chore: work done, not yet working

.gitignore

@@ -37,3 +37,4 @@ terraform.rc
 
 # ENV
 .env
+admin.*config

k8s.tjo.cloud/dashboard.tf

@@ -12,10 +12,6 @@ resource "helm_release" "dashboard" {
 }
 
 resource "kubernetes_manifest" "dashoard-http-route" {
-  depends_on = [
-    kubernetes_manifest.gateway,
-  ]
-
   manifest = {
     apiVersion = "gateway.networking.k8s.io/v1"
     kind       = "HTTPRoute"
@@ -26,7 +22,7 @@ resource "kubernetes_manifest" "dashoard-http-route" {
     spec = {
       parentRefs = [
         {
-          name : "gateway"
+          name : kubernetes_manifest.gateway.object.metadata.name
         }
       ]
       hostnames = [
@@ -120,7 +116,7 @@ resource "kubernetes_manifest" "dashboard-oidc" {
       targetRef = {
         group : "gateway.networking.k8s.io"
         kind : "HTTPRoute"
-        name : "dashboard"
+        name : kubernetes_manifest.dashoard-http-route.object.metadata.name
       }
       oidc = {
         provider = {
@@ -128,7 +124,7 @@ resource "kubernetes_manifest" "dashboard-oidc" {
         }
         clientID : var.oidc_client_id
         clientSecret : {
-          name : "dashboard-oidc"
+          name : kubernetes_secret.dashboard-oidc.metadata[0].name
         }
         scopes : ["openid", "email", "profile"]
         forwardAccessToken : true

k8s.tjo.cloud/gateway.tf

@@ -9,6 +9,8 @@ resource "kubernetes_secret" "digitalocean-token" {
 }
 
 resource "helm_release" "cert-manager" {
+  depends_on = [helm_release.envoy]
+
   name       = "cert-manager"
   chart      = "cert-manager"
   repository = "https://charts.jetstack.io"
@@ -27,6 +29,8 @@ resource "helm_release" "cert-manager" {
 }
 
 resource "kubernetes_manifest" "tjo-cloud-issuer" {
+  depends_on = [helm_release.cert-manager]
+
   manifest = {
     apiVersion = "cert-manager.io/v1"
     kind       = "Issuer"
@@ -90,7 +94,9 @@ resource "helm_release" "envoy" {
   ]
 }
 
-resource "kubernetes_manifest" "gateway-class" {
+resource "kubernetes_manifest" "gateway_class" {
+  depends_on = [helm_release.envoy]
+
   manifest = {
     apiVersion = "gateway.networking.k8s.io/v1"
     kind       = "GatewayClass"
@@ -115,7 +121,7 @@ resource "kubernetes_manifest" "gateway" {
       }
     }
     spec = {
-      gatewayClassName = "envoy"
+      gatewayClassName = kubernetes_manifest.gateway_class.object.metadata.name
       listeners = [
         {
           name : "http"

k8s.tjo.cloud/kubeconfig

@@ -0,0 +1,26 @@
+apiVersion: v1
+kind: Config
+clusters:
+- name: tjo-cloud
+  cluster:
+    server: https://api.k8s.tjo.cloud:6443
+    certificate-authority-data: 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
+contexts:
+- name: oidc@tjo-cloud
+  context:
+    cluster: tjo-cloud
+    namespace: default
+    user: oidc
+current-context: oidc@tjo-cloud
+users:
+- name: oidc
+  user:
+    exec:
+      apiVersion: client.authentication.k8s.io/v1beta1
+      command: kubectl
+      args:
+      - oidc-login
+      - get-token
+      - --oidc-issuer-url=https://id.tjo.space/application/o/k8stjocloud/
+      - --oidc-client-id=HAI6rW0EWtgmSPGKAJ3XXzubQTUut2GMeTRS2spg
+      - --oidc-extra-scope=profile

