Ignition service for bootstrapping kubeadm and kubelet.

README.md

@@ -11,9 +11,7 @@
 ```bash
 make -C capi-lab
 eval $(make -C capi-lab --silent dev-env)
-kubectl apply -f ../firewall-controller-manager/config/crds
-kubectl create -f https://github.com/prometheus-operator/prometheus-operator/releases/download/v0.77.1/bundle.yaml
-clusterctl init
+EXP_KUBEADM_BOOTSTRAP_FORMAT_IGNITION=true clusterctl init
 make push-to-capi-lab
 
 kubectl apply -k config/samples

capi-lab/deploy.yaml

@@ -3,8 +3,16 @@
   hosts: localhost
   connection: local
   gather_facts: false
+  vars:
+    setup_yaml:
+      - url: https://raw.githubusercontent.com/metal-stack/releases/develop/release.yaml
+        meta_var: metal_stack_release
   roles:
     - name: ansible-common
     - name: metal-roles
     - name: cert-manager
+    - name: prometheus
+    - name: firewall-controller-manager
+      vars:
+        firewall_controller_manager_namespace: cap-metal-stack
     - name: cluster-api-provider-metal-stack

capi-lab/firewall-rules.yaml

@@ -0,0 +1,23 @@
+egress:
+- comment: allow outgoing traffic for HTTP and HTTPS and DNS
+  ports:
+  - 443
+  - 80
+  - 53
+  protocol: TCP
+  to:
+  - 0.0.0.0/0
+- comment: allow outgoing DNS and NTP via UDP
+  ports:
+  - 53
+  - 123
+  protocol: UDP
+  to:
+  - 0.0.0.0/0
+ingress:
+- comment: allow incoming HTTPS to kube-apiserver
+  ports:
+  - 443
+  protocol: TCP
+  from:
+  - 0.0.0.0/0

capi-lab/requirements.yaml

@@ -6,4 +6,4 @@
   version: v0.2.9
 - src: https://github.com/metal-stack/metal-roles
   name: metal-roles
-  version: v0.14.2
+  version: v0.14.5

capi-lab/roles/firewall-controller-manager/defaults/main.yaml

@@ -0,0 +1,2 @@
+---
+firewall_controller_manager_namespace: "firewall-controller-manager"

capi-lab/roles/firewall-controller-manager/tasks/main.yaml

@@ -0,0 +1,29 @@
+---
+- name: Gather release versions
+  setup_yaml:
+
+- name: Create namespace
+  k8s:
+    definition:
+      apiVersion: v1
+      kind: Namespace
+      metadata:
+        name: "{{ firewall_controller_manager_namespace }}"
+        labels:
+          name: "{{ firewall_controller_manager_namespace }}"
+
+- name: Deploy firewall-controller-manager CRDs
+  k8s:
+    definition: "{{ lookup('url', 'https://raw.githubusercontent.com/metal-stack/firewall-controller-manager/refs/tags/' + firewall_controller_manager_image_tag + '/config/crds/' + item, split_lines=False) }}"
+    namespace: "{{ firewall_controller_manager_namespace }}"
+  loop:
+    - firewall.metal-stack.io_firewalldeployments.yaml
+    - firewall.metal-stack.io_firewallmonitors.yaml
+    - firewall.metal-stack.io_firewalls.yaml
+    - firewall.metal-stack.io_firewallsets.yaml
+
+# - name: Deploy firewall-controller-manager
+#   k8s:
+#     definition:
+
+#     namespace: "{{ firewall_controller_manager_namespace }}"

capi-lab/roles/prometheus/defaults/main.yaml

@@ -0,0 +1,3 @@
+---
+prometheus_namespace: prometheus
+prometheus_helm_chart_version: "66.2.1"

capi-lab/roles/prometheus/tasks/main.yaml

@@ -0,0 +1,22 @@
+---
+- name: Create namespace
+  k8s:
+    definition:
+      apiVersion: v1
+      kind: Namespace
+      metadata:
+        name: "{{ prometheus_namespace }}"
+        labels:
+          name: "{{ prometheus_namespace }}"
+
+- name: Deploy Prometheus
+  include_role:
+    name: ansible-common/roles/helm-chart
+  vars:
+    helm_chart: kube-prometheus-stack
+    helm_repo: https://prometheus-community.github.io/helm-charts
+    helm_force: false
+    helm_release_name: kube-prometheus-stack
+    helm_target_namespace: "{{ prometheus_namespace }}"
+    helm_chart_version: "{{ prometheus_helm_chart_version }}"
+    # helm_value_file_template: values.yaml

config/samples/example-kubeadm.yaml

@@ -0,0 +1,121 @@
