Bogus use_ip_aliases state reported on upgrade to v3.7.0

[akilman]

Community Note

• Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request.
• Please do not leave +1 or me too comments, they generate extra noise for issue followers and do not help prioritize the request.
• If you are interested in working on this issue or have submitted a pull request, please leave a comment.
• If an issue is assigned to the modular-magician user, it is either in the process of being autogenerated, or is planned to be autogenerated soon. If an issue is assigned to a user, that user is claiming responsibility for the issue. If an issue is assigned to hashibot, a community member has claimed the issue already.

Terraform Version

$ terraform -v
Terraform v0.12.20
+ provider.google v3.7.0
+ provider.kubernetes v1.10.0
+ provider.random v2.2.1

Affected Resource(s)

• google_container_cluster

Terraform Configuration Files

resource "google_container_cluster" "clever-cluster-name" {
    name = "${var.application_instance_name}-cluster"
    depends_on = [google_project_service.kubernetes]

    remove_default_node_pool = true
    initial_node_count = 1

    ip_allocation_policy {
    }

    network = google_compute_network.clever-vpc.self_link
}

Debug Output

https://gist.github.com/akilman/f06c0d2a00f86ae2aeac0f3e37f34023

Panic Output

Expected Behavior

GCP reports useIpAliases == true in the UI after upgrade to v3.7.0 of the Google provider and terraform apply. Terraform should reflect the same correct state.

Actual Behavior

terraform show | grep ip_allocation_policy -A 8 details use_ip_aliases == false

Steps to Reproduce

1. terraform apply w/ v2.5.0 Google Provider with

    ip_allocation_policy {
        use_ip_aliases = true
    }

2. Upgrade to 3.7.0 provider, following the upgrade guide which instructs to remove use_ip_aliases = true as true is the default with the presence of the ip_allocation_policy
3. terraform apply with newly upgraded Provider + updated config
4. Observe use_ip_aliases has changed to false in Terraform. However, in GCP useIpAliases is still set to true. Can be confirmed via:

$ gcloud container clusters describe clever-cluster | grep useIpAliases

Important Factoids

References

• Confusing upgrade instructions regarding container_cluster.ip_allocation_policy.use_ip_aliases #5154

[akilman]

For good measure, attempted to spin up a fresh cluster to see if it was just a state reporting issue on upgrade. But, was able to reproduce the same behavior on a newly provisioned environment.

terraform show reports use_ip_aliases = false, and GCP's UI reports VPC native (Alias IP) attribute of the cluster is set to true 🤷‍♂

[edwardmedia]

@c2thorn after tf init, I see this in the plan

      + ip_allocation_policy {
          + cluster_ipv4_cidr_block       = (known after apply)
          + cluster_secondary_range_name  = (known after apply)
          + create_subnetwork             = (known after apply)
          + node_ipv4_cidr_block          = (known after apply)
          + services_ipv4_cidr_block      = (known after apply)
          + services_secondary_range_name = (known after apply)
          + subnetwork_name               = (known after apply)
          + use_ip_aliases                = (known after apply)
        }

after tf apply I see this in the show (it is false)

    ip_allocation_policy {
        cluster_ipv4_cidr_block       = "10.4.0.0/14"
        cluster_secondary_range_name  = "gke-name-cluster-pods-432f38af"
        create_subnetwork             = false
        services_ipv4_cidr_block      = "10.0.0.0/20"
        services_secondary_range_name = "gke-name-cluster-services-432f38af"
        use_ip_aliases                = false
    }

In the meantime, GCP UI show

VPC-native (alias IP)	Enabled

[c2thorn]

use_ip_aliases is a removed field since v3.0.0, so ideally it would not appear in state at all. Filed an issue with the Terraform SDK, but in the meantime I'll add a workaround for this field.

[ghost]

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 feel this issue should be reopened, we encourage creating a new issue linking back to this one for added context. If you feel I made an error 🤖 🙉 , please reach out to my human friends 👉 hashibot-feedback@hashicorp.com. Thanks!