wild-cloud/api/internal/node/node.go

package node

import (
	"context"
	"fmt"
	"log/slog"
	"os"
	"os/exec"
	"path/filepath"
	"strings"
	"sync"
	"time"

	"github.com/wild-cloud/wild-central/daemon/internal/config"
	"github.com/wild-cloud/wild-central/daemon/internal/setup"
	"github.com/wild-cloud/wild-central/daemon/internal/tools"
	"gopkg.in/yaml.v3"
)

// Manager handles node configuration and state management
type Manager struct {
	dataDir   string
	configMgr *config.Manager
	talosctl  *tools.Talosctl
}

// NewManager creates a new node manager
func NewManager(dataDir string, instanceName string) *Manager {
	var talosctl *tools.Talosctl

	// If instanceName is provided, use instance-specific talosconfig
	// Otherwise, create basic talosctl (will use --insecure mode)
	if instanceName != "" {
		talosconfigPath := tools.GetTalosconfigPath(dataDir, instanceName)
		talosctl = tools.NewTalosconfigWithConfig(talosconfigPath)
	} else {
		talosctl = tools.NewTalosctl()
	}

	return &Manager{
		dataDir:   dataDir,
		configMgr: config.NewManager(),
		talosctl:  talosctl,
	}
}

// Node represents a cluster node configuration
type Node struct {
	Hostname    string `yaml:"hostname" json:"hostname"`
	Role        string `yaml:"role" json:"role"` // controlplane or worker
	TargetIP    string `yaml:"targetIp" json:"target_ip"`
	CurrentIP   string `yaml:"currentIp,omitempty" json:"current_ip,omitempty"` // For maintenance mode detection
	Interface   string `yaml:"interface,omitempty" json:"interface,omitempty"`
	Disk        string `yaml:"disk" json:"disk"`
	Version     string `yaml:"version,omitempty" json:"version,omitempty"`
	SchematicID string `yaml:"schematicId,omitempty" json:"schematic_id,omitempty"`
	Maintenance bool   `yaml:"maintenance,omitempty" json:"maintenance"` // Explicit maintenance mode flag
	Configured  bool   `yaml:"configured,omitempty" json:"configured"`
	Applied     bool   `yaml:"applied,omitempty" json:"applied"`
}

// HardwareInfo contains discovered hardware information
type HardwareInfo struct {
	IP              string           `json:"ip"`
	Interface       string           `json:"interface"`
	Disks           []tools.DiskInfo `json:"disks"`
	SelectedDisk    string           `json:"selected_disk"`
	MaintenanceMode bool             `json:"maintenance_mode"`
}

// ApplyOptions contains options for node apply
type ApplyOptions struct {
	// No options needed - apply always regenerates and auto-fetches templates
}

// GetInstancePath returns the path to an instance's nodes directory
func (m *Manager) GetInstancePath(instanceName string) string {
	return tools.GetInstancePath(m.dataDir, instanceName)
}

// nodeConfigFile is the operator-authored portion stored in config/nodes/{hostname}.yaml
type nodeConfigFile struct {
	Role        string `yaml:"role"`
	Disk        string `yaml:"disk"`
	Interface   string `yaml:"interface,omitempty"`
	TargetIP    string `yaml:"targetIp,omitempty"`
	SchematicID string `yaml:"schematicId,omitempty"`
	Version     string `yaml:"version,omitempty"`
	Maintenance bool   `yaml:"maintenance,omitempty"`
}

// nodeCacheFile is the daemon-written state stored in cache/nodes/{hostname}.yaml
type nodeCacheFile struct {
	Applied    bool   `yaml:"applied,omitempty"`
	Configured bool   `yaml:"configured,omitempty"`
	CurrentIP  string `yaml:"currentIp,omitempty"`
}

// List returns all nodes for an instance
func (m *Manager) List(instanceName string) ([]Node, error) {
	return m.listFromNodeFiles(instanceName)
}

