feat: make kubeadm workflows auto-scale with terraform outputs

nixos/kubeadm/scripts/rebuild-and-bootstrap.sh

@@ -7,7 +7,7 @@ INVENTORY_FILE="${1:-$SCRIPT_DIR/inventory.env}"
 
 if [ ! -f "$INVENTORY_FILE" ]; then
   echo "Missing inventory file: $INVENTORY_FILE"
-  echo "Copy $SCRIPT_DIR/inventory.example.env to $SCRIPT_DIR/inventory.env and edit IPs."
+  echo "Copy $SCRIPT_DIR/inventory.example.env to $SCRIPT_DIR/inventory.env and edit node mappings."
   exit 1
 fi
 
@@ -17,21 +17,61 @@ source "$INVENTORY_FILE"
 SSH_USER="${SSH_USER:-micqdf}"
 SSH_OPTS="${SSH_OPTS:- -o BatchMode=yes -o StrictHostKeyChecking=accept-new }"
 
-required=(CP_1 CP_2 CP_3 WK_1 WK_2 WK_3)
-for key in "${required[@]}"; do
-  if [ -z "${!key:-}" ]; then
-    echo "Missing required inventory variable: $key"
+declare -A NODE_IPS=()
+declare -a CP_NAMES=()
+declare -a WK_NAMES=()
+
+add_node_pair() {
+  local role="$1"
+  local pair="$2"
+  local name="${pair%%=*}"
+  local ip="${pair#*=}"
+
+  if [ -z "$name" ] || [ -z "$ip" ] || [ "$name" = "$ip" ]; then
+    echo "Invalid node pair '$pair' (expected name=ip)."
     exit 1
   fi
-done
 
-cluster_has_node() {
-  local node_name="$1"
-  remote "$CP_1" "sudo kubectl --kubeconfig /etc/kubernetes/admin.conf get node $node_name >/dev/null 2>&1"
+  NODE_IPS["$name"]="$ip"
+  if [ "$role" = "cp" ]; then
+    CP_NAMES+=("$name")
+  else
+    WK_NAMES+=("$name")
+  fi
 }
 
-cluster_ready() {
-  remote "$CP_1" "test -f /etc/kubernetes/admin.conf && sudo kubectl --kubeconfig /etc/kubernetes/admin.conf get nodes >/dev/null 2>&1"
+populate_nodes() {
+  if [ -n "${CONTROL_PLANES:-}" ]; then
+    for pair in $CONTROL_PLANES; do
+      add_node_pair "cp" "$pair"
+    done
+  else
+    while IFS= read -r var_name; do
+      idx="${var_name#CP_}"
+      add_node_pair "cp" "cp-$idx=${!var_name}"
+    done < <(compgen -A variable | grep -E '^CP_[0-9]+$' | sort -V)
+  fi
+
+  if [ -n "${WORKERS:-}" ]; then
+    for pair in $WORKERS; do
+      add_node_pair "wk" "$pair"
+    done
+  else
+    while IFS= read -r var_name; do
+      idx="${var_name#WK_}"
+      add_node_pair "wk" "wk-$idx=${!var_name}"
+    done < <(compgen -A variable | grep -E '^WK_[0-9]+$' | sort -V)
+  fi
+
+  if [ "${#CP_NAMES[@]}" -eq 0 ]; then
+    echo "No control planes found in inventory."
+    exit 1
+  fi
+
+  if [ "${#WK_NAMES[@]}" -eq 0 ]; then
+    echo "No workers found in inventory."
+    exit 1
+  fi
 }
 
 remote() {
@@ -40,6 +80,15 @@ remote() {
   ssh $SSH_OPTS "$SSH_USER@$host_ip" "$cmd"
 }
 
+cluster_has_node() {
+  local node_name="$1"
+  remote "$PRIMARY_CP_IP" "sudo kubectl --kubeconfig /etc/kubernetes/admin.conf get node $node_name >/dev/null 2>&1"
+}
+
+cluster_ready() {
+  remote "$PRIMARY_CP_IP" "test -f /etc/kubernetes/admin.conf && sudo kubectl --kubeconfig /etc/kubernetes/admin.conf get nodes >/dev/null 2>&1"
+}
+
 rebuild_node() {
   local node_name="$1"
   local node_ip="$2"
@@ -51,28 +100,38 @@ rebuild_node() {
     --use-remote-sudo
 }
 
