fix(redis-cluster): ignore changes on secondary cluster's non modifiable variables post creation

[alexmathieu22]

Hi team!

I am using the memorystore module for redis cluster and found a little unwanted behavior. We have setup our current setup to use a shard_count variable that is then used by both our primary and secondary cluster, e.g.:

module "redis_cluster_primary" {
  source  = "terraform-google-modules/memorystore/google//modules/redis-cluster"
  version = "~> 14.0"

  name                    = var.name
  project_id              = var.project_id
  shard_count             = var.redis_architecture.shard_count
  region                  = var.primary_region
  network                 = var.network
  node_type               = var.redis_architecture.node_type
  enable_apis             = var.enable_apis
  replica_count           = var.redis_architecture.replica_count
  transit_encryption_mode = var.transit_encryption_mode
  authorization_mode      = var.authorization_mode
  redis_configs           = var.redis_configs

  persistence_config = var.persistence_config

  service_connection_policies = var.primary_service_connection_policies

  depends_on = [
    module.enable_apis,
  ]
}

module "redis_cluster_secondary" {
  source  = "terraform-google-modules/memorystore/google//modules/redis-cluster"
  version = "~> 14.0"

  count = var.secondary_region != null ? 1 : 0

  name                    = var.name
  project_id              = var.project_id
  shard_count             = var.redis_architecture.shard_count
  region                  = var.secondary_region
  ...

This works perfectly on creation, but I realized that on modification of the number of shards, to scale out for example, our terraform apply failed. This is because we received the following error:

│ Error: Error updating Cluster "projects/my-gcp-prj/locations/northamerica-northeast2/clusters/my-cluster": googleapi: Error 400: unsupported path in fieldMask: shard_count for secondary. Allowed values are maintenance_window, maintenance_policy, simulate_maintenance_event, deletion_protection_enabled, replica_count, maintenance_policy.weekly_maintenance_window, cross_cluster_replication_config, automated_backup_config, cluster_endpoints
│ Details:
│ [
│   {
│     "@type": "type.googleapis.com/google.rpc.BadRequest",
│     "fieldViolations": [
│       {
│         "field": "shard_count"
│       }
│     ]
│   }
│ ]
│ 
│   with [...].google_redis_cluster.redis_cluster,
│   on .terraform/.../main.tf line 17, in resource "google_redis_cluster" "redis_cluster":
│   17: resource "google_redis_cluster" "redis_cluster" {
│ 

The error is pretty self explanatory! Since the fields are required on creation, I think it would be pertinent to stop tracking changes on them after creation using ignore_changes.

Currently it makes my TF pipelines fail, and I thought about the fact that we could split the variables so that it doesn't change in lab, but I feel like that would lead us to have code that does not represent the real tf state, so I think using a variable for both and ignoring changes on secondary is a better solution.

Let me know how that sounds, or if you would have a better idea to fix this problem. Thanks!

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.