deploy-felhom-compose/controller/internal/setup/scanner.go

package setup

import (
	"bufio"
	"context"
	"encoding/json"
	"fmt"
	"log"
	"os"
	"os/exec"
	"path/filepath"
	"strings"
	"time"

	"gitea.dooplex.hu/admin/felhom-controller/internal/backup"
)

// DriveBackup represents a found infra backup on a drive.
type DriveBackup struct {
	Device         string   `json:"device"`
	Label          string   `json:"label"`
	MountPoint     string   `json:"mount_point"`
	CustomerID     string   `json:"customer_id"`
	Timestamp      string   `json:"timestamp"`
	CtrlVersion    string   `json:"controller_version"`
	IntegrityOK    bool     `json:"integrity_ok"`
	Error          string   `json:"error,omitempty"`
	StackCount     int      `json:"stack_count"`
	StackNames     []string `json:"stack_names,omitempty"`
	DiskCount      int      `json:"disk_count"`
	IsHistory      bool     `json:"is_history"`
	HistoryFile    string   `json:"history_file,omitempty"`
	WasTempMounted bool     `json:"-"`
}

// lsblkOutput represents the JSON output of lsblk.
type lsblkOutput struct {
	Blockdevices []lsblkDevice `json:"blockdevices"`
}

type lsblkDevice struct {
	Name       string         `json:"name"`
	Path       string         `json:"path"`
	FSType     *string        `json:"fstype"`
	MountPoint *string        `json:"mountpoint"`
	Label      *string        `json:"label"`
	Size       interface{}    `json:"size"` // string or int
	Type       string         `json:"type"` // "disk", "part"
	Children   []lsblkDevice  `json:"children,omitempty"`
}

// ScanDrivesForInfraBackups scans all block devices for .felhom-infra-backup/ directories.
func ScanDrivesForInfraBackups(logger *log.Logger, debug bool) ([]DriveBackup, error) {
	logger.Printf("[INFO] Setup: scanning drives for infra backups...")

	// Read currently mounted filesystems
	mountedFS := readMountedFilesystems()

	// Get root device to skip
	rootDevices := getRootDevices()

	// Run lsblk
	ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
	defer cancel()

	out, err := exec.CommandContext(ctx, "lsblk", "-J", "-o", "NAME,PATH,FSTYPE,MOUNTPOINT,LABEL,SIZE,TYPE").Output()
	if err != nil {
		return nil, fmt.Errorf("lsblk failed: %w", err)
	}

	var lsblk lsblkOutput
	if err := json.Unmarshal(out, &lsblk); err != nil {
		return nil, fmt.Errorf("parsing lsblk: %w", err)
	}

	if debug {
		logger.Printf("[DEBUG] Setup scan: lsblk returned %d block devices", len(lsblk.Blockdevices))
	}

	var results []DriveBackup

	// Flatten all partitions
	var partitions []lsblkDevice
	for _, disk := range lsblk.Blockdevices {
		if disk.Type == "part" {
			partitions = append(partitions, disk)
		}
		for _, child := range disk.Children {
			if child.Type == "part" {
				partitions = append(partitions, child)
			}
		}
	}

	if debug {
		logger.Printf("[DEBUG] Setup scan: found %d partitions to check, %d root devices to skip", len(partitions), len(rootDevices))
	}

	for _, part := range partitions {
		// Skip partitions without filesystem
		if part.FSType == nil || *part.FSType == "" || *part.FSType == "swap" {
			continue
		}

		// Skip LUKS encrypted partitions
		if *part.FSType == "crypto_LUKS" {
			logger.Printf("[DEBUG] Setup: skipping LUKS partition %s", part.Path)
			continue
		}

		// Skip LVM
		if part.Type == "lvm" {
			logger.Printf("[DEBUG] Setup: skipping LVM volume %s", part.Path)
			continue
		}

		// Skip root partitions
		if isRootPartition(part.Path, rootDevices) {
			continue
		}

		partResults := scanPartition(part, mountedFS, logger)
		results = append(results, partResults...)
	}

	logger.Printf("[INFO] Setup: drive scan complete — found %d backup(s)", countValid(results))
	return results, nil
}

// CleanupTempMounts unmounts any partitions that were temporarily mounted during scanning.
func CleanupTempMounts(results []DriveBackup, logger *log.Logger) {
	for _, r := range results {
		if r.WasTempMounted && r.MountPoint != "" {
			ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
			exec.CommandContext(ctx, "umount", r.MountPoint).Run()
			cancel()
			os.Remove(r.MountPoint)
			logger.Printf("[DEBUG] Setup: unmounted temp mount %s", r.MountPoint)
		}
	}
}

func scanPartition(part lsblkDevice, mountedFS map[string]string, logger *log.Logger) []DriveBackup {
	label := ""
	if part.Label != nil {
		label = *part.Label
	}

	// Check if already mounted
	var mountPoint string
	var tempMounted bool

	if part.MountPoint != nil && *part.MountPoint != "" {
		mountPoint = *part.MountPoint
	} else if mp, ok := mountedFS[part.Path]; ok {
		mountPoint = mp
	} else {
		// Try to mount temporarily
		tmpDir := filepath.Join("/mnt", ".felhom-scan", part.Name)
		if err := os.MkdirAll(tmpDir, 0700); err != nil {
			logger.Printf("[DEBUG] Setup: skip %s — cannot create temp dir: %v", part.Path, err)
			return nil
		}

		ctx, cancel := context.WithTimeout(context.Background(), 30*time.Second)
		defer cancel()

		// Try read-only mount
		err := exec.CommandContext(ctx, "mount", "-o", "ro", part.Path, tmpDir).Run()
		if err != nil {
			// Retry with noload for journal errors
			err = exec.CommandContext(ctx, "mount", "-o", "ro,noload", part.Path, tmpDir).Run()
		}
		if err != nil {
			os.Remove(tmpDir)
			logger.Printf("[DEBUG] Setup: skip %s — mount failed: %v", part.Path, err)
			return nil
		}
		mountPoint = tmpDir
		tempMounted = true
	}

	// Check for .felhom-infra-backup/
	infraDir := backup.InfraBackupDir(mountPoint)
	if _, err := os.Stat(infraDir); os.IsNotExist(err) {
		if tempMounted {
			exec.Command("umount", mountPoint).Run()
			os.Remove(mountPoint)
		}
		return nil
	}

	var results []DriveBackup

	// Read current backup
	backupData, meta, err := backup.ReadLocalInfraBackup(mountPoint)

	current := DriveBackup{
		Device:         part.Path,
		Label:          label,
		MountPoint:     mountPoint,
		WasTempMounted: tempMounted,
	}

	if err != nil {
		current.IntegrityOK = false
		current.Error = err.Error()
		if meta != nil {
			current.CustomerID = meta.CustomerID
			current.Timestamp = meta.Timestamp
			current.CtrlVersion = meta.ControllerVersion
		}
	} else {
		current.IntegrityOK = true
		current.CustomerID = meta.CustomerID
		current.Timestamp = meta.Timestamp
		current.CtrlVersion = meta.ControllerVersion
		backup.ParseBackupCounts(backupData, &current.StackCount, &current.StackNames, &current.DiskCount)
	}

	results = append(results, current)

	logger.Printf("[INFO] Setup: found infra backup on %s (%s) — customer=%s, integrity=%v",
		part.Path, label, current.CustomerID, current.IntegrityOK)

	// Also scan history directory for older versions
	history := backup.ReadLocalInfraHistory(mountPoint)
	for _, hv := range history {
		hResult := DriveBackup{
			Device:      part.Path,
			Label:       label,
			MountPoint:  mountPoint,
			CustomerID:  hv.CustomerID,
			Timestamp:   hv.Timestamp,
			CtrlVersion: hv.ControllerVersion,
			IntegrityOK: hv.IntegrityOK,
			Error:       hv.Error,
			StackCount:  hv.StackCount,
			StackNames:  hv.StackNames,
			DiskCount:   hv.DiskCount,
			IsHistory:   true,
			HistoryFile: hv.HistoryFile,
		}
		results = append(results, hResult)
	}

	if len(history) > 0 {
		logger.Printf("[INFO] Setup: found %d historical backup version(s) on %s", len(history), part.Path)
	}

	return results
}

func readMountedFilesystems() map[string]string {
	result := make(map[string]string)

	f, err := os.Open("/proc/mounts")
	if err != nil {
		return result
	}
	defer f.Close()

	scanner := bufio.NewScanner(f)
	for scanner.Scan() {
		fields := strings.Fields(scanner.Text())
		if len(fields) >= 2 {
			result[fields[0]] = fields[1]
		}
	}
	return result
}

func getRootDevices() map[string]bool {
	result := make(map[string]bool)
	mountedFS := readMountedFilesystems()
	for dev, mp := range mountedFS {
		if mp == "/" || mp == "/boot" || mp == "/boot/efi" {
			result[dev] = true
		}
	}
	return result
}

func isRootPartition(devPath string, rootDevices map[string]bool) bool {
	return rootDevices[devPath]
}

func countValid(results []DriveBackup) int {
	n := 0
	for _, r := range results {
		if r.IntegrityOK {
			n++
		}
	}
	return n
}

// runDriveScan runs the scan asynchronously and stores results on the Server.
func (s *Server) runDriveScan() {
	results, err := ScanDrivesForInfraBackups(s.logger, s.isDebug())

	// Clean up any temporary mounts created during scan
	if results != nil {
		CleanupTempMounts(results, s.logger)
	}

	s.scanMu.Lock()
	defer s.scanMu.Unlock()

	s.scanRunning = false
	s.scanDone = true
	if err != nil {
		s.scanError = err.Error()
	} else {
		s.scanResults = results
	}
}