EKS ASG Worker Desired Capacity Change being ignored

[datadarius]

I have issues

For an existing Cluster that has been deployed using version 1.6 I would like to modify the worker counts.

I'm submitting a...

• support request

What is the current behavior?

When editing asg_desired_capacity it does not affect terraform plan and the ASG is not modified.

module "eks_dev" {
  source  = "terraform-aws-modules/eks/aws"
  version = "1.6.0"
...
  workers_group_defaults = {
   key_name = "exploration-zone"
   asg_desired_capacity          = "4"                             # Desired worker capacity in the autoscaling group.
   asg_max_size                  = "5"                             # Maximum worker capacity in the autoscaling group.
   asg_min_size                  = "2"                             # Minimum worker capacity in the autoscaling group.
   instance_type = "m4.large" # Size of the workers instances.
...

If this is a bug, how to reproduce? Please include a code sample if relevvant.

• Deploy an EKS Cluster with custom worke_group_defaults
• After succesfull deployment, try to modify the asg_desired_capacity

What's the expected behavior?

• The asg parameter change should be reflected in terraform plan and applicable.

Are you able to fix this problem and submit a PR? Link here if you have already.

sry, no.

Environment details

AWS region us-east-1

• Affected module version:
• OS: Terraform executed in both ubuntu 16.04 as well as hashicorp/terraform:light
• Terraform version: 0.11.8

Any other relevant info

Also testes with 1.7 release. No impact.
.terraform was also removed and terraform init performed.

[max-rocket-internet]

This is because desired_capacity is ignored due not wanting it to interfere with the cluster-autoscaler: https://github.com/terraform-aws-modules/terraform-aws-eks/blob/master/workers.tf#L22

So currently you would need to set this manually outside of Terraform.

Feel free to make a PR to make this setting optional 🙂

[datadarius]

Thank you for the quick and helpful response. I think the auto-scaler aspect is a fair point. I am fine if this is working as intended for the time being and will make use of the autoscaler once the environment has been bootstrapped.

[max-rocket-internet]

OK cool, thanks for the update.

[larslevie]

Would it make sense to exclude desired_capacity from the ignore lifecycle hook unless autoscaling_enabled is true for a given worker group?

[max-rocket-internet]

Sure

[dpiddockcmp]

lifecycle block does not support interpolations hashicorp/terraform#3116

[github-actions]

I'm going to lock this issue because it has been closed for 30 days ⏳. This helps our maintainers find and focus on the active issues. If you have found a problem that seems similar to this, please open a new issue and complete the issue template so we can capture all the details necessary to investigate further.