// listFromNodeFiles reads nodes from config/nodes/*.yaml + cache/nodes/*.yaml
func (m *Manager) listFromNodeFiles(instanceName string) ([]Node, error) {
	configDir := tools.GetNodeConfigDirPath(m.dataDir, instanceName)
	entries, err := os.ReadDir(configDir)
	if os.IsNotExist(err) {
		return []Node{}, nil
	}
	if err != nil {
		return nil, fmt.Errorf("failed to read node config dir: %w", err)
	}

	var nodes []Node
	for _, entry := range entries {
		if entry.IsDir() || filepath.Ext(entry.Name()) != ".yaml" {
			continue
		}
		hostname := entry.Name()[:len(entry.Name())-5] // strip .yaml

		data, err := os.ReadFile(filepath.Join(configDir, entry.Name()))
		if err != nil {
			slog.Warn("failed to read node config", "hostname", hostname, "error", err)
			continue
		}
		var cfg nodeConfigFile
		if err := yaml.Unmarshal(data, &cfg); err != nil {
			slog.Warn("failed to parse node config", "hostname", hostname, "error", err)
			continue
		}

		node := Node{
			Hostname:    hostname,
			Role:        cfg.Role,
			Disk:        cfg.Disk,
			Interface:   cfg.Interface,
			TargetIP:    cfg.TargetIP,
			SchematicID: cfg.SchematicID,
			Version:     cfg.Version,
			Maintenance: cfg.Maintenance,
		}

		// Merge cache
		cacheData, err := os.ReadFile(tools.GetNodeCachePath(m.dataDir, instanceName, hostname))
		if err == nil {
			var cache nodeCacheFile
			if err := yaml.Unmarshal(cacheData, &cache); err == nil {
				node.Applied = cache.Applied
				node.Configured = cache.Configured
				node.CurrentIP = cache.CurrentIP
			}
		}

		nodes = append(nodes, node)
	}
	return nodes, nil
}


// Get returns a specific node by hostname
func (m *Manager) Get(instanceName, hostname string) (*Node, error) {
	// Get all nodes
	nodes, err := m.List(instanceName)
	if err != nil {
		return nil, err
	}

	// Find node by hostname
	for _, node := range nodes {
		if node.Hostname == hostname {
			return &node, nil
		}
	}

	return nil, fmt.Errorf("node %s not found", hostname)
}

// Add registers a new node
func (m *Manager) Add(instanceName string, node *Node) error {
	slog.Info("adding node", "component", "node", "instance", instanceName, "hostname", node.Hostname, "role", node.Role)

	// Validate node data
	if node.Hostname == "" {
		return fmt.Errorf("hostname is required")
	}
	if node.Role != "controlplane" && node.Role != "worker" {
		return fmt.Errorf("role must be 'controlplane' or 'worker'")
	}
	if node.Disk == "" {
		return fmt.Errorf("disk is required")
	}

	// Check if node already exists - ERROR if yes
	existing, err := m.Get(instanceName, node.Hostname)
	if err == nil && existing != nil {
		return fmt.Errorf("node %s already exists", node.Hostname)
	}

	// Set maintenance=true if currentIP provided (node in maintenance mode)
	if node.CurrentIP != "" {
		node.Maintenance = true
	}

	return m.addNodeFile(instanceName, node)
}

// addNodeFile writes a new node's config to config/nodes/{hostname}.yaml (new format)
func (m *Manager) addNodeFile(instanceName string, node *Node) error {
	// Fill talos defaults from instance.yaml if missing
	if node.SchematicID == "" || node.Version == "" {
		instanceYaml := tools.GetInstanceYamlPath(m.dataDir, instanceName)
		if data, err := os.ReadFile(instanceYaml); err == nil {
			var root struct {
				Cluster struct {
					Nodes struct {
						Talos struct {
							SchematicID string `yaml:"schematicId"`
							Version     string `yaml:"version"`
						} `yaml:"talos"`
					} `yaml:"nodes"`
				} `yaml:"cluster"`
			}
			if err := yaml.Unmarshal(data, &root); err == nil {
				if node.SchematicID == "" {
					node.SchematicID = root.Cluster.Nodes.Talos.SchematicID
				}
				if node.Version == "" {
					node.Version = root.Cluster.Nodes.Talos.Version
				}
			}
		}
	}

	configDir := tools.GetNodeConfigDirPath(m.dataDir, instanceName)
	if err := os.MkdirAll(configDir, 0755); err != nil {
		return fmt.Errorf("failed to create node config dir: %w", err)
	}

	cfg := nodeConfigFile{
		Role:        node.Role,
		Disk:        node.Disk,
		Interface:   node.Interface,
		TargetIP:    node.TargetIP,
		SchematicID: node.SchematicID,
		Version:     node.Version,
		Maintenance: node.Maintenance,
	}
	data, err := yaml.Marshal(cfg)
	if err != nil {
		return fmt.Errorf("failed to marshal node config: %w", err)
	}
	return os.WriteFile(tools.GetNodeConfigPath(m.dataDir, instanceName, node.Hostname), data, 0644)
}

