[2] Inconsistent conditional result types error

[raviranjanelastisys]

What should be investigated.

While upgrading one cluster of version kubespray 2.21 - we found some problem with inconsistencies . Below is the error

module.compute.openstack_networking_floatingip_associate_v2.k8s_master[1]: Refreshing state... [id=baf25378-8047-4273-9334-18f4eaf8d71c]
module.compute.openstack_compute_instance_v2.k8s_master[1]: Refreshing state... [id=e4e3588e-1915-43c6-bfe7-4e0a0e3d756e]
module.compute.openstack_compute_instance_v2.k8s_master[0]: Refreshing state... [id=a6c1133d-6a08-4507-9c14-62dc9f88d4aa]
module.compute.openstack_compute_instance_v2.k8s_master[2]: Refreshing state... [id=4bf27f0b-f56b-4ab3-bfcc-1d3f68d122dd]
╷
│ Warning: Experimental feature "module_variable_optional_attrs" is active
│ 
│   on versions.tf line 13, in terraform:
│   13:   experiments      = [module_variable_optional_attrs]
│ 
│ Experimental features are subject to breaking changes in future minor or patch releases, based on feedback.
│ 
│ If you have feedback on the design of this feature, please open a GitHub issue to discuss it.
│ 
│ (and one more similar warning elsewhere)
╵
╷
│ Error: Inconsistent conditional result types
│ 
│   on modules/ips/main.tf line 43, in resource "openstack_networking_floatingip_v2" "k8s_nodes":
│   43:   for_each   = var.number_of_k8s_nodes == 0 ? { for key, value in var.k8s_nodes : key => value if value.floating_ip } : {}
│     ├────────────────
│     │ var.k8s_nodes is object with 21 attributes
│     │ var.number_of_k8s_nodes is 0
│ 
│ The true and false result expressions must have consistent types. The 'true' value includes object attribute "elastisys-0", which is absent in the 'false' value.
╵
Terraform found changes for wc-cluster, review the changes.
Continuing here will not apply anything, it will just create a temporary state file.
Continue? [y/N] 

To fix the problem we had to add here - https://github.com/elastisys/kubespray/compare/5b3d5b26fbe0f0aad06d16ee6a4488476e2d29da...80e63ed55da04184f1fcb456f875732d9f43fe56#diff-e13952be66e4a9b780c6d5bf82e76f413b148dea98fda505ed19ce21ffbf16c8

variable "k8s_nodes" {
  default = {}
  type = map(object({
    az                     = string
    flavor                 = string
    floating_ip            = bool
    extra_groups           = optional(string)
    image_id               = optional(string)
    root_volume_size_in_gb = optional(number)
    volume_type            = optional(string)
    network_id             = optional(string)
    server_group           = optional(string)
    cloudinit              = optional(object({
      extra_partitions = list(object({
        volume_path     = string
        partition_path  = string
        partition_start = string
        partition_end   = string
        mount_path      = string
      }))
    }))
  }))
}

Investigate if this needs to be included in general in our main branch ?

What artifacts should this produce.

[Ajarmar]

Timebox investigation to 1 day. If you can replicate the issue but can't find the root cause, just implement the suggested fix.