4xlarge as per the configuration. But thats not happening. Kindly check. Thanks !

[afritzler]

I would recommend to update the worker/master flavors via the terraform.tfvars file

master = {
  count=3
  flavor_name="medium_4_8"
}

worker = {
  count=3
  flavor_name="medium_2_4"
}

We also have a rolling mechanism in place. So what should happen when you do the first terraform apply, the first master and the first worker is updated. You need then to run apply as long as all the machines have been replaced.