// Delete removes a node from config.yaml
// If skipReset is false, the node will be reset before deletion (with 30s timeout)
func (m *Manager) Delete(instanceName, nodeIdentifier string, skipReset bool) error {
	slog.Info("deleting node", "component", "node", "instance", instanceName, "node", nodeIdentifier, "skipReset", skipReset)

	// Get node to find hostname
	node, err := m.Get(instanceName, nodeIdentifier)
	if err != nil {
		return err
	}

	// Reset node first unless skipReset is true
	if !skipReset {
		ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
		defer cancel()

		// Use goroutine to respect context timeout
		done := make(chan error, 1)
		go func() {
			done <- m.Reset(instanceName, nodeIdentifier)
		}()

		select {
		case err := <-done:
			if err != nil {
				return fmt.Errorf("failed to reset node before deletion (use skip_reset=true to force delete): %w", err)
			}
		case <-ctx.Done():
			return fmt.Errorf("node reset timed out after 30 seconds (use skip_reset=true to force delete)")
		}
	}

	// Remove the k8s node object (best-effort — node may already be gone)
	kubeconfigPath := tools.GetKubeconfigPath(m.dataDir, instanceName)
	cmd := exec.Command("kubectl", "delete", "node", node.Hostname, "--ignore-not-found")
	tools.WithKubeconfig(cmd, kubeconfigPath)
	if out, err := cmd.CombinedOutput(); err != nil {
		slog.Warn("failed to delete k8s node object (continuing)", "component", "node",
			"hostname", node.Hostname, "error", err, "output", strings.TrimSpace(string(out)))
	}

	// Delete node from config.yaml
	return m.deleteFromConfig(instanceName, node.Hostname)
}

// deleteFromConfig removes a node entry from config
func (m *Manager) deleteFromConfig(instanceName, hostname string) error {
	configPath := tools.GetNodeConfigPath(m.dataDir, instanceName, hostname)
	if err := os.Remove(configPath); err != nil && !os.IsNotExist(err) {
		return fmt.Errorf("failed to delete node config: %w", err)
	}
	cachePath := tools.GetNodeCachePath(m.dataDir, instanceName, hostname)
	if err := os.Remove(cachePath); err != nil && !os.IsNotExist(err) {
		slog.Warn("failed to delete node cache", "component", "node", "hostname", hostname, "error", err)
	}
	return nil
}

// DetectHardware queries node hardware information via talosctl
// Automatically detects maintenance mode by trying insecure first, then secure
func (m *Manager) DetectHardware(nodeIP string) (*HardwareInfo, error) {
	// Try insecure first (maintenance mode)
	hwInfo, err := m.detectHardwareWithMode(nodeIP, true)
	if err == nil {
		return hwInfo, nil
	}

	// Fall back to secure (configured node)
	return m.detectHardwareWithMode(nodeIP, false)
}

// detectHardwareWithMode queries node hardware with specified connection mode
func (m *Manager) detectHardwareWithMode(nodeIP string, insecure bool) (*HardwareInfo, error) {
	// Try to get default interface (with default route)
	iface, err := m.talosctl.GetDefaultInterface(nodeIP, insecure)
	if err != nil {
		// Fall back to physical interface
		iface, err = m.talosctl.GetPhysicalInterface(nodeIP, insecure)
		if err != nil {
			return nil, fmt.Errorf("failed to detect interface: %w", err)
		}
	}

	// Get disks
	disks, err := m.talosctl.GetDisks(nodeIP, insecure)
	if err != nil {
		return nil, fmt.Errorf("failed to detect disks: %w", err)
	}

	// Select first disk as default
	var selectedDisk string
	if len(disks) > 0 {
		selectedDisk = disks[0].Path
	}

	return &HardwareInfo{
		IP:              nodeIP,
		Interface:       iface,
		Disks:           disks,
		SelectedDisk:    selectedDisk,
		MaintenanceMode: insecure, // If we used insecure, it's in maintenance mode
	}, nil
}

