feat: ✨ add support for multiple worker_node types and labels

README.md

@@ -204,6 +204,64 @@ module "talos" {
 }
 ```
 
+### Mixed Worker Node Types
+
+For more advanced use cases, you can define different types of worker nodes with individual configurations using the `worker_nodes` variable:
+
+```hcl
+module "talos" {
+  source  = "hcloud-talos/talos/hcloud"
+  version = "<latest-version>"
+
+  talos_version      = "v1.10.3"
+  kubernetes_version = "1.30.3"
+
+  hcloud_token            = "your-hcloud-token"
+  firewall_use_current_ip = true
+
+  cluster_name    = "mixed-cluster"
+  datacenter_name = "fsn1-dc14"
+
+  control_plane_count       = 1
+  control_plane_server_type = "cx22"
+
+  # Define different worker node types
+  worker_nodes = [
+    # Standard x86 workers
+    {
+      type  = "cx22"
+      labels = {
+        "node.kubernetes.io/instance-type" = "cx22"
+      }
+    },
+    # ARM workers for specific workloads with taints
+    {
+      type   = "cax22"
+      labels = {
+        "node.kubernetes.io/arch"          = "arm64"
+        "affinity.example.com" = "example"
+      }
+      taints = [
+        {
+          key    = "arm64-only"
+          value  = "true"
+          effect = "NoSchedule"
+        }
+      ]
+    }
+  ]
+}
+```
+
+> [!NOTE]
+> The `worker_nodes` variable allows you to:
+> - Mix different server types (x86 and ARM)
+> - Add custom labels to nodes
+> - Apply taints for workload isolation
+> - Control the count of each node type independently
+> 
+> The legacy `worker_count` and `worker_server_type` variables are still supported for backward compatibility but are deprecated in favor of `worker_nodes`.
+
 You need to pipe the outputs of the module:
 
 ```hcl

network.tf

@@ -84,7 +84,7 @@ resource "hcloud_primary_ip" "control_plane_ipv6" {
 }
 
 resource "hcloud_primary_ip" "worker_ipv4" {
-  count         = var.worker_count
+  count         = local.total_worker_count
   name          = "${local.cluster_prefix}worker-${count.index + 1}-ipv4"
   datacenter    = data.hcloud_datacenter.this.name
   type          = "ipv4"
@@ -97,7 +97,7 @@ resource "hcloud_primary_ip" "worker_ipv4" {
 }
 
 resource "hcloud_primary_ip" "worker_ipv6" {
-  count         = var.enable_ipv6 ? var.worker_count > 0 ? var.worker_count : 1 : 0
+  count         = var.enable_ipv6 ? local.total_worker_count > 0 ? local.total_worker_count : 1 : 0
   name          = "${local.cluster_prefix}worker-${count.index + 1}-ipv6"
   datacenter    = data.hcloud_datacenter.this.name
   type          = "ipv6"
@@ -139,6 +139,6 @@ locals {
     for index in range(var.control_plane_count > 0 ? var.control_plane_count : 1) : cidrhost(hcloud_network_subnet.nodes.ip_range, index + 101)
   ]
   worker_private_ipv4_list = [
-    for index in range(var.worker_count > 0 ? var.worker_count : 1) : cidrhost(hcloud_network_subnet.nodes.ip_range, index + 201)
+    for index in range(local.total_worker_count > 0 ? local.total_worker_count : 1) : cidrhost(hcloud_network_subnet.nodes.ip_range, index + 201)
   ]
 }

outputs.tf

@@ -9,7 +9,8 @@ output "kubeconfig" {
 }
 
 output "talos_client_configuration" {
-  value = data.talos_client_configuration.this
+  value     = data.talos_client_configuration.this
+  sensitive = true
 }
 
 output "talos_machine_configurations_control_plane" {
@@ -40,8 +41,9 @@ output "hetzner_network_id" {
 
 output "talos_worker_ids" {
   description = "Server IDs of the hetzner talos workers machines"
-  value = {
-    for id, server in hcloud_server.workers : id => server.id
-  }
+  value = merge(
+    { for id, server in hcloud_server.workers_new : id => server.id },
+    { for id, server in hcloud_server.workers : id => server.id }
+  )
 }
 

