Files
deploy-felhom-compose/controller/internal/monitor/watchdog.go
T
admin db83db383c fix: deep bug hunt II — concurrency, security & optimization (25 files)
Critical: watchdog mutex panic safety, SetGeoAppOverride nil guard,
SSD-only app DB restore fallback.

High: double deploy race (atomic Deploying flag), delete/remove during
deploy guard, ScanStacks overwrite protection, FileBrowser mount mutex,
PushEvent history, PushOnce error handling, DB dump sync+close before
rename, restic retry fresh context, encrypt failure logging, cross-backup
path traversal validation, deepCopyStack completeness.

Security: constant-time API key comparison, login rate limiting (5/min),
git credential masking in logs, storage path prefix traversal fix.

Concurrency: MigrateEncryption lock ordering, SubdomainInUse I/O outside
lock, scheduler late-registered jobs, SQLite WAL verification, metrics
shutdown context, telemetry scan error logging, asset sync lock scope.

Optimization: streaming file copy for DB dumps, restic stats dedup,
atomic infra config copy.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-25 14:21:09 +01:00

903 lines
27 KiB
Go

package monitor
import (
"context"
"fmt"
"log"
"os"
"os/exec"
"path/filepath"
"strings"
"sync"
"time"
"gitea.dooplex.hu/admin/felhom-controller/internal/config"
"gitea.dooplex.hu/admin/felhom-controller/internal/notify"
"gitea.dooplex.hu/admin/felhom-controller/internal/settings"
"gitea.dooplex.hu/admin/felhom-controller/internal/system"
)
const (
// probeThreshold is the number of consecutive probe failures before declaring disconnected.
probeThreshold = 3
// defaultProbeInterval is the normal probe interval for connected drives.
defaultProbeInterval = 5 * time.Second
// disconnectedProbeInterval is the slower probe interval for disconnected drives
// (checking for UUID reappearance, not I/O probing).
disconnectedProbeInterval = 30 * time.Second
// hostFstabPath is where the host's fstab is mounted inside the container.
hostFstabPath = "/host-fstab"
// hostDevUUIDPath is where the host's /dev/disk/by-uuid is accessible.
hostDevUUIDPath = "/host-dev/disk/by-uuid"
// primaryResticSubpath is the relative path to the primary restic repo under a drive.
primaryResticSubpath = "backups/primary/restic"
)
// WatchdogStackInfo holds minimal stack info for the watchdog.
type WatchdogStackInfo struct {
Name string
}
// WatchdogStackProvider provides stack operations needed by the watchdog.
// Defined here to avoid circular imports with the backup package.
type WatchdogStackProvider interface {
ListDeployedStacks() []WatchdogStackInfo
GetStackHDDPath(name string) string
StopStack(name string) error
StartStack(name string) error
}
// pathProbeState tracks in-memory probe state for a single storage path.
type pathProbeState struct {
mu sync.Mutex
consecutiveFailures int
lastStatus string // "connected", "disconnected"
lastProbeTime time.Time
probeInterval time.Duration
// Debug counters for summary logging
probeCount int
probeOKCount int
lastSummaryTime time.Time
totalLatency time.Duration
}
// StorageWatchdog monitors registered storage paths and reacts to disconnection/reconnection.
type StorageWatchdog struct {
settings *settings.Settings
stackProvider WatchdogStackProvider
notifier *notify.Notifier
cfg *config.Config
logger *log.Logger
// Callbacks to break import cycles — set via SetXxx methods after construction
alertRefresh func()
pushHubReport func()
unlockRepo func(ctx context.Context, repoPath string) error
mu sync.Mutex
pathState map[string]*pathProbeState
// Debug simulation state
simulatedMu sync.RWMutex
simulatedPaths map[string]bool
}
// NewStorageWatchdog creates a new storage watchdog.
func NewStorageWatchdog(
sett