k8s.tjo.cloud/main.tf

@@ -22,7 +22,7 @@ module "cluster" {
   proxmox = {
     name           = "tjo-cloud"
     url            = "https://proxmox.tjo.cloud/api2/json"
-    iso_storage_id = "proxmox-backup-tjo-cloud"
+    common_storage = "proxmox-backup-tjo-cloud"
   }
 
   tailscale_authkey = var.tailscale_authkey
@@ -30,52 +30,52 @@ module "cluster" {
   allow_scheduling_on_control_planes = true
   nodes = {
     pink = {
-      public    = true
-      type      = "controlplane"
-      host      = "hetzner"
-      boot_pool = "hetzner-main-data"
-      cores     = 4
-      memory    = 4096
+      public  = true
+      type    = "controlplane"
+      host    = "hetzner"
+      storage = "local-zfs"
+      cores   = 4
+      memory  = 4096
     }
     purple = {
-      public    = true
-      type      = "controlplane"
-      host      = "hetzner"
-      boot_pool = "hetzner-main-data"
-      cores     = 4
-      memory    = 4096
+      public  = true
+      type    = "controlplane"
+      host    = "hetzner"
+      storage = "local-zfs"
+      cores   = 4
+      memory  = 4096
     }
     violet = {
-      public    = true
-      type      = "controlplane"
-      host      = "hetzner"
-      boot_pool = "hetzner-main-data"
-      cores     = 4
-      memory    = 4096
+      public  = true
+      type    = "controlplane"
+      host    = "hetzner"
+      storage = "local-zfs"
+      cores   = 4
+      memory  = 4096
     }
     blue = {
-      public    = false
-      type      = "worker"
-      host      = "hetzner"
-      boot_pool = "hetzner-main-data"
-      cores     = 4
-      memory    = 16384
+      public  = false
+      type    = "worker"
+      host    = "hetzner"
+      storage = "local-zfs"
+      cores   = 4
+      memory  = 16384
     }
     cyan = {
-      public    = false
-      type      = "worker"
-      host      = "hetzner"
-      boot_pool = "hetzner-main-data"
-      cores     = 4
-      memory    = 16384
+      public  = false
+      type    = "worker"
+      host    = "hetzner"
+      storage = "local-zfs"
+      cores   = 4
+      memory  = 16384
     }
     green = {
-      public    = false
-      type      = "worker"
-      host      = "hetzner"
-      boot_pool = "hetzner-main-data"
-      cores     = 4
-      memory    = 16384
+      public  = false
+      type    = "worker"
+      host    = "hetzner"
+      storage = "local-zfs"
+      cores   = 4
+      memory  = 16384
     }
   }
 }

k8s.tjo.cloud/providers.tf

@@ -20,10 +20,6 @@ terraform {
       source  = "hashicorp/random"
       version = "3.6.2"
     }
-    macaddress = {
-      source  = "ivoronin/macaddress"
-      version = "0.3.2"
-    }
     helm = {
       source  = "hashicorp/helm"
       version = "2.14.0"

modules/cluster/main.tf

@@ -21,9 +21,8 @@ locals {
   talos_controlplane_config = {
     cluster : {
       etcd : {
-        #advertisedSubnets : [
-        #  local.tailscaleSubnet
-        #]
+        advertisedSubnets : local.tailscaleSubnets
+        listenSubnets : local.tailscaleSubnets
       }
       allowSchedulingOnControlPlanes : var.allow_scheduling_on_control_planes,
       apiServer : {
@@ -67,11 +66,8 @@ locals {
           contents : data.helm_template.ccm.manifest
         }
       ]
-      externalCloudProvider : {
-        enabled : true
-      }
       extraManifests : [
-        #"https://raw.githubusercontent.com/alex1989hu/kubelet-serving-cert-approver/v0.8.5/deploy/standalone-install.yaml",
+        "https://raw.githubusercontent.com/alex1989hu/kubelet-serving-cert-approver/v0.8.5/deploy/standalone-install.yaml",
         "https://github.com/kubernetes-sigs/metrics-server/releases/download/v0.7.1/components.yaml",
       ]
     }
@@ -79,6 +75,9 @@ locals {
 
   talos_worker_config = {
     cluster : {
+      #externalCloudProvider : {
+      #  enabled : true
+      #}
       controlPlane : {
         endpoint : local.cluster_endpoint
         localAPIServerPort : var.cluster.api.port
@@ -101,6 +100,7 @@ locals {
         }
         extraArgs : {
           rotate-server-certificates : true
+          #cloud-provider : "external"
         }
       }
       install = {
@@ -109,6 +109,33 @@ locals {
       }
     }
   }
+
+  talos_node_config = {
+    for k, node in local.nodes_with_address : k => [
+      yamlencode({
+        machine = {
+          network = {
+            hostname = node.name
+          }
+          nodeLabels = {
+            "k8s.tjo.cloud/public"  = node.public ? "true" : "false"
+            "k8s.tjo.cloud/host"    = node.host
+            "k8s.tjo.cloud/proxmox" = var.proxmox.name
+          }
+        }
+      }),
+      yamlencode(
+        {
+          apiVersion : "v1alpha1"
+          kind : "ExtensionServiceConfig"
+          name : "tailscale"
+          environment : [
+            "TS_AUTHKEY=${var.tailscale_authkey}",
+            "TS_HOSTNAME=${node.name}",
+          ]
+      })
+    ]
+  }
 }
 
 resource "digitalocean_record" "controlplane-A" {
@@ -131,7 +158,9 @@ resource "digitalocean_record" "controlplane-AAAA" {
   ttl    = 30
 }
 
-resource "talos_machine_secrets" "this" {}
+resource "talos_machine_secrets" "this" {
+  talos_version = var.talos.version
+}
 
 data "talos_machine_configuration" "controlplane" {
   cluster_name     = var.cluster.name
@@ -262,14 +291,30 @@ data "helm_template" "csi" {
         - url: ${var.proxmox.url}
           insecure: ${var.proxmox.insecure}
           token_id: "${proxmox_virtual_environment_user_token.csi.id}"
-          token_secret: "${proxmox_virtual_environment_user_token.csi.value}"
+          token_secret: "${split("=", proxmox_virtual_environment_user_token.csi.value)[1]}"
           region: "${var.proxmox.name}"
 
     storageClass:
-      - name: default
+      - name: proxmox
         storage: local-storage
         reclaimPolicy: Delete
-        fstype: xfs
+        fstype: ext4
+        ssd: true
+        cache: none
+
+    replicaCount: 1
+
+    nodeSelector:
+      node-role.kubernetes.io/control-plane: ""
+      node.cloudprovider.kubernetes.io/platform: nocloud
+    tolerations:
+      - key: node-role.kubernetes.io/control-plane
+        effect: NoSchedule
+    node:
+      nodeSelector:
+        node.cloudprovider.kubernetes.io/platform: nocloud
+      tolerations:
+        - operator: Exists
   EOF
   ]
 }
@@ -285,13 +330,22 @@ data "helm_template" "ccm" {
   kube_version = var.talos.kubernetes
 
   values = [<<-EOF
+    affinity:
+      nodeAffinity:
+        requiredDuringSchedulingIgnoredDuringExecution:
+          nodeSelectorTerms:
+          - matchExpressions:
+            - key: node-role.kubernetes.io/control-plane
+              operator: Exists
+    enabledControllers:
+      - cloud-node-lifecycle
     config:
       clusters:
         - url: ${var.proxmox.url}
           insecure: ${var.proxmox.insecure}
-          token_id: "${proxmox_virtual_environment_user_token.ccm.id}"
-          token_secret: "${proxmox_virtual_environment_user_token.ccm.value}"
-          region: "${var.proxmox.name}"
+          token_id: ${proxmox_virtual_environment_user_token.ccm.id}
+          token_secret: ${split("=", proxmox_virtual_environment_user_token.ccm.value)[1]}
+          region: ${var.proxmox.name}
   EOF
   ]
 }
@@ -307,31 +361,13 @@ resource "talos_machine_configuration_apply" "controlplane" {
 
   apply_mode = "reboot"
 
