terraform-aws-eks-workers

What's Changed

• Support other AWS partitions other than Commercial by @RothAndrew in https://github.com/cloudposse/terraform-aws-eks-workers/pull/93

New Contributors

• @RothAndrew made their first contribution in https://github.com/cloudposse/terraform-aws-eks-workers/pull/93

Full Changelog: https://github.com/cloudposse/terraform-aws-eks-workers/compare/1.0.0-rc1...1.0.0-rc2

• Add kubelet_extra_args to allow passing extra arguments to kubelet
• Restore behavior of bootstrap_extra_args to allow passing extra arguments to bootstrap.sh
• Add autoscaling_group_tags to allow passing tags only to the Auto Scaling Group that will not be applied elsewhere.

what

• adds mixed instance policy configuration

why

• allows to use mixed group of on demand/spot of differing types for EKS nodes

what

• fix for block_duration_minutes to use correct type

why

• block_duration_minutes should be a number. See https://www.terraform.io/docs/providers/aws/r/launch_template.html#block_duration_minutes

what

• Fix bootstrap-extra-args: Use KUBELET_EXTRA_ARGS env variable to pass extra kubelet args

why

• There was an issue passing node labels to bootstrap.sh, the bootstrap could not complete when we pass extra kubelet args as an argument, and thus node won't join to the kubernetes cluster. Passing extra kubelet args via KUBELET_EXTRA_ARGS solves this

what

• Add additional userdata to execute on each worker node before and after joining the EKS cluster

why

• https://kubedex.com/90-days-of-aws-eks-in-production

what

• Add var.cluster_security_group_ingress_enabled

why

• To fix count can't be computed error

• To be able to use this module in standalone tests and with terraform-aws-eks-cluster module:

  • When this module is connected to terraform-aws-eks-cluster module, the condition var.enabled && var.cluster_security_group_id != "" in count throws count can't be computed error
  • When this module is used in tests, terraform-aws-eks-cluster module is not present and we don't have var.cluster_security_group_id defined, so we need to disable resource "aws_security_group_rule" "ingress_cluster" when testing

what

• Port module to Terraform 0.12
• Pin all providers
• Add example for testing
• Add bats and terratest for the example
• Add Codefresh badge to point to the test pipeline in terraform-modules project
• Update README
• Rename outpute
  • from worker_role_arn to workers_role_arn
  • from worker_role_name to workers_role_name

why

• Module currently does not work with 0.12. Much easier syntax
• Better regression control
• Automatically test the example on every commit and pull request
• Provision resources on AWS in the test account and check the outputs for the correct values
• terraform-modules project contains pipelines for all terraform modules
• Rename outputs to reflect that the cluster has many worker nodes

what

• Fix typo in README

what

• Add variable workers_role_policy_arns_count

why

• Prevents terraform errors count cannot be computed when the module is used from top-level modules

what

• Add possibility to use an existing security group
• Add possibility to add additional list of security groups to autoscaling group

why

• Useful when users want to use an existing security group and instance profile
• Additional SGs can be used to have access to the EKS workers

what

• Remove duplicate most_recent attribute

why

• Regression from Pull Requests

what

• Add worker_role_name output
• Add possibility to use existing instance profile for workers

why

• worker_role_name can be used for attaching additional policy to existing worker role
• Add possibility to use existing instance profile for workers - useful when you have multiple EKS workers group

what

• Add use_custom_image_id variable
• Update logic to decide if the provided image_id should be used, or the module should lookup the EKS worker AMI

why

• The old logic produces count cannot be computed errors when this modules is used with other modules (e.g. a modules to create a custom AMI and then use it in the image_id variable)

what

• Update eks_worker_ami_name_filter variable
• Add eks_worker_ami_name_regex variable

why

• Improve the AMI name filter to return correct instances
• A regex string to apply to the AMI list returned by AWS

what

• Add output for worker role ARN

why

• This may be useful for example if using two workers set within a cluster

what

• Add missing splat+join for resources with counts
• Rebuild README

why

• Some resources with counts were used without splat+join