-for node in cp-1 cp-2 cp-3 wk-1 wk-2 wk-3; do
-  key="${node^^}"
-  key="${key//-/_}"
-  rebuild_node "$node" "${!key}"
+populate_nodes
+
+PRIMARY_CONTROL_PLANE="${PRIMARY_CONTROL_PLANE:-cp-1}"
+if [ -z "${NODE_IPS[$PRIMARY_CONTROL_PLANE]:-}" ]; then
+  PRIMARY_CONTROL_PLANE="${CP_NAMES[0]}"
+fi
+PRIMARY_CP_IP="${NODE_IPS[$PRIMARY_CONTROL_PLANE]}"
+
+for node in "${CP_NAMES[@]}"; do
+  rebuild_node "$node" "${NODE_IPS[$node]}"
 done
 
-echo "==> Initializing control plane on cp-1"
-if cluster_ready; then
-  echo "==> Existing cluster detected on cp-1; skipping kubeadm init"
-else
-  remote "$CP_1" "sudo th-kubeadm-init"
+for node in "${WK_NAMES[@]}"; do
+  rebuild_node "$node" "${NODE_IPS[$node]}"
+done
 
-  echo "==> Installing Cilium on cp-1"
-  remote "$CP_1" "helm repo add cilium https://helm.cilium.io >/dev/null 2>&1 || true"
-  remote "$CP_1" "helm repo update >/dev/null"
-  remote "$CP_1" "kubectl create namespace kube-system >/dev/null 2>&1 || true"
-  remote "$CP_1" "helm upgrade --install cilium cilium/cilium --namespace kube-system --set kubeProxyReplacement=true"
+echo "==> Initializing control plane on $PRIMARY_CONTROL_PLANE"
+if cluster_ready; then
+  echo "==> Existing cluster detected on $PRIMARY_CONTROL_PLANE; skipping kubeadm init"
+else
+  remote "$PRIMARY_CP_IP" "sudo th-kubeadm-init"
+
+  echo "==> Installing Cilium on $PRIMARY_CONTROL_PLANE"
+  remote "$PRIMARY_CP_IP" "helm repo add cilium https://helm.cilium.io >/dev/null 2>&1 || true"
+  remote "$PRIMARY_CP_IP" "helm repo update >/dev/null"
+  remote "$PRIMARY_CP_IP" "kubectl create namespace kube-system >/dev/null 2>&1 || true"
+  remote "$PRIMARY_CP_IP" "helm upgrade --install cilium cilium/cilium --namespace kube-system --set kubeProxyReplacement=true"
 fi
 
 echo "==> Building kubeadm join commands"
-JOIN_CMD="$(remote "$CP_1" "sudo kubeadm token create --print-join-command")"
-CERT_KEY="$(remote "$CP_1" "sudo kubeadm init phase upload-certs --upload-certs | tail -n 1")"
+JOIN_CMD="$(remote "$PRIMARY_CP_IP" "sudo kubeadm token create --print-join-command")"
+CERT_KEY="$(remote "$PRIMARY_CP_IP" "sudo kubeadm init phase upload-certs --upload-certs | tail -n 1")"
 CP_JOIN_CMD="$JOIN_CMD --control-plane --certificate-key $CERT_KEY"
 
 join_control_plane() {
@@ -90,36 +149,26 @@ join_worker() {
 }
 
 echo "==> Joining remaining control planes"
-if cluster_has_node "cp-2"; then
-  echo "cp-2 already joined; skipping"
-else
-  join_control_plane "$CP_2"
-fi
+for node in "${CP_NAMES[@]}"; do
+  if [ "$node" = "$PRIMARY_CONTROL_PLANE" ]; then
+    continue
+  fi
 
-if cluster_has_node "cp-3"; then
-  echo "cp-3 already joined; skipping"
-else
-  join_control_plane "$CP_3"
-fi
+  if cluster_has_node "$node"; then
+    echo "$node already joined; skipping"
+  else
+    join_control_plane "${NODE_IPS[$node]}"
+  fi
+done
 
 echo "==> Joining workers"
-if cluster_has_node "wk-1"; then
-  echo "wk-1 already joined; skipping"
-else
-  join_worker "$WK_1"
-fi
-
-if cluster_has_node "wk-2"; then
-  echo "wk-2 already joined; skipping"
-else
-  join_worker "$WK_2"
-fi
-
-if cluster_has_node "wk-3"; then
-  echo "wk-3 already joined; skipping"
-else
-  join_worker "$WK_3"
-fi
+for node in "${WK_NAMES[@]}"; do
+  if cluster_has_node "$node"; then
+    echo "$node already joined; skipping"
+  else
+    join_worker "${NODE_IPS[$node]}"
+  fi
+done
 
 echo "==> Final node list"
-remote "$CP_1" "kubectl get nodes -o wide"
+remote "$PRIMARY_CP_IP" "kubectl get nodes -o wide"