// Apply generates configuration and applies it to node
// This follows the wild-node-apply flow:
// 1. Auto-fetch templates if missing
// 2. Generate node-specific patch file from template
// 3. Merge base config + patch → final config (talosctl machineconfig patch)
// 4. Apply final config to node (talosctl apply-config --insecure if maintenance mode)
// 5. Update state: currentIP=targetIP, maintenance=false, applied=true
func (m *Manager) Apply(instanceName, nodeIdentifier string, opts ApplyOptions) error {
	// Get node configuration
	node, err := m.Get(instanceName, nodeIdentifier)
	if err != nil {
		return err
	}

	setupDir := tools.GetNodeSetupCachePath(m.dataDir, instanceName)

	// Ensure node has version and schematicId (use cluster defaults if missing)
	if node.Version == "" || node.SchematicID == "" {
		m.fillTalosDefaults(instanceName, node)
	}

	// Always auto-extract templates from embedded files if they don't exist
	templatesDir := filepath.Join(setupDir, "patch.templates")
	if !m.templatesExist(templatesDir) {
		if err := m.extractEmbeddedTemplates(templatesDir); err != nil {
			return fmt.Errorf("failed to extract templates: %w", err)
		}
	}

	// Determine base configuration file (generated by cluster config generation)
	var baseConfig string
	if node.Role == "controlplane" {
		baseConfig = tools.GetControlplaneConfigPath(m.dataDir, instanceName)
	} else {
		baseConfig = tools.GetWorkerConfigPath(m.dataDir, instanceName)
	}

	// Check if base config exists
	if _, err := os.Stat(baseConfig); err != nil {
		return fmt.Errorf("base configuration not found: %s (run cluster config generation first)", baseConfig)
	}

	// Generate node-specific patch file
	patchFile, err := m.generateNodePatch(instanceName, node, setupDir)
	if err != nil {
		return fmt.Errorf("failed to generate node patch: %w", err)
	}

	// Generate final machine configuration (base + patch)
	finalConfig, err := m.generateFinalConfig(node, baseConfig, patchFile, setupDir)
	if err != nil {
		return fmt.Errorf("failed to generate final configuration: %w", err)
	}

	// Mark as configured
	node.Configured = true
	if err := m.updateNodeStatus(instanceName, node); err != nil {
		return fmt.Errorf("failed to update node status: %w", err)
	}

	slog.Info("applying node config", "component", "node", "instance", instanceName, "hostname", node.Hostname, "role", node.Role)

	// Apply configuration to node
	// Determine which IP to use and whether node is in maintenance mode
	//
	// Three scenarios:
	// 1. Production node (already applied, maintenance=false): use targetIP, no --insecure
	// 2. IP changing (currentIP != targetIP): use currentIP, --insecure (always maintenance)
	// 3. Fresh/maintenance node (never applied OR maintenance=true): use targetIP, --insecure
	var deployIP string
	var maintenanceMode bool

	if node.CurrentIP != "" && node.CurrentIP != node.TargetIP {
		// Scenario 2: IP is changing - node is at currentIP, moving to targetIP
		deployIP = node.CurrentIP
		maintenanceMode = true
	} else if node.Maintenance || !node.Applied {
		// Scenario 3: Explicit maintenance mode OR never been applied (fresh node)
		// Fresh nodes need --insecure because they have self-signed certificates
		deployIP = node.TargetIP
		maintenanceMode = true
	} else {
		// Scenario 1: Production node at target IP (already applied, not in maintenance)
		deployIP = node.TargetIP
		maintenanceMode = false
	}

	// Apply config
	talosconfigPath := tools.GetTalosconfigPath(m.dataDir, instanceName)
	if err := m.talosctl.ApplyConfig(deployIP, finalConfig, maintenanceMode, talosconfigPath); err != nil {
		return fmt.Errorf("failed to apply config to %s: %w", deployIP, err)
	}

