Skip to content

[Bug]: Terraform cluster deployment missing coredns and kube-proxy addons #1782

@r-repka

Description

@r-repka

Installation method

Own AWS account

What happened?

When creating the EKS cluster using the Terraform configuration in cluster/terraform/eks.tf, the cluster is provisioned without coredns and kube-proxy addons. Only vpc-cni is defined in cluster_addons.

The module terraform-aws-modules/eks/aws version 21.0.9 (used in this repo) hardcodes bootstrap_self_managed_addons = false (ref), meaning all addons must be explicitly declared via the EKS addons API. Since coredns and kube-proxy are not listed in cluster_addons, they are never installed.

This results in no DNS resolution inside the cluster — pods cannot resolve Service names (e.g. java.net.UnknownHostException: carts-dynamodb).

What did you expect to happen?

The cluster should have coredns and kube-proxy installed as EKS managed addons.

How can we reproduce it?

Run terraform apply using cluster/terraform/ as-is
Run aws eks list-addons --cluster-name eks-workshop — only vpc-cni is listed
Deploy sample application — DNS resolution fails, e.g. between carts App and DB

Anything else we need to know?

Invalid parameter names in eks.tf
cluster_name -> name
cluster_version -> kubernetes_version
cluster_endpoint_public_access -> endpoint_public_access
create_cluster_security_group -> create_security_group
cluster_security_group_additional_rules -> security_group_additional_rules
cluster_remote_network_config -> remote_network_config

EKS version

1.33, as defined in workshop terraform

Metadata

Metadata

Assignees

No one assigned

    Labels

    bugSomething isn't working

    Type

    No type

    Projects

    Status

    ✅ Done

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions