Kubernetes NodePool

Creates and manages IonosCloud Kubernetes Node Pools.

Manages a Managed Kubernetes Node Pool, part of a managed Kubernetes cluster on IonosCloud.

Example Usage

resource "ionoscloud_datacenter" "example" {
  name                  = "Datacenter Example"
  location              = "us/las"
  description           = "datacenter description"
  sec_auth_protection   = false
}

resource "ionoscloud_lan" "example" {
  datacenter_id         = ionoscloud_datacenter.example.id
  public                = false
  name                  = "Lan Example"
  lifecycle {
    create_before_destroy = true
  }
}

resource "ionoscloud_ipblock" "example" {
  location              = "us/las"
  size                  = 3
  name                  = "IP Block Example"
}

resource "ionoscloud_k8s_cluster" "example" {
  name                  = "k8sClusterExample"
  k8s_version           = "1.28.6"
  maintenance_window {
    day_of_the_week     = "Sunday"
    time                = "09:00:00Z"
  }
  api_subnet_allow_list = ["1.2.3.4/32"]
  s3_buckets { 
     name               = <your_s3_bucket>
  }
}

resource "ionoscloud_k8s_node_pool" "example" {
  datacenter_id         = ionoscloud_datacenter.example.id
  k8s_cluster_id        = ionoscloud_k8s_cluster.example.id
  name                  = "k8sNodePoolExample"
  k8s_version           = ionoscloud_k8s_cluster.example.k8s_version
  maintenance_window {
    day_of_the_week     = "Monday"
    time                = "09:00:00Z"
  } 
  auto_scaling {
    min_node_count      = 1
    max_node_count      = 2
  }
  cpu_family            = "INTEL_XEON"
  availability_zone     = "AUTO"
  storage_type          = "SSD"
  node_count            = 1
  cores_count           = 2
  ram_size              = 2048
  storage_size          = 40
  public_ips            = [ ionoscloud_ipblock.example.ips[0], ionoscloud_ipblock.example.ips[1], ionoscloud_ipblock.example.ips[2] ]
  lans {
    id                  = ionoscloud_lan.example.id
    dhcp                = true
	routes {
       network          = "1.2.3.5/24"
       gateway_ip       = "10.1.5.17"
     }
   }  
  labels                = {
    lab1                = "value1"
    lab2                = "value2"
  }
  annotations           = {
    ann1                = "value1"
    ann2                = "value2"
  }
}

Note: Set create_before_destroy on the lan resource if you want to remove it from the nodepool during an update. This is to ensure that the nodepool is updated before the lan is destroyed.

Argument Reference

The following arguments are supported:

• name - (Required)[string] The name of the Kubernetes Cluster. This attribute is immutable.

• k8s_version - (Optional)[string] The desired Kubernetes Version. For supported values, please check the API documentation. Downgrades are not supported. The provider will ignore downgrades of patch level.

• auto_scaling - (Optional)[string] Wether the Node Pool should autoscale. For more details, please check the API documentation

  • min_node_count - (Optional)[int] The minimum number of worker nodes the node pool can scale down to. Should be less than max_node_count

  • max_node_count - (Optional)[int] The maximum number of worker nodes that the node pool can scale to. Should be greater than min_node_count

• lans - (Optional)[list] A list of numeric LAN id's you want this node pool to be part of. For more details, please check the API documentation, as well as the example above

  • id - (Required)[int] The LAN ID of an existing LAN at the related datacenter

  • dhcp - (Optional)[boolean] Indicates if the Kubernetes Node Pool LAN will reserve an IP using DHCP. Default value is true

  • routes - (Optional) An array of additional LANs attached to worker nodes

    • network - (Required)[string] IPv4 or IPv6 CIDR to be routed via the interface

    • gateway_ip - (Required)[string] IPv4 or IPv6 Gateway IP for the route

• maintenance_window - (Optional) See the maintenance_window section in the example above

  • time - (Required)[string] A clock time in the day when maintenance is allowed

  • day_of_the_week - (Required)[string] Day of the week when maintenance is allowed

• datacenter_id - (Required)[string] A Datacenter's UUID

• k8s_cluster_id- (Required)[string] A k8s cluster's UUID

• cpu_family - (Required)[string] The desired CPU Family - See the API documentation for more information. This attribute is immutable.

• availability_zone - (Required)[string] - The desired Compute availability zone - See the API documentation for more information. This attribute is immutable.

• storage_type -(Required)[string] - The desired storage type - SSD/HDD. This attribute is immutable.

• node_count -(Required)[int] - The desired number of nodes in the node pool

• cores_count -(Required)[int] - The CPU cores count for each node of the node pool. This attribute is immutable.

• ram_size -(Required)[int] - The desired amount of RAM, in MB. This attribute is immutable.

• storage_size -(Required)[int] - The size of the volume in GB. The size should be greater than 10GB. This attribute is immutable.

• public_ips - (Optional)[list] A list of public IPs associated with the node pool; must have at least node_count + 1 elements

• labels - (Optional)[map] A key/value map of labels

• annotations - (Optional)[map] A key/value map of annotations

• allow_replace - (Optional)[bool] When set to true, allows the update of immutable fields by first destroying and then re-creating the node pool.

⚠️ Warning: allow_replace - lets you update immutable fields, but it first destroys and then re-creates the node pool in order to do it. Set the field to true only if you know what you are doing. This will cause a downtime for all pods on that nodepool. Consider adding multiple nodepools and update one after the other for downtime free nodepool upgrade.

Immutable fields list: name, cpu_family, availability_zone, cores_count, ram_size, storage_size, storage_type.

⚠️ Note:

Be careful when using auto_scaling since the number of nodes can change. Because of that, when running terraform plan, Terraform will think that an update is required (since node_count from the tf plan will be different from the number of nodes set by the scheduler). To avoid that, you can use:

lifecycle {
    ignore_changes = [
      node_count
    ]
  }

This will also ignore the manual changes for node_count made in the tf plan. You can read more details about the ignore_changes attribute here.

Import

A Kubernetes Node Pool resource can be imported using its Kubernetes cluster's uuid as well as its own UUID, both of which you can retrieve from the cloud API: resource id, e.g.:

terraform import ionoscloud_k8s_node_pool.demo {k8s_cluster_uuid}/{k8s_nodepool_id}

This can be helpful when you want to import kubernetes node pools which you have already created manually or using other means, outside of terraform, towards the goal of managing them via Terraform

⚠️ Warning: **During a maintenance window, k8s can update your k8s_version if the old one reaches end of life. This upgrade will not be shown in the plan, as we prevent terraform from doing a downgrade, as downgrading k8s_version is not supported.

⚠️ Warning: If you are upgrading from v5.x.x to v6.x.x: You have to modify you plan for lans to match the new structure, by putting the ids from the old slice in lans.id fields. This is not backwards compatible.