server.tf

@@ -19,14 +19,57 @@ locals {
     (var.disable_arm ? null : data.hcloud_image.arm[0].id) : // Use ARM image if not disabled
     (var.disable_x86 ? null : data.hcloud_image.x86[0].id)   // Use x86 image if not disabled
   )
-  worker_image_id = (
-    var.worker_count > 0 ? # Only calculate if workers exist
-    (
-      substr(var.worker_server_type, 0, 3) == "cax" ?
-      (var.disable_arm ? null : data.hcloud_image.arm[0].id) : // Use ARM image if not disabled
-      (var.disable_x86 ? null : data.hcloud_image.x86[0].id)   // Use x86 image if not disabled
-    ) : null                                                   # No workers, no image needed
-  )
+
+  # Calculate total worker count from both old and new variables
+  legacy_worker_count = var.worker_count
+  new_worker_count    = length(var.worker_nodes)
+  total_worker_count  = local.legacy_worker_count + local.new_worker_count
+
+  # Generate worker node configurations from both old and new variables
+  legacy_workers = var.worker_count > 0 ? [
+    for i in range(var.worker_count) : {
+      index       = i
+      name        = "${local.cluster_prefix}worker-${i + 1}"
+      server_type = var.worker_server_type
+      image_id = (
+        substr(var.worker_server_type, 0, 3) == "cax" ?
+        (var.disable_arm ? null : data.hcloud_image.arm[0].id) :
+        (var.disable_x86 ? null : data.hcloud_image.x86[0].id)
+      )
+      ipv4_public         = local.worker_public_ipv4_list[i]
+      ipv6_public         = var.enable_ipv6 ? local.worker_public_ipv6_list[i] : null
+      ipv6_public_subnet  = var.enable_ipv6 ? local.worker_public_ipv6_subnet_list[i] : null
+      ipv4_private        = local.worker_private_ipv4_list[i]
+      labels              = {}
+      taints              = []
+      node_group_index    = 0
+      node_in_group_index = i
+    }
+  ] : []
+
+
+  new_workers = [
+    for i, worker in var.worker_nodes : {
+      index       = local.legacy_worker_count + i
+      name        = "${local.cluster_prefix}worker-${local.legacy_worker_count + i + 1}"
+      server_type = worker.type
+      image_id = (
+        substr(worker.type, 0, 3) == "cax" ?
+        (var.disable_arm ? null : data.hcloud_image.arm[0].id) :
+        (var.disable_x86 ? null : data.hcloud_image.x86[0].id)
+      )
+      ipv4_public        = local.worker_public_ipv4_list[local.legacy_worker_count + i]
+      ipv6_public        = var.enable_ipv6 ? local.worker_public_ipv6_list[local.legacy_worker_count + i] : null
+      ipv6_public_subnet = var.enable_ipv6 ? local.worker_public_ipv6_subnet_list[local.legacy_worker_count + i] : null
+      ipv4_private       = local.worker_private_ipv4_list[local.legacy_worker_count + i]
+      labels             = worker.labels
+      taints             = worker.taints
+    }
+  ]
+
+  # Combine legacy and new workers
+  workers = concat(local.legacy_workers, local.new_workers)
+
   control_planes = [
     for i in range(var.control_plane_count) : {
       index              = i
@@ -37,16 +80,6 @@ locals {
       ipv4_private       = local.control_plane_private_ipv4_list[i]
     }
   ]
-  workers = [
-    for i in range(var.worker_count) : {
-      index              = i
-      name               = "${local.cluster_prefix}worker-${i + 1}"
-      ipv4_public        = local.worker_public_ipv4_list[i],
-      ipv6_public        = var.enable_ipv6 ? local.worker_public_ipv6_list[i] : null
-      ipv6_public_subnet = var.enable_ipv6 ? local.worker_public_ipv6_subnet_list[i] : null
-      ipv4_private       = local.worker_private_ipv4_list[i]
-    }
-  ]
 }
 
 resource "tls_private_key" "ssh_key" {
@@ -108,20 +141,69 @@ resource "hcloud_server" "control_planes" {
 }
 
 resource "hcloud_server" "workers" {
-  for_each           = { for worker in local.workers : worker.name => worker }
+  for_each           = { for worker in local.legacy_workers : worker.name => worker }
   datacenter         = data.hcloud_datacenter.this.name
   name               = each.value.name
-  image              = local.worker_image_id
-  server_type        = var.worker_server_type
+  image              = each.value.image_id
+  server_type        = each.value.server_type
   user_data          = data.talos_machine_configuration.worker[each.value.name].machine_configuration
   ssh_keys           = [hcloud_ssh_key.this.id]
   placement_group_id = hcloud_placement_group.worker.id
 
-  labels = {
-    "cluster" = var.cluster_name,
-    "role"    = "worker"
+  labels = merge({
+    "cluster"     = var.cluster_name,
+    "role"        = "worker"
+    "server_type" = each.value.server_type
+  }, each.value.labels)
+
+  firewall_ids = [
+    hcloud_firewall.this.id
+  ]
+
+  public_net {
+    ipv4_enabled = true
+    ipv4         = hcloud_primary_ip.worker_ipv4[each.value.index].id
+    ipv6_enabled = var.enable_ipv6
+    ipv6         = var.enable_ipv6 ? hcloud_primary_ip.worker_ipv6[each.value.index].id : null
   }
 
