Migrating a container results in it being recreated

[duntonr]

Describe the bug
Providor attempts to remake an existing container if the container was migrated to another node

To Reproduce
Requires a 2+ node cluster

1. Instantiate the provider using node A's IP, eg

provider "proxmox" {
  username = var.proxmox_username
  password = var.proxmox_password
  endpoint = "https://${local.proxmox_hosts.px01.ip}:8006/"
  insecure = true
}

2. Create a proxmox_virtual_environment_container resource on node A and ignore changes to the node_name on lifecycle

resource "proxmox_virtual_environment_container" "atlantis-container" {
  cpu {
    cores = 2
  }
  node_name   = "px01"
  description = local.cts.atlantis.description
  vm_id       = local.cts.atlantis.vm_id
  disk {
    datastore_id = local.infra.ceph_pool_01_name
    size         = 25
  }
  operating_system {
    template_file_id = "px01-dir:vztmpl/debian-12-standard_12.7-1_amd64.tar.zst"
    type             = "debian"
  }
  memory {
    dedicated = 1024
    swap      = 512
  }
  network_interface {
    name    = "eth0"
    bridge  = local.cts.atlantis.bridge
    enabled = "true"
  }
  pool_id = proxmox_virtual_environment_pool.lxc-util-pool.pool_id
  features {
    nesting = true
  }
  initialization {
    dns {
      domain  = local.infra.primary_search_domain
      servers = [local.infra.dns_piehole_ip, local.cts.atlantis.gateway, local.infra.dns_fallback_ip]
    }
    ip_config {
      ipv4 {
        address = "${local.cts.atlantis.ip}/24"
        gateway = local.cts.atlantis.gateway
      }
    }
    hostname = local.cts.atlantis.hostname
    user_account {
      keys     = [var.ssh_public_key]
      password = var.root_password
    }
  }

  lifecycle {
    ignore_changes = [
      node_name,
      operating_system,
      disk["datastore_id"]
    ]
  }
}

3. Run tofu apply to bring the container up
4. manually migrate the container from Node A to Node B
5. Run tofu plan again and note it wants to recreate the resource
   

Expected behavior
No new container creation as the container was already created, it just exists on a different node

Screenshots
If applicable, add screenshots to help explain your problem.

Additional context
The bug seems to be that the provider will get the old node_name from tfstate and only check that node to see if the container is present. If not, it will try to recreate it on that node. With the ignore_changes on the node_name field set, it doesn't matter what goes in the tf application itsself, the provider will pull the node to check from the state and only check that node. I confirmed this by manually editing the state to contain the new node and a create was NOT triggered with a new plan.

With normal migrations, the container could be on any node in the cluster, so it may be more robust to see if the vm_id for the container already exists on any node in the cluster when deciding if a new create is needed or not

• Clustered Proxmox
• Proxmox version: 8.2.7
• Provider version (ideally it should be the latest version): 0.66.3
• Terraform/OpenTofu version: tofu 1.8.6
• OS (where you run Terraform/OpenTofu from): ubuntu
• Debug logs (TF_LOG=DEBUG terraform apply):

[duntonr]

This looks similar to #658 but with the addition of use of ignore_changes

[bpg]

Yes seems to be the same as #658, so I'm going to close this one.