	// Post-application updates: move to production IP, exit maintenance mode
	node.Applied = true
	node.CurrentIP = node.TargetIP // Node now on production IP
	node.Maintenance = false       // Exit maintenance mode
	if err := m.updateNodeStatus(instanceName, node); err != nil {
		return fmt.Errorf("failed to update node status: %w", err)
	}

	slog.Info("node config applied", "component", "node", "instance", instanceName, "hostname", node.Hostname, "ip", node.TargetIP)
	return nil
}

// generateNodePatch creates a node-specific patch file from template
func (m *Manager) generateNodePatch(instanceName string, node *Node, setupDir string) (string, error) {
	// Determine template file based on role
	var templateFile string
	if node.Role == "controlplane" {
		templateFile = filepath.Join(setupDir, "patch.templates", "controlplane.yaml")
	} else {
		templateFile = filepath.Join(setupDir, "patch.templates", "worker.yaml")
	}

	// Read template
	templateContent, err := os.ReadFile(templateFile)
	if err != nil {
		return "", fmt.Errorf("failed to read template %s: %w", templateFile, err)
	}

	// Stage 1: Apply simple variable substitutions (like v.PoC does with sed)
	patchContent := string(templateContent)
	patchContent = strings.ReplaceAll(patchContent, "{{NODE_NAME}}", node.Hostname)
	patchContent = strings.ReplaceAll(patchContent, "{{NODE_IP}}", node.TargetIP)
	patchContent = strings.ReplaceAll(patchContent, "{{SCHEMATIC_ID}}", node.SchematicID)
	patchContent = strings.ReplaceAll(patchContent, "{{VERSION}}", node.Version)

	// Stage 2: Process through gomplate with merged global+instance config context
	merged, err := config.LoadMergedInstanceConfig(m.dataDir, instanceName)
	if err != nil {
		return "", fmt.Errorf("failed to load merged config: %w", err)
	}

	mergedYAML, err := yaml.Marshal(merged)
	if err != nil {
		return "", fmt.Errorf("failed to marshal merged config: %w", err)
	}

	gomplate := tools.NewGomplate()
	processedPatch, err := gomplate.RenderTemplate(patchContent, string(mergedYAML))
	if err != nil {
		return "", fmt.Errorf("failed to process template with gomplate: %w", err)
	}

	// Wipe disk when node needs a fresh install (maintenance or never applied)
	if node.Maintenance || !node.Applied {
		var patchMap map[string]interface{}
		if err := yaml.Unmarshal([]byte(processedPatch), &patchMap); err == nil {
			if machine, ok := patchMap["machine"].(map[string]interface{}); ok {
				if install, ok := machine["install"].(map[string]interface{}); ok {
					install["wipe"] = true
					if patched, err := yaml.Marshal(patchMap); err == nil {
						processedPatch = string(patched)
					}
				}
			}
		}
	}

	// Create patch directory
	patchDir := filepath.Join(setupDir, "patch")
	if err := os.MkdirAll(patchDir, 0755); err != nil {
		return "", fmt.Errorf("failed to create patch directory: %w", err)
	}

	// Write patch file
	patchFile := filepath.Join(patchDir, node.Hostname+".yaml")
	if err := os.WriteFile(patchFile, []byte(processedPatch), 0644); err != nil {
		return "", fmt.Errorf("failed to write patch file: %w", err)
	}

	return patchFile, nil
}

// generateFinalConfig merges base config + patch to create final machine config
func (m *Manager) generateFinalConfig(node *Node, baseConfig, patchFile, setupDir string) (string, error) {
	// Create final config directory
	finalDir := filepath.Join(setupDir, "final")
	if err := os.MkdirAll(finalDir, 0755); err != nil {
		return "", fmt.Errorf("failed to create final directory: %w", err)
	}

	finalConfig := filepath.Join(finalDir, node.Hostname+".yaml")

	// Use talosctl machineconfig patch to merge base + patch
	// talosctl machineconfig patch base.yaml --patch @patch.yaml -o final.yaml
	cmd := exec.Command("talosctl", "machineconfig", "patch", baseConfig,
		"--patch", "@"+patchFile,
		"-o", finalConfig)

	output, err := cmd.CombinedOutput()
	if err != nil {
		return "", fmt.Errorf("failed to patch machine config: %w\nOutput: %s", err, string(output))
	}

	return finalConfig, nil
}

// templatesExist checks if patch templates exist in the instance directory
func (m *Manager) templatesExist(templatesDir string) bool {
	controlplaneTemplate := filepath.Join(templatesDir, "controlplane.yaml")
	workerTemplate := filepath.Join(templatesDir, "worker.yaml")