+---
+apiVersion: cluster.x-k8s.io/v1beta1
+kind: Cluster
+metadata:
+  name: metal-test
+spec:
+  clusterNetwork:
+    pods:
+      cidrBlocks: ["192.168.0.0/16"]
+  controlPlaneRef:
+    apiVersion: controlplane.cluster.x-k8s.io/v1beta1
+    kind: KubeadmControlPlane
+    name: metal-test-controlplane
+  infrastructureRef:
+    apiVersion: infrastructure.cluster.x-k8s.io/v1alpha1
+    kind: MetalStackCluster
+    name: metal-test
+---
+apiVersion: infrastructure.cluster.x-k8s.io/v1alpha1
+kind: MetalStackCluster
+metadata:
+  name: metal-test
+spec:
+  projectID: 00000000-0000-0000-0000-000000000001
+  partition: mini-lab
+  firewall:
+    size: v1-small-x86
+    image: firewall-ubuntu-3.0
+    networks:
+    - internet-mini-lab
+---
+apiVersion: infrastructure.cluster.x-k8s.io/v1alpha1
+kind: MetalStackMachineTemplate
+metadata:
+  name: metal-test-controlplane
+spec:
+  template:
+    spec:
+      image: ubuntu-24.04
+      size: v1-small-x86
+---
+kind: KubeadmControlPlane
+apiVersion: controlplane.cluster.x-k8s.io/v1beta1
+metadata:
+  name: metal-test-controlplane
+spec:
+  replicas: 1
+  version: v1.30.6
+  machineTemplate:
+    nodeDrainTimeout: 10m
+    infrastructureRef:
+      kind: MetalStackMachineTemplate
+      apiVersion: infrastructure.cluster.x-k8s.io/v1alpha1
+      name: metal-test-controlplane
+  kubeadmConfigSpec:
+    format: ignition
+    clusterConfiguration:
+      controlPlaneEndpoint: 203.0.113.129:443
+    initConfiguration:
+      localAPIEndpoint:
+        advertiseAddress: 10.0.0.2
+        bindPort: 443
+      nodeRegistration: {}
+    joinConfiguration:
+      controlPlane: {}
+      nodeRegistration: {}
+    ignition:
+      containerLinuxConfig:
+        additionalConfig: |
+          systemd:
+            units:
+            - name: cluster-api-init.service
+              enable: true
+              contents: |-
+                [Unit]
+                Description=Prepares the node for bootstrapping with cluster-api kubeadm
+                Before=kubeadm.service
+                After=network-online.target
+                Wants=network-online.target
+                [Service]
+                Type=oneshot
+                Restart=on-failure
+                RestartSec=5
+                StartLimitBurst=0
+                EnvironmentFile=/etc/environment
+                ExecStart=/var/lib/cluster-api-init/bootstrap.sh
+                [Install]
+                WantedBy=multi-user.target
+    files:
+      - path: /var/lib/cluster-api-init/bootstrap.sh
+        owner: "root:root"
+        permissions: "0744"
+        content: |
+          #!/usr/bin/env bash
+          set -eo pipefail
+          set +x
+
+          apt update
+          apt install conntrack
+
+          CNI_PLUGINS_VERSION="v1.3.0"
+          DEST="/opt/cni/bin"
+          mkdir -p "$DEST"
+          curl -L "https://github.com/containernetworking/plugins/releases/download/${CNI_PLUGINS_VERSION}/cni-plugins-linux-amd64-${CNI_PLUGINS_VERSION}.tgz" | sudo tar -C "$DEST" -xz
+
+          RELEASE="v1.30.6"
+          cd /usr/local/bin
+          sudo curl -L --remote-name-all https://dl.k8s.io/release/${RELEASE}/bin/linux/amd64/{kubeadm,kubelet,kubectl}
+          sudo chmod +x {kubeadm,kubelet,kubectl}
+
+          RELEASE_VERSION="v0.16.2"
+          curl -sSL "https://raw.githubusercontent.com/kubernetes/release/${RELEASE_VERSION}/cmd/krel/templates/latest/kubelet/kubelet.service" | sed "s:/usr/bin:/usr/local/bin:g" | sudo tee /usr/lib/systemd/system/kubelet.service
+          sudo mkdir -p /usr/lib/systemd/system/kubelet.service.d
+          curl -sSL "https://raw.githubusercontent.com/kubernetes/release/${RELEASE_VERSION}/cmd/krel/templates/latest/kubeadm/10-kubeadm.conf" | sed "s:/usr/bin:/usr/local/bin:g" | sudo tee /usr/lib/systemd/system/kubelet.service.d/10-kubeadm.conf
