Change in naming pattern is reported to cause resource re-creation

[vburckhardt]

From this change:
https://github.com/terraform-ibm-modules/terraform-ibm-landing-zone-vpc/commit/33309f8a6ec200fded1b3f16d6ff735f2a9f92dc#diff-dc46acf[…]9bf2532fbbL72

Report that the removal of -vpc suffix on the vpc resource is causing resource to be recreated.

32 new resources, 25 changed, 32 destroyed.

• Validate that this is actually the root cause - eg: need to see in detail whether the vpc name change is causing a recreate (it is possible to change name of vpc without any resource being recreated in UI). If so, is there an issue at provider level?

[maheshwarishikha]

The PR mentioned above created 7.21.0 version.

I compared basic example using version 7.20.2 and 7.21.0. Basic example is just creating a vpc with one zone subnet.

Created infra using 7.20.2 and then upgrading it using 7.21.0, I get these:

Terraform will perform the following actions:

  # module.slz_vpc.ibm_is_vpc.vpc[0] will be updated in-place
  ~ resource "ibm_is_vpc" "vpc" {
        id                          = "r022-bed819b7-82ec-4534-b312-d9a8c62a9c13"
      ~ name                        = "sm1-vpc-vpc" -> "sm1-vpc"
        tags                        = []
        # (25 unchanged attributes hidden)

        # (1 unchanged block hidden)
    }

Plan: 0 to add, 1 to change, 0 to destroy.

There is no destroy as such. So the changes in naming convention for VPC - is not destroying VPC.

To check further, I'll try creating more resources and then upgrade.

Do we have any resource name which is getting destroyed for the customer?

[maheshwarishikha]

Added some more resources like 2 VPCs, transit gateway, flow log collector (cos instance and bucket)...
Tested with versions 7.20.2 , 7.21.0 and the latest one.

Still resources are not being marked as update-in-place

with 7.21.0 version -

Plan: 2 to add, 5 to change, 0 to destroy.

Changes to Outputs:
  ~ management_vpc_name = "test-management-management-vpc" -> "test-management-management"
  ~ workload_vpc_name   = "test-workload-workload-vpc" -> "test-workload-workload"

with latest version -

Plan: 3 to add, 5 to change, 1 to destroy.

Changes to Outputs:
  ~ management_vpc_name = "test-management-management-vpc" -> "test-management-management"
  ~ workload_vpc_name   = "test-workload-workload-vpc" -> "test-workload-workload"

Here,

• destroy is for # module.workload_vpc.module.workload_vpc.ibm_iam_authorization_policy.policy[0] must be replaced because of scoping auth policy for flow logs to the exact cos bucket.
• add is for ibm_cos_bucket_lifecycle_configuration.cos_bucket_lifecycle and for time_sleep.

I dont think that the removal of -vpc suffix is really causing destroy resources.