	_, err1 := os.Stat(controlplaneTemplate)
	_, err2 := os.Stat(workerTemplate)

	return err1 == nil && err2 == nil
}

// extractEmbeddedTemplates extracts patch templates from embedded files to instance directory
func (m *Manager) extractEmbeddedTemplates(destDir string) error {
	// Create destination directory
	if err := os.MkdirAll(destDir, 0755); err != nil {
		return fmt.Errorf("failed to create templates directory: %w", err)
	}

	// Get embedded template files
	controlplaneData, err := setup.GetClusterNodesFile("patch.templates/controlplane.yaml")
	if err != nil {
		return fmt.Errorf("failed to get controlplane template: %w", err)
	}

	workerData, err := setup.GetClusterNodesFile("patch.templates/worker.yaml")
	if err != nil {
		return fmt.Errorf("failed to get worker template: %w", err)
	}

	// Write templates
	if err := os.WriteFile(filepath.Join(destDir, "controlplane.yaml"), controlplaneData, 0644); err != nil {
		return fmt.Errorf("failed to write controlplane template: %w", err)
	}

	if err := os.WriteFile(filepath.Join(destDir, "worker.yaml"), workerData, 0644); err != nil {
		return fmt.Errorf("failed to write worker template: %w", err)
	}

	return nil
}

// updateNodeStatus updates node status flags
func (m *Manager) updateNodeStatus(instanceName string, node *Node) error {
	return m.updateNodeStatusFiles(instanceName, node)
}

// updateNodeStatusFiles writes cache (applied/configured/currentIp) and updates maintenance in config (new format)
func (m *Manager) updateNodeStatusFiles(instanceName string, node *Node) error {
	cacheDir := filepath.Dir(tools.GetNodeCachePath(m.dataDir, instanceName, node.Hostname))
	if err := os.MkdirAll(cacheDir, 0755); err != nil {
		return fmt.Errorf("failed to create node cache dir: %w", err)
	}
	cache := nodeCacheFile{
		Applied:    node.Applied,
		Configured: node.Configured,
		CurrentIP:  node.CurrentIP,
	}
	cacheData, err := yaml.Marshal(cache)
	if err != nil {
		return fmt.Errorf("failed to marshal node cache: %w", err)
	}
	if err := os.WriteFile(tools.GetNodeCachePath(m.dataDir, instanceName, node.Hostname), cacheData, 0644); err != nil {
		return fmt.Errorf("failed to write node cache: %w", err)
	}

	// Update maintenance in config/nodes/{hostname}.yaml
	configPath := tools.GetNodeConfigPath(m.dataDir, instanceName, node.Hostname)
	configData, err := os.ReadFile(configPath)
	if err != nil {
		return fmt.Errorf("failed to read node config: %w", err)
	}
	var cfg nodeConfigFile
	if err := yaml.Unmarshal(configData, &cfg); err != nil {
		return fmt.Errorf("failed to parse node config: %w", err)
	}
	cfg.Maintenance = node.Maintenance
	updatedData, err := yaml.Marshal(cfg)
	if err != nil {
		return fmt.Errorf("failed to marshal node config: %w", err)
	}
	return os.WriteFile(configPath, updatedData, 0644)
}

// Update modifies existing node configuration with partial updates
func (m *Manager) Update(instanceName string, hostname string, updates map[string]interface{}) error {
	return m.updateNodeFile(instanceName, hostname, updates)
}

// updateNodeFile applies partial updates to config/nodes/{hostname}.yaml (new format)
func (m *Manager) updateNodeFile(instanceName, hostname string, updates map[string]interface{}) error {
	configPath := tools.GetNodeConfigPath(m.dataDir, instanceName, hostname)
	data, err := os.ReadFile(configPath)
	if err != nil {
		return fmt.Errorf("node %s not found: %w", hostname, err)
	}
	var cfg nodeConfigFile
	if err := yaml.Unmarshal(data, &cfg); err != nil {
		return fmt.Errorf("failed to parse node config: %w", err)
	}

