oci_database_cloud_autonomous_vm_cluster fails to update due to unused/invalid property

Community Note

Please vote on this issue by adding a 👍 reaction to the original issue to help the community and maintainers prioritize this request
Please do not leave "+1" or "me too" comments, they generate extra noise for issue followers and do not help prioritize the request
If you are interested in working on this issue or have submitted a pull request, please leave a comment

Terraform Version and Provider Version

Terraform v1.4.2
on linux_amd64
+ provider registry.terraform.io/hashicorp/azurerm v3.49.0
+ provider registry.terraform.io/oracle/oci v4.113.0

Affected Resource(s)

oci_database_cloud_autonomous_vm_cluster

Terraform Configuration Files

resource "oci_database_cloud_autonomous_vm_cluster" "cluster" {
  cloud_exadata_infrastructure_id = var.infra_id
  compartment_id                  = var.compartment_id
  # (snip)

  # Maintenance window
  maintenance_window_details {
    preference                       = "CUSTOM_PREFERENCE"
    is_custom_action_timeout_enabled = false
    custom_action_timeout_in_mins    = 0
    weeks_of_month                   = var.maintenance_schedule.week_of_month
    hours_of_day                     = [var.maintenance_schedule.start_hour_utc]

    dynamic "months" {
      for_each = local.maintenance_months
      content {
        name = months.value
      }
    }

    dynamic "days_of_week" {
      for_each = var.maintenance_schedule.day_of_week
      content {
        name = days_of_week.value
      }
    }
  }

Expected Behavior

After creation of the VM cluster, the updatable properties can be modified. As an example, the display name.

Actual Behavior

Upon making any change and running apply, the following occurs:

Error: 400-InvalidParameter, MaintenanceWindow leadTimeInWeeks  : is only supported for Exadata DbSystems.
│ Suggestion: Please update the parameter(s) in the Terraform config as per error message MaintenanceWindow leadTimeInWeeks  : is only supported for Exadata DbSystems.
│ Documentation: https://registry.terraform.io/providers/oracle/oci/latest/docs/resources/database_cloud_autonomous_vm_cluster 
│ API Reference: https://docs.oracle.com/iaas/api/#/en/database/20160918/CloudAutonomousVmCluster/UpdateCloudAutonomousVmCluster 
│ Request Target: PUT https://database.eu-amsterdam-1.oraclecloud.com/snip/cloudAutonomousVmClusters/ocid1.cloudautonomousvmcluster.oc1.eu-amsterdam-1.snipsnip
│ Provider version: 4.113.0, released on 2023-03-22. 
│ Service: Database Cloud Autonomous Vm Cluster 
│ Operation Name: UpdateCloudAutonomousVmCluster 
│ OPC request ID: snip

Steps to Reproduce

Create a oci_database_cloud_autonomous_vm_cluster instance on X9M-2. Specify a custom maintenance window.
Change an updatable property related not related to the maintenance window.
The plan should show that property being updated, but no modifications to the maintenance schedule (as expected)
The apply fails with the message above.

Important Factoids

It appears related to how the provider deals with 'lead time in weeks'. Note that in the code we never specify this property and Terraform will create the cluster accordingly. Afterwards, I verified through oci db cloud-autonomous-vm-cluster get --cloud-autonomous-vm-cluster-id that when you request the cluster from the API, the corresponding 'lead time' property is set to null:

 "maintenance-window": {
      "custom-action-timeout-in-mins": null,
      "days-of-week": [
        {
          "name": "SUNDAY"
        }
      ],
      "hours-of-day": [
        4
      ],
      "is-custom-action-timeout-enabled": null,
      "is-monthly-patching-enabled": null,
      "lead-time-in-weeks": null,
      "months": [
        {
          "name": "FEBRUARY"
        },
        {
          "name": "MAY"
        },
        {
          "name": "AUGUST"
        },
        {
          "name": "NOVEMBER"
        }
      ],
      "patching-mode": null,
      "preference": "CUSTOM_PREFERENCE",
      "weeks-of-month": [
        1
      ]
    },

However, when you issue a terraform state show it will show for that same resource the value '0' for the lead time:

maintenance_window_details {
        custom_action_timeout_in_mins    = 0
        hours_of_day                     = [
            4,
        ]
        is_custom_action_timeout_enabled = false
        is_monthly_patching_enabled      = false
        lead_time_in_weeks               = 0
        preference                       = "CUSTOM_PREFERENCE"
        weeks_of_month                   = [
            1,
        ]

        days_of_week {
            name = "SUNDAY"
        }

        months {
            name = "FEBRUARY"
        }
        months {
            name = "MAY"
        }
        months {
            name = "AUGUST"
        }
        months {
            name = "NOVEMBER"
        }
    }

With TF_LOG=DEBUG OCI_GO_SDK_DEBUG=v I was able to verify that when Terraform issues the PUT request to update the resource, it has the following in the body of the request:

"maintenanceWindowDetails":{"customActionTimeoutInMins":0,"daysOfWeek":[{"name":"SUNDAY"}],"hoursOfDay":[4],"isCustomActionTimeoutEnabled":false,"isMonthlyPatchingEnabled":false,"leadTimeInWeeks":0,"months":[{"name":"FEBRUARY"},{"name":"MAY"},{"name":"AUGUST"},{"name":"NOVEMBER"}],"preference":"CUSTOM_PREFERENCE","weeksOfMonth":[1]}

It appears that the null from the API is imported into the Terraform state as 0 which is then propagated in subsequent updates.

oracle / terraform-provider-oci