`aci_fabric_node_member` resource do not remove node from inventory

[edudppaz]

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 other comments that do not add relevant new information or questions, 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

v1.5.6

APIC version and APIC Platform

• V 6.0(3e) on prem

Affected Resource(s)

• aci_fabric_node_member

Terraform Configuration Files

resource "aci_fabric_node_member" "fabricNodes" {
  for_each = { for node in try(local.nodes, []) : node.id => node }
  name     = each.value.name
  serial   = each.value.serial_number
  node_id  = each.value.id
  pod_id   = each.value.pod
  role     = each.value.role
}

Debug Output

Panic Output

Expected Behavior

Upon destruction of the resource, the aci node should be removed from the fabric inventory.

Actual Behavior

A terraform apply reports the destruction of the resource, and the resource gets removed from the terraform state. However, the fabric node member is not removed from the fabric inventory.

2024-05-13T09:11:58.3126327Z module.tf-aci-setup[0].aci_fabric_node_member.fabricNodes["1136"]: Destruction complete after 0s

Steps to Reproduce

1. terraform plan
2. terraform apply

Important Factoids

The proper API call to remove the node from inventory is: https://<IP>/api/node/mo/uni/fabric/outofsvc.json

With body (example):

 {
  "fabricRsDecommissionNode": {
    "attributes": {
      "tDn": "topology/pod-1/node-1136",
      "status": "created,modified",
      "removeFromController": "true"
    }
  }
}

References

• 0000

[shrsr]

@edudppaz Thank you for opening the issue. Our team will look into this and get back to you.

[bardahlm]

Just a comment regarding use of aci_fabric_node_member. When one switch needed to be replaced we got into trouble as Terraform wanted to remove the old switch, which did no longer exist. After this we changed to use serial number as key instead of hostname. The next time we had to replace a switch we had no problems, just imported the new switch into state after adding it to the fabric using the gui.