	for key, value := range updates {
		switch key {
		case "target_ip":
			if strVal, ok := value.(string); ok {
				cfg.TargetIP = strVal
			}
		case "current_ip":
			// currentIp is cache; update via cache file (handled by updateNodeStatus)
		case "disk":
			if strVal, ok := value.(string); ok {
				cfg.Disk = strVal
			}
		case "interface":
			if strVal, ok := value.(string); ok {
				cfg.Interface = strVal
			}
		case "version":
			if strVal, ok := value.(string); ok {
				cfg.Version = strVal
			}
		case "schematic_id":
			if strVal, ok := value.(string); ok {
				cfg.SchematicID = strVal
			}
		case "maintenance":
			if boolVal, ok := value.(bool); ok {
				cfg.Maintenance = boolVal
			}
		}
	}

	out, err := yaml.Marshal(cfg)
	if err != nil {
		return fmt.Errorf("failed to marshal node config: %w", err)
	}
	return os.WriteFile(configPath, out, 0644)
}

// FetchTemplates extracts patch templates from embedded files to instance
func (m *Manager) FetchTemplates(instanceName string) error {
	destDir := filepath.Join(tools.GetNodeSetupCachePath(m.dataDir, instanceName), "patch.templates")
	return m.extractEmbeddedTemplates(destDir)
}

// fillTalosDefaults reads cluster talos version/schematicId defaults into node from instance.yaml
func (m *Manager) fillTalosDefaults(instanceName string, node *Node) {
	instanceYaml := tools.GetInstanceYamlPath(m.dataDir, instanceName)
	if data, err := os.ReadFile(instanceYaml); err == nil {
		var root struct {
			Cluster struct {
				Nodes struct {
					Talos struct {
						SchematicID string `yaml:"schematicId"`
						Version     string `yaml:"version"`
					} `yaml:"talos"`
				} `yaml:"nodes"`
			} `yaml:"cluster"`
		}
		if err := yaml.Unmarshal(data, &root); err == nil {
			if node.SchematicID == "" {
				node.SchematicID = root.Cluster.Nodes.Talos.SchematicID
			}
			if node.Version == "" {
				node.Version = root.Cluster.Nodes.Talos.Version
			}
		}
	}
}

// NodeHealth represents the health status of a node
type NodeHealth struct {
	Node        string               `json:"node"`
	Services    []tools.ServiceStatus `json:"services"`
	DmesgErrors []tools.DmesgError   `json:"dmesgErrors"`
	Healthy     bool                 `json:"healthy"`
}

// Health checks node health by querying Talos service statuses and scanning dmesg for errors
func (m *Manager) Health(instanceName, nodeIdentifier string) (*NodeHealth, error) {
	node, err := m.Get(instanceName, nodeIdentifier)
	if err != nil {
		return nil, fmt.Errorf("node not found: %w", err)
	}

	if !node.Applied || node.Maintenance {
		return nil, fmt.Errorf("health check requires an applied, non-maintenance node")
	}

	ip := node.TargetIP
	if ip == "" {
		return nil, fmt.Errorf("no IP address available for node %s", node.Hostname)
	}

	// Fetch services and dmesg concurrently
	var services []tools.ServiceStatus
	var dmesgRaw string
	var svcErr, dmesgErr error

	var wg sync.WaitGroup
	wg.Add(2)

	go func() {
		defer wg.Done()
		services, svcErr = m.talosctl.GetServices(ip)
	}()

	go func() {
		defer wg.Done()
		dmesgRaw, dmesgErr = m.talosctl.GetDmesg(ip)
	}()

	wg.Wait()

	if svcErr != nil {
		return nil, fmt.Errorf("failed to get services: %w", svcErr)
	}

	var dmesgErrors []tools.DmesgError
	if dmesgErr == nil {
		dmesgErrors = tools.ParseDmesgErrors(dmesgRaw)
	}
	if dmesgErrors == nil {
		dmesgErrors = []tools.DmesgError{}
	}

	// Compute overall health
	healthy := len(dmesgErrors) == 0
	for _, svc := range services {
		if !svc.Healthy && svc.HealthMessage != "" {
			healthy = false
			break
		}
	}

	return &NodeHealth{
		Node:        node.Hostname,
		Services:    services,
		DmesgErrors: dmesgErrors,
		Healthy:     healthy,
	}, nil
}