-  config_patches = [
-    yamlencode(merge(local.talos_worker_config, local.talos_controlplane_config, {
-      machine = {
-        network = {
-          hostname = each.value.name
-        }
-        nodeLabels = {
-          "k8s.tjo.cloud/public"  = each.value.public ? "true" : "false"
-          "k8s.tjo.cloud/host"    = each.value.host
-          "k8s.tjo.cloud/proxmox" = var.proxmox.name
-        }
-      }
-
-    })),
-    yamlencode(
-      {
-        apiVersion : "v1alpha1"
-        kind : "ExtensionServiceConfig"
-        name : "tailscale"
-        environment : [
-          "TS_AUTHKEY=${var.tailscale_authkey}",
-          "TS_HOSTNAME=${each.value.name}",
-        ]
-    })
-  ]
+  config_patches = concat(
+    [
+      yamlencode(local.talos_worker_config),
+      yamlencode(local.talos_controlplane_config)
+    ],
+    local.talos_node_config[each.key]
+  )
 }
 
 resource "talos_machine_configuration_apply" "worker" {
@@ -345,30 +381,12 @@ resource "talos_machine_configuration_apply" "worker" {
 
   apply_mode = "reboot"
 
-  config_patches = [
-    yamlencode(merge(local.talos_worker_config, {
-      machine = {
-        network = {
-          hostname = each.value.name
-        }
-        nodeLabels = {
-          "k8s.tjo.cloud/public"  = each.value.public ? "true" : "false"
-          "k8s.tjo.cloud/host"    = each.value.host
-          "k8s.tjo.cloud/proxmox" = var.proxmox.name
-        }
-      }
-    })),
-    yamlencode(
-      {
-        apiVersion : "v1alpha1"
-        kind : "ExtensionServiceConfig"
-        name : "tailscale"
-        environment : [
-          "TS_AUTHKEY=${var.tailscale_authkey}",
-          "TS_HOSTNAME=${each.value.name}",
-        ]
-    })
-  ]
+  config_patches = concat(
+    [
+      yamlencode(local.talos_worker_config)
+    ],
+    local.talos_node_config[each.key]
+  )
 }
 
 resource "talos_machine_bootstrap" "this" {
@@ -390,3 +408,23 @@ data "talos_cluster_kubeconfig" "this" {
   client_configuration = talos_machine_secrets.this.client_configuration
   node                 = local.first_controlplane_node.ipv4
 }
+
+resource "local_file" "kubeconfig" {
+  content  = data.talos_cluster_kubeconfig.this.kubeconfig_raw
+  filename = "${path.root}/admin.kubeconfig"
+}
+
+data "talos_client_configuration" "this" {
+  count = length(values({ for k, v in local.nodes_with_address : k => v if v.type == "controlplane" })) > 0 ? 1 : 0
+
+  cluster_name         = var.cluster.name
+  client_configuration = talos_machine_secrets.this.client_configuration
+  endpoints            = values({ for k, v in local.nodes_with_address : k => v if v.type == "controlplane" })[*].ipv4
+}
+
+resource "local_file" "talosconfig" {
+  count = length(values({ for k, v in local.nodes : k => v if v.type == "controlplane" })) > 0 ? 1 : 0
+
+  content  = nonsensitive(data.talos_client_configuration.this[0].talos_config)
+  filename = "${path.root}/admin.talosconfig"
+}

