Why does Terraform's GCP google_project_iam_member resource take so long to deploy?

Question

I have 15 google_project_iam_member resources being deployed in a single count loop on Terraform. It's taking around 8 minutes for each resource to deploy (running in parallel) and I think there's probably something wrong. No other resource takes that long. The code:

## This code converts the user_permissions map to the format [ {members = x, role = y}, ...  ]

locals {
  permissions = flatten([for group in var.user_permissions :
    [for tuple in setproduct(group.emails, group.roles) : {
      members = tuple[0],
      roles   = tuple[1]
  }]])
}

resource "google_project_iam_member" "ib_admins_access" {
  count   = length(local.permissions)
  project = var.project_id
  role    = "roles/${local.permissions[count.index].roles}"
  member  = local.permissions[count.index].members
}

And the user_permissions map is in the following format:

user_permissions = {
 group_x = {
   emails = [<emails>]
   roles = [
     <roles>
   ]
 }

These 2 policies took 20+minutes. They were storage.objectAdmin and viewer for 2 different emails

module.iam.google_project_iam_member.ib_access_predefined_roles[6]: Still creating... [22m1s elapsed]
module.iam.google_project_iam_member.ib_access_predefined_roles[15]: Still creating... [22m1s elapsed]

I'm trying to construe a dictionary like above and have it deployed in an non-authoritative manner. I figured it has probably something to do with policy updating, like the policies are fighting each other. I thought about delaying or doing it sequentially but could not find resources or terraform commands to do that inside the loop. Is there a better way to do this? What is wrong with my code?

Answer 1

I was using WSL 2 and there is probably some problem with the internet connection there. When I changed to PowerShell, it ran much faster.