Open tcrowder-koerber opened 3 weeks ago
Regarding the creation of RB/CRB, it may be that some of the commands are silently failing. Please check the note in the terraform documentation for remote-exec
.
Please append set -o errexit
to the list of inline commands executed to create the CRB, here, and attach any log that may explain why the resource creation fails.
Community Note
Terraform Version and Provider Version
Terraform 1.7.5 OCI provider 6.9.0
Affected Resource(s)
metrics server, autoscaler and custom role bindings
Terraform Configuration Files
OKE Module 5.1.8
Attributes: cluster_type = "enhanced" create_service_account = true service_accounts = merge({ kubeconfig-sa = { sa_namespace = "kube-system" sa_name = "kubeconfig-sa" sa_cluster_role = "cluster-admin" sa_cluster_role_binding = "sa-crb" } }, var.cluster_service_accounts) create_iam_resources = true create_operator = true operator_install_helm = true cluster_autoscaler_install = true metrics_server_install = true
... nodepool-1 = { allow_autoscaler = true }
Debug Output
No output related. Generally the autoscaler and metrics-server will install OK but the CRB failed but reports OK. Before adding the CRB, the metrics-server would have issues. The clusters create fine, kubectl is fine, but metrics-server is missing or the CRB is missing. I run post installation validation checks on these resources now due to how often it fails.
The resolution for me is to set the values to false for the affected resources and apply. Turn them back to true and apply.
Panic Output
Expected Behavior
It should add the resources or error it failed.
Actual Behavior
The apply succeeds but some resources such as metrics-server or custom role bindings are not present.
Steps to Reproduce
terraform apply
Important Factoids
The resolution for me is to set the values to false for the affected resources and apply. Turn them back to true and apply.
References