// Reboot restarts a node without wiping state
func (m *Manager) Reboot(instanceName, nodeIdentifier string) error {
	slog.Info("rebooting node", "component", "node", "instance", instanceName, "node", nodeIdentifier)

	node, err := m.Get(instanceName, nodeIdentifier)
	if err != nil {
		return fmt.Errorf("node not found: %w", err)
	}

	rebootIP := node.TargetIP
	if rebootIP == "" {
		rebootIP = node.CurrentIP
	}
	if rebootIP == "" {
		return fmt.Errorf("no IP address available for node %s", node.Hostname)
	}

	if err := m.talosctl.Reboot(rebootIP); err != nil {
		return fmt.Errorf("failed to reboot node: %w", err)
	}

	slog.Info("node reboot initiated", "component", "node", "instance", instanceName, "hostname", node.Hostname, "ip", rebootIP)
	return nil
}

// Reset resets a node to maintenance mode with resilient handling.
// For control plane nodes, it first removes the etcd member from the cluster
// via a healthy peer, then resets the node. The node stays in config.yaml
// so it can be reconfigured and rejoined.
func (m *Manager) Reset(instanceName, nodeIdentifier string) error {
	slog.Info("resetting node", "component", "node", "instance", instanceName, "node", nodeIdentifier)

	// Get node
	node, err := m.Get(instanceName, nodeIdentifier)
	if err != nil {
		return fmt.Errorf("node not found: %w", err)
	}

	// Determine IP to reset
	resetIP := node.CurrentIP
	if resetIP == "" {
		resetIP = node.TargetIP
	}
	if resetIP == "" {
		return fmt.Errorf("no IP address available for node %s", node.Hostname)
	}

	// For control plane nodes, remove from etcd first via a healthy peer
	if node.Role == "controlplane" {
		if err := m.removeEtcdMember(instanceName, node); err != nil {
			// Log but don't fail — the node may already be removed from etcd,
			// or etcd may be unreachable on this node. The reset should still proceed.
			slog.Warn("etcd member removal failed (continuing with reset)", "component", "node",
				"instance", instanceName, "hostname", node.Hostname, "error", err)
		}
	}

	// Reset the node via talosctl
	if err := m.talosctl.Reset(resetIP); err != nil {
		return fmt.Errorf("failed to reset node: %w", err)
	}

	// Update node status to maintenance mode but keep in config for reconfiguration
	node.Maintenance = true
	node.Configured = false
	node.Applied = false
	node.CurrentIP = ""
	node.Version = ""
	if err := m.updateNodeStatus(instanceName, node); err != nil {
		return fmt.Errorf("failed to update node status: %w", err)
	}

	slog.Info("node reset to maintenance mode", "component", "node",
		"instance", instanceName, "hostname", node.Hostname)
	return nil
}

// removeEtcdMember removes a control plane node's etcd member via a healthy peer.
func (m *Manager) removeEtcdMember(instanceName string, targetNode *Node) error {
	// Find a healthy control plane peer to run the etcd remove-member command
	nodes, err := m.List(instanceName)
	if err != nil {
		return fmt.Errorf("failed to list nodes: %w", err)
	}

	for _, peer := range nodes {
		// Skip the target node itself, non-control-plane nodes, and unapplied nodes
		if peer.Hostname == targetNode.Hostname || peer.Role != "controlplane" || !peer.Applied || peer.Maintenance {
			continue
		}

		peerIP := peer.CurrentIP
		if peerIP == "" {
			peerIP = peer.TargetIP
		}
		if peerIP == "" {
			continue
		}

		slog.Info("removing etcd member via peer", "component", "node",
			"instance", instanceName, "target", targetNode.Hostname, "peer", peer.Hostname)

		if err := m.talosctl.EtcdRemoveMember(peerIP, targetNode.Hostname); err != nil {
			slog.Warn("etcd remove-member failed via peer, trying next", "component", "node",
				"peer", peer.Hostname, "error", err)
			continue
		}

		slog.Info("etcd member removed successfully", "component", "node",
			"instance", instanceName, "target", targetNode.Hostname, "via", peer.Hostname)
		return nil
	}

	return fmt.Errorf("no healthy control plane peer available to remove etcd member for %s", targetNode.Hostname)
}