modules/cluster/proxmox.tf

@@ -1,6 +1,18 @@
 locals {
-  nodes = { for k, v in var.nodes : k => merge(v, { name = replace("${k}.${v.type}.${var.cluster.domain}", ".", "-") }) }
-
+  nodes_with_names = {
+    for k, v in var.nodes : k => merge(v, {
+      id   = 1000 + index(keys(var.nodes), k)
+      name = replace("${k}.${v.type}.${var.cluster.domain}", ".", "-")
+    })
+  }
+  hashes = {
+    for k, v in local.nodes_with_names : k => sha1("${v.name}:${var.cluster.name}")
+  }
+  nodes = {
+    for k, v in local.nodes_with_names : k => merge(v, {
+      mac_address = "AA:BB:CC:DD:${format("%v:%v", substr(local.hashes[k], 0, 2), substr(local.hashes[k], 2, 2))}"
+    })
+  }
 
 
   first_controlplane_node = values({ for k, v in local.nodes_with_address : k => v if v.type == "controlplane" })[0]
@@ -21,49 +33,44 @@ locals {
   nodes_with_address = {
     for k, v in local.nodes :
     k => merge(v, {
-      ipv4 = try(local.ipv4_addresses[k]["eth0"][0], null)
-      ipv6 = try(local.ipv6_addresses[k]["eth0"][0], null)
+      ipv4 = local.ipv4_addresses[k]["eth0"][0]
+      ipv6 = local.ipv6_addresses[k]["eth0"][0]
     })
   }
 }
 
 resource "proxmox_virtual_environment_download_file" "talos" {
   content_type = "iso"
-  datastore_id = var.proxmox.iso_storage_id
+  datastore_id = var.proxmox.common_storage
   node_name    = values(var.nodes)[0].host
-  file_name    = "talos-${var.talos.version}-amd64.iso"
+  file_name    = "talos-${var.talos.schematic_id}-${var.talos.version}-amd64.iso"
   url          = "https://factory.talos.dev/image/${var.talos.schematic_id}/${var.talos.version}/nocloud-amd64.iso"
 }
 
-resource "proxmox_virtual_environment_file" "controlplane" {
-  for_each     = { for k, v in local.nodes_with_address : k => v if v.type == "controlplane" }
+resource "proxmox_virtual_environment_file" "metadata" {
+  for_each = local.nodes
+
   node_name    = each.value.host
   content_type = "snippets"
-  datastore_id = each.value.boot_pool
+  datastore_id = var.proxmox.common_storage
 
   source_raw {
     data      = <<-EOF
       hostname: ${each.value.name}
-      instance-id: 1000
-      instance-type: ${each.value.cpu}VCPU-${floor(each.value.memory / 1024)}GB
-      provider-id: "proxmox://${var.proxmox.name}/1000"
-      region: ${var.proxmox.name}
+      id: ${each.value.id}
+      providerID: proxmox://${var.proxmox.name}/${each.value.id}
+      type: ${each.value.cores}VCPU-${floor(each.value.memory / 1024)}GB
       zone: ${each.value.host}
+      region: ${var.proxmox.name}
     EOF
-    file_name = "${each.value.name}.metadata.yaml"
+    file_name = "cluster-${var.cluster.name}-${each.value.name}.metadata.yaml"
   }
 }
 
-resource "macaddress" "private" {
-  for_each = local.nodes
-}
-resource "macaddress" "public" {
-  for_each = local.nodes
-}
-
 resource "proxmox_virtual_environment_vm" "nodes" {
   for_each = local.nodes
 
+  vm_id     = each.value.id
   name      = each.value.name
   node_name = each.value.host
 
