Terraform Provider for VMware vSphere

The vsphere_resource_pool resource allows you to create and manage resource pools on DRS-enabled vSphere clusters or standalone ESXi hosts. You can use these pools to partition CPU and memory resources for virtual machines or nested resource pools.

Key Concepts

• Parenting: Every resource pool must have a parent_resource_pool_id. This can be the root resource pool of a cluster, a standalone host, or another resource pool (for nesting).
• Moving Pools: When moving a resource pool from one parent to another, both the old and new parents must share a common root resource pool.
• Targeting VMs: To assign a virtual machine to a specific resource pool, set the resource_pool_id argument within the vsphere_virtual_machine resource.

resource "vsphere_resource_pool" "resource_pool" {
  name                    = "resource-pool-01"
  parent_resource_pool_id = data.vsphere_compute_cluster.compute_cluster.resource_pool_id
}

resource "vsphere_virtual_machine" "vm" {
  # ... other configuration ...
  resource_pool_id = data.vsphere_compute_cluster.cluster.resource_pool_id
  # ... other configuration ...
}

The vsphere_compute_cluster_vm_group resource manages groups of virtual machines within a specific vSphere cluster. These groups are primarily used as inputs for defining rules in the cluster, such as vsphere_compute_cluster_vm_dependency_rule and vsphere_compute_cluster_vm_host_rule resources.

Important Requirements:

• This resource requires a vCenter connection; it is not available for direct ESXi connections.
• The name argument shares a namespace with vsphere_compute_cluster_host_group. Ensure names are unique across both resource types to avoid conflicts.
• To update an existing VM group, you must first import it using the terraform import command.

resource "vsphere_compute_cluster_vm_group" "cluster_vm_group" {
  name                = "terraform-test-cluster-vm-group"
  compute_cluster_id  = data.vsphere_compute_cluster.cluster.id
  virtual_machine_ids = ["${vsphere_virtual_machine.vm.*.id}"]
}

For templates originating from OVF/OVA files, you can supply configuration parameters using the vapp block. This uses the properties map to set GuestInfo or ISO transport parameters.

Important Usage Notes:

• Boolean Values: While Terraform uses bool (true/false), vSphere OVF properties often require the strings "True" or "False" (title case). It is recommended to define these as strings in your Terraform variables to ensure correct formatting.
• Compatibility: You can only set existing user-configurable keys that were present in the original template. You cannot set properties on VMs created from scratch or those lacking vApp configuration.

variable "ssh_enabled" {
  type        = string
  description = "Enable SSH on the virtual appliance. One of `True` or `False`."
  default     = "False"
}

resource "vsphere_virtual_machine" "vm" {
  # ... other configurations ...
  vapp {
    properties = {
      "ssh_enabled" = var.ssh_enabled
    }
  }
}

The vsphere_virtual_machine resource supports vSphere Storage DRS by allowing you to assign a virtual machine to a datastore cluster instead of a single datastore. Use the datastore_cluster_id argument to achieve this.

Important Note on Race Conditions: When managing datastore clusters, member datastores, and virtual machines in the same configuration, a race condition can occur. The vsphere_virtual_machine resource does not have an implicit dependency on the individual datastores within a cluster. To prevent errors, use the depends_on meta-argument to create an explicit dependency on the datastores in the cluster, or manage the datastore cluster and its datastores in a separate Terraform configuration.

data "vsphere_datastore_cluster" "datastore_cluster" {
  name          = "datastore-cluster-01"
  datacenter_id = data.vsphere_datacenter.datacenter.id
}

resource "vsphere_virtual_machine" "vm" {
  name                 = "foo"
  resource_pool_id     = data.vsphere_compute_cluster.cluster.resource_pool_id
  datastore_cluster_id = data.vsphere_datastore_cluster.datastore_cluster.id
  num_cpus             = 1
  memory               = 1024
  guest_id             = "otherLinux64Guest"
  network_interface {
    network_id = data.vsphere_network.network.id
  }
  disk {
    label = "Hard Disk 1"
    size  = 20
  }
}