fix(storagedirective): fix storage directive validator

[anvial]

Description

This fix add the following:

• Add validation logic to check for unknown or null values in the map elements.
• Parse and validate storage directives using juju/storage package (use jujustorage name instead storage).

Fixes: #534

Type of change

• Bug fix (non-breaking change which fixes an issue)

Environment

• Juju controller version:

• Terraform version:

QA steps

terraform {
  required_providers {
    juju = {
      source  = "juju/juju"
      version = "0.14.0"
    }
  }
}
provider "juju" {}

resource "juju_model" "tf-test" {
  name = "tf-test"
}

variable "blocksize"{
  type    = string
  default = "101M"
}

variable "filessize"{
  type    = string
  default = "102M"
}

variable "pgstorages" {
  type = map
  default = {
    "files"  = "2G"
  }
}

resource "juju_application" "test-app-storage" {
  model = juju_model.tf-test.name
  name = "test-app-storage"
  charm {
    name = "ubuntu"
    channel = "latest/stable"
    revision = 24
  }

  # storage_directives = var.pgstorages
  storage_directives = {
    files = var.filessize
    block = var.blocksize
  }

  units = 1
}
...

[hmlanigan]

I haven't completed QA yet, but I have a question.

[hmlanigan]

thought: Should this be continue instead? What if there is more than one storage directive provided?

[anvial]

Looks like we just need to stop validating if we have at least one unknown among variables.

I've updated QA to touch case with multiple vars and and storage directives.
It looks like it's working.

[hmlanigan]

todo: update the comment at line 33 to describe what is going on so we know if it breaks again. Below describes what I found and why this works.

It's only working because the method is called twice by terraform. I added some logging to see what is going on

Warning: CALLED ValidateMap "{\"block\":<unknown>,\"files\":\"73heather\"}"
Warning: CALLED ValidateMap "{\"block\":\"101M\",\"files\":\"73heather\"}"

The second call has the values filled in. I have no idea why terraform is doing this. If the behavior changes, we won't validate all the values.

[hmlanigan]

Once the comment is appropriately updated, approved.

[anvial]

/usr/bin/sg microk8s -c microk8s kubectl -n kube-system rollout status deployment/coredns
  Waiting for deployment "coredns" rollout to finish: 0 of 1 updated replicas are available...
  error: deployment "coredns" exceeded its progress deadline
  Command microk8s kubectl -n kube-system rollout status deployment/coredns failed with return code 1. Will retry after 10000ms

It looks like the GitHub action fails, not because of the PR. So I'll merge it.