@@ -75,8 +82,11 @@ resource "proxmox_virtual_environment_vm" "nodes" {
   )
 
   stop_on_destroy     = true
+  timeout_start_vm    = 60
   timeout_stop_vm     = 60
   timeout_shutdown_vm = 60
+  timeout_reboot      = 60
+  timeout_create      = 120
 
   cpu {
     cores = each.value.cores
@@ -97,27 +107,28 @@ resource "proxmox_virtual_environment_vm" "nodes" {
 
   network_device {
     bridge      = each.value.public ? "vmpublic0" : "vmprivate0"
-    mac_address = macaddress.private[each.key].address
+    mac_address = each.value.mac_address
   }
 
-  disk {
-    file_format  = "raw"
-    interface    = "scsi0"
-    datastore_id = each.value.boot_pool
-    file_id      = proxmox_virtual_environment_download_file.talos.id
-    backup       = false
+  cdrom {
+    enabled = true
+    file_id = proxmox_virtual_environment_download_file.talos.id
   }
 
+  scsi_hardware = "virtio-scsi-single"
   disk {
     file_format  = "raw"
     interface    = "virtio0"
-    datastore_id = each.value.boot_pool
+    datastore_id = each.value.storage
     size         = each.value.boot_size
     backup       = true
+    cache        = "none"
+    iothread     = true
   }
 
   initialization {
-    meta_data_file_id = proxmox_virtual_environment_file.controlplane[each.key].id
+    datastore_id      = each.value.storage
+    meta_data_file_id = proxmox_virtual_environment_file.metadata[each.key].id
   }
 }
 
@@ -136,12 +147,24 @@ resource "proxmox_virtual_environment_user" "csi" {
   comment = "Managed by Terraform"
   user_id = "kubernetes-csi@pve"
   enabled = true
+  acl {
+    path      = "/"
+    propagate = true
+    role_id   = proxmox_virtual_environment_role.csi.role_id
+  }
 }
 resource "proxmox_virtual_environment_user_token" "csi" {
   comment    = "Managed by Terraform"
   token_name = "terraform"
   user_id    = proxmox_virtual_environment_user.csi.user_id
 }
+resource "proxmox_virtual_environment_acl" "csi" {
+  token_id = proxmox_virtual_environment_user_token.csi.id
+  role_id  = proxmox_virtual_environment_role.csi.role_id
+
+  path      = "/"
+  propagate = true
+}
 
 resource "proxmox_virtual_environment_role" "ccm" {
   role_id = "kubernetes-ccm"
@@ -154,9 +177,21 @@ resource "proxmox_virtual_environment_user" "ccm" {
   comment = "Managed by Terraform"
   user_id = "kubernetes-ccm@pve"
   enabled = true
+  acl {
+    path      = "/"
+    propagate = true
+    role_id   = proxmox_virtual_environment_role.ccm.role_id
+  }
 }
 resource "proxmox_virtual_environment_user_token" "ccm" {
   comment    = "Managed by Terraform"
   token_name = "terraform"
   user_id    = proxmox_virtual_environment_user.ccm.user_id
 }
+resource "proxmox_virtual_environment_acl" "ccm" {
+  token_id = proxmox_virtual_environment_user_token.ccm.id
+  role_id  = proxmox_virtual_environment_role.ccm.role_id
+
+  path      = "/"
+  propagate = true
+}

modules/cluster/variables.tf

@@ -7,7 +7,7 @@ variable "nodes" {
     cores  = optional(number, 4)
     memory = optional(number, 4096)
 
-    boot_pool = string
+    storage   = string
     boot_size = optional(number, 32)
   }))
 }
@@ -60,7 +60,7 @@ variable "proxmox" {
     name           = string
     url            = string
     insecure       = optional(bool, false)
-    iso_storage_id = string
+    common_storage = string
   })
   sensitive = true
 }

modules/cluster/versions.tf

@@ -20,10 +20,6 @@ terraform {
       source  = "hashicorp/random"
       version = "3.6.2"
     }
-    macaddress = {
-      source  = "ivoronin/macaddress"
-      version = "0.3.2"
-    }
     helm = {
       source  = "hashicorp/helm"
       version = "2.14.0"