terraform-provider-ibm breaks and cannot handle LB creation with count greater than 1

[faseyiks]

Terraform Version

0.13.5
Terraform v0.13.5

• provider registry.terraform.io/hashicorp/null v3.0.0
• provider registry.terraform.io/ibm-cloud/ibm v1.19.0

Affected Resource(s)

Please list the resources as a list, for example:

• ibm_is_lb.application
• ibm_is_lb_listener.applistener
• ibm_is_lb_listener.applistener[0]
• ibm_is_lb_listener.applistener[1]
• ibm_is_lb_pool.apppool[0]
• ibm_is_lb_pool.apppool[1]

If this issue appears to affect multiple resources, it may be an issue with Terraform's core, so please mention this.

Terraform Configuration Files

resource "ibm_is_lb" "application" {
  count          = ! var.enable_strict_layer_4 ? 1 : 0
  name           = format("%s-load-balancer-%s-%s", "application", var.tags["Environment"], local.protocol_selector)
  subnets        = var.subnets_names != "" ? [data.ibm_is_subnet.subnet[0].id] : data.ibm_is_subnets.infrasubnets.subnets.*.id
  resource_group = data.ibm_resource_group.resourcegroup.id
  type           = local.block_in_private_range ? "private" : "public"
  #tags          = var.tags
}

resource "ibm_is_lb_pool" "apppool" {
  count                           = ! var.enable_strict_layer_4 ? length(var.inbound_destination_ports[local.protocol_selector]) : 0
  lb                              = ibm_is_lb.application[0].id
  name                            = format("app-pool-%s-%s-%s", var.tags["Environment"], local.protocol_selector, count.index + 1)
  protocol                        = local.protocol_selector
  algorithm                       = lookup(var.lb_pool_params, "algorithm")
  health_delay                    = lookup(var.lb_pool_params, "delay")
  health_retries                  = lookup(var.lb_pool_params, "retries")
  health_timeout                  = lookup(var.lb_pool_params, "timeout")
  health_type                     = var.https_certificate_crn != "" ? "http" : "tcp"
  health_monitor_url              = lookup(var.lb_pool_params, "monitor_url")
  health_monitor_port             = lookup(var.lb_pool_params, "health_port")
  session_persistence_type        = lookup(var.lb_pool_params, "persistence_type", "http_cookie")
  session_persistence_cookie_name = lookup(var.lb_pool_params, "cookie_name", "http_cookie")
  depends_on                      = [ibm_is_lb.application, null_resource.app_lb_wait_for_time]
}

resource "ibm_is_lb_listener" "applistener" {
  depends_on           = [ibm_iam_authorization_policy.authorization, ibm_is_lb_pool.apppool, null_resource.app_lb_wait_for_time]
  count                = length(var.inbound_destination_ports[local.protocol_selector])
  lb                   = ibm_is_lb.application[0].id
  port                 = element(var.inbound_destination_ports[local.protocol_selector], count.index)
  protocol             = local.protocol_selector
  default_pool         = element(split(local.separator, ibm_is_lb_pool.apppool[count.index].id), 1)
  certificate_instance = var.https_certificate_crn

Expected Behavior

Terraform will provision one LB, 2 lb_pools and 2 lb_listeners

Actual Behavior

There is an issue with the LB implementation itself such that after creating the first LB_LISTENER, it then breaks terraform (obviously a race condition with the underlying update of the LB ) with the error below
Error: Error while creating Load Balanacer Listener err The load balancer with ID 'r018-78abc19d-4a20-40d2-aae7-f50fb05de2ae' cannot be updated because its status is 'UPDATE_PENDING'.
{
....
"errors": [
{
"code": "load_balancer_update_conflict",
"message": "The load balancer with ID 'rxxx-xxxxxxx-4a20-40d2-aae7-xxxxxxxxx' cannot be updated because its status is 'UPDATE_PENDING'.",
"more_info": "https://cloud.ibm.com/docs/infrastructure/vpc-on-classic?topic=vpc-on-classic-rias-error-messages#load_balancer_update_conflict"
}
],
"trace": "xxxxxx-9852-4213-bfb2-xxxxxxxxxxxxxx"
},
"RawResult": null

Steps to Reproduce

Please list the steps required to reproduce the issue, for example:

1. Copy the code and fill in for the variables or substitute with raw values
2. terraform apply

Important Factoids

None

References

I have seen several links where the discussion of IBM LB implementation is discussed. I have also seen the below
#295 (comment).
It seems to me that a generic way to implement a "wait_for_status_update" on the lb_listener is required as a workaround.

[hkantare]

Released in https://github.com/IBM-Cloud/terraform-provider-ibm/releases/tag/v1.20.0
Cloud docs will be updated soon
Registry docs:
https://registry.terraform.io/providers/IBM-Cloud/ibm/latest/docs