+  network {
+    network_id = hcloud_network_subnet.nodes.network_id
+    ip         = each.value.ipv4_private
+    alias_ips  = [] # fix for https://github.com/hetznercloud/terraform-provider-hcloud/issues/650
+  }
+
+  depends_on = [
+    hcloud_network_subnet.nodes,
+    data.talos_machine_configuration.worker
+  ]
+
+  lifecycle {
+    ignore_changes = [
+      user_data,
+      image
+    ]
+  }
+}
+
+
+
+resource "hcloud_server" "workers_new" {
+  for_each           = { for worker in local.new_workers : worker.name => worker }
+  datacenter         = data.hcloud_datacenter.this.name
+  name               = each.value.name
+  image              = each.value.image_id
+  server_type        = each.value.server_type
+  user_data          = data.talos_machine_configuration.worker[each.value.name].machine_configuration
+  ssh_keys           = [hcloud_ssh_key.this.id]
+  placement_group_id = hcloud_placement_group.worker.id
+
+  labels = merge({
+    "cluster"     = var.cluster_name,
+    "role"        = "worker"
+    "server_type" = each.value.server_type
+  }, each.value.labels)
+
   firewall_ids = [
     hcloud_firewall.this.id
   ]

talos.tf

@@ -103,20 +103,7 @@ data "talos_machine_configuration" "dummy_control_plane" {
   examples           = false
 }
 
-# Dummy configuration generated when worker_count is 0 for debugging purposes
-# tflint-ignore: terraform_unused_declarations
-data "talos_machine_configuration" "dummy_worker" {
-  count              = var.worker_count == 0 ? 1 : 0
-  talos_version      = var.talos_version
-  cluster_name       = var.cluster_name
-  cluster_endpoint   = local.cluster_endpoint_url_internal # Uses dummy endpoint when count is 0
-  kubernetes_version = var.kubernetes_version
-  machine_type       = "worker"
-  machine_secrets    = talos_machine_secrets.this.machine_secrets
-  config_patches     = concat([yamlencode(local.worker_yaml["dummy-worker-0"])], var.talos_worker_extra_config_patches) # Use dummy yaml + extra patches
-  docs               = false
-  examples           = false
-}
+
 
 resource "talos_machine_bootstrap" "this" {
   count                = var.control_plane_count > 0 ? 1 : 0

talos_patch_worker.tf

@@ -1,15 +1,22 @@
 locals {
   # Define a dummy worker entry for when count is 0
-  dummy_workers = var.worker_count == 0 ? [{
-    index              = 0
-    name               = "dummy-worker-0"
-    ipv4_public        = "0.0.0.0"                           # Fallback
-    ipv6_public        = null                                # Fallback
-    ipv6_public_subnet = null                                # Fallback
-    ipv4_private       = cidrhost(local.node_ipv4_cidr, 200) # Use a predictable dummy private IP
+  dummy_workers = local.total_worker_count == 0 ? [{
+    index               = 0
+    name                = "dummy-worker-0"
+    server_type         = "cx11"
+    image_id            = null
+    ipv4_public         = "0.0.0.0"                           # Fallback
+    ipv6_public         = null                                # Fallback
+    ipv6_public_subnet  = null                                # Fallback
+    ipv4_private        = cidrhost(local.node_ipv4_cidr, 200) # Use a predictable dummy private IP
+    labels              = {}
+    taints              = []
+    node_group_index    = 0
+    node_in_group_index = 0
   }] : []
 
-  # Combine real and dummy workers
+  # Combine real and dummy workers - always include dummy when no workers exist
+  #  merged_workers = local.total_worker_count == 0 ? local.dummy_workers : local.workers
   merged_workers = concat(local.workers, local.dummy_workers)
 
   # Generate YAML for all (real or dummy) workers
@@ -23,20 +30,34 @@ locals {
           ]
         }
         certSANs = local.cert_SANs
-        kubelet = {
-          extraArgs = merge(
-            {
-              "cloud-provider"             = "external"
-              "rotate-server-certificates" = true
-            },
-            var.kubelet_extra_args
-          )
-          nodeIP = {
-            validSubnets = [
-              local.node_ipv4_cidr
-            ]
-          }
-        }
+        kubelet = merge(
+          {
+            extraArgs = merge(
+              {
+                "cloud-provider"             = "external"
+                "rotate-server-certificates" = true
+              },
+              var.kubelet_extra_args
+            )
+            nodeIP = {
+              validSubnets = [
+                local.node_ipv4_cidr
+              ]
+            }
+          },
+          # Add registerWithTaints if taints are defined
+          length(worker.taints) > 0 ? {
+            extraConfig = {
+              registerWithTaints = [
+                for taint in worker.taints : {
+                  key    = taint.key
+                  value  = taint.value
+                  effect = taint.effect
+                }
+              ]
+            }
+          } : {}
+        )
         network = {
           extraHostEntries = local.extra_host_entries
           kubespan = {
@@ -70,6 +91,7 @@ locals {
             "time.cloudflare.com"
           ]
         }
+        nodeLabels = worker.labels
         registries = var.registries
       }
       cluster = {