+
+          systemctl enable kubelet.service
+      - path: /etc/containerd/config.toml
+        owner: "root:root"
+        permissions: "0644"
+        content: |
+          disabled_plugins = []

config/samples/kustomization.yaml

@@ -2,5 +2,5 @@
 namespace: default
 
 resources:
-- example.yaml
+- example-kubeadm.yaml
 # +kubebuilder:scaffold:manifestskustomizesamples

internal/controller/metalstackcluster_controller.go

@@ -166,12 +166,12 @@ func (r *MetalStackClusterReconciler) SetupWithManager(mgr ctrl.Manager) error {
 }
 
 func (r *clusterReconciler) reconcile() error {
-	nodeCIDR, err := r.ensureNodeNetwork()
+	nodeNetworkID, err := r.ensureNodeNetwork()
 	if err != nil {
 		return fmt.Errorf("unable to ensure node network: %w", err)
 	}
 
-	r.log.Info("reconciled node network", "cidr", nodeCIDR)
+	r.log.Info("reconciled node network", "network-id", nodeNetworkID)
 
 	ip, err := r.ensureControlPlaneIP()
 	if err != nil {
@@ -197,7 +197,7 @@ func (r *clusterReconciler) reconcile() error {
 		return fmt.Errorf("failed to update infra cluster control plane endpoint: %w", err)
 	}
 
-	fwdeploy, err := r.ensureFirewallDeployment(nodeCIDR)
+	fwdeploy, err := r.ensureFirewallDeployment(nodeNetworkID)
 	if err != nil {
 		return fmt.Errorf("unable to ensure firewall deployment: %w", err)
 	}
@@ -259,15 +259,15 @@ func (r *clusterReconciler) ensureNodeNetwork() (string, error) {
 			return "", fmt.Errorf("error creating node network: %w", err)
 		}
 
-		return resp.Payload.Prefixes[0], nil
+		return *resp.Payload.ID, nil
 	case 1:
 		nw := nws[0]
 
 		if len(nw.Prefixes) == 0 {
 			return "", errors.New("node network exists but the prefix is gone")
 		}
 
-		return nw.Prefixes[0], nil
+		return *nw.ID, nil
 	default:
 		return "", fmt.Errorf("more than a single node network exists for this cluster, operator investigation is required")
 	}
@@ -399,7 +399,7 @@ func (r *clusterReconciler) findControlPlaneIP() ([]*models.V1IPResponse, error)
 	return resp.Payload, nil
 }
 
-func (r *clusterReconciler) ensureFirewallDeployment(nodeCIDR string) (*fcmv2.FirewallDeployment, error) {
+func (r *clusterReconciler) ensureFirewallDeployment(nodeNetworkID string) (*fcmv2.FirewallDeployment, error) {
 	deploy := &fcmv2.FirewallDeployment{
 		ObjectMeta: metav1.ObjectMeta{
 			Name:      r.infraCluster.Name,
@@ -440,7 +440,7 @@ func (r *clusterReconciler) ensureFirewallDeployment(nodeCIDR string) (*fcmv2.Fi
 
 		deploy.Spec.Template.Spec.Size = r.infraCluster.Spec.Firewall.Size
 		deploy.Spec.Template.Spec.Image = r.infraCluster.Spec.Firewall.Image
-		deploy.Spec.Template.Spec.Networks = append(r.infraCluster.Spec.Firewall.AdditionalNetworks, nodeCIDR)
+		deploy.Spec.Template.Spec.Networks = append(r.infraCluster.Spec.Firewall.AdditionalNetworks, nodeNetworkID)
 		deploy.Spec.Template.Spec.RateLimits = r.infraCluster.Spec.Firewall.RateLimits
 		deploy.Spec.Template.Spec.EgressRules = r.infraCluster.Spec.Firewall.EgressRules
 		deploy.Spec.Template.Spec.LogAcceptedConnections = ptr.Deref(r.infraCluster.Spec.Firewall.LogAcceptedConnections, false)
@@ -452,6 +452,10 @@ func (r *clusterReconciler) ensureFirewallDeployment(nodeCIDR string) (*fcmv2.Fi
 		deploy.Spec.Template.Spec.NftablesExporterVersion = ""
 		deploy.Spec.Template.Spec.NftablesExporterURL = ""
 
+		// TODO: we need to allow internet connection for the nodes before the firewall-controller can connect to the control-plane
+		// the FCM currently does not support this
+		deploy.Spec.Template.Spec.Userdata = ""
+
 		// TODO: do we need to generate ssh keys for the machines and the firewall in this controller?
 		deploy.Spec.Template.Spec.SSHPublicKeys = nil