[BUG] Elasticache endpoints are `null` under some circumstances

[spkane]

What is the current behavior?

This code applies, but producesnull outputs for elasticache_replication_group_primary_endpoint_address and elasticache_replication_group_reader_endpoint_address

module "elasticache_redis" {
  source  = "umotif-public/elasticache-redis/aws"
  version = "3.0.0"

  name_prefix        = test
  num_cache_clusters = 2
  node_type          = cache.t3.small

  cluster_mode_enabled    = true
  replicas_per_node_group = 1
  num_node_groups         = 2

  engine_version           = "6.x"
  port                     = 6379
  maintenance_window       = "mon:03:00-mon:04:00"
  snapshot_window          = "04:00-06:00"
  snapshot_retention_limit = 7

  automatic_failover_enabled = true

  at_rest_encryption_enabled = true
  transit_encryption_enabled = true
  auth_token                 = random_password.auth_token.result

  apply_immediately = true
  family            = "redis6.x"
  description       = var.description

  subnet_ids = var.subnet_ids
  vpc_id     = var.vpc_id

  ingress_cidr_blocks = ["0.0.0.0/0"]

  parameter = [
    {
      name  = "repl-backlog-size"
      value = "16384"
    }
  ]

  tags = var.tags
}

Either this config is wrong-headed somehow that AWS still allows (which is possible as I am not super familiar with Elasticache) or the logic here is broken, because my cluster appears to have this defined in the state file:

"configuration_endpoint_address": "clustercfg.foo.bar.use1.cache.amazonaws.com",

but not these:

            "primary_endpoint_address": null,
            "reader_endpoint_address": null,

If the current behavior is a bug, please provide the steps to reproduce and if possible a minimal demo of the problem.

Use this module with the code above.

What is the expected behavior?

The redis endpoint is set correctly in the outputs.

Software versions?

3.0.0

[Ohid25]

Thank you @spkane for raising this as a bug.

The fix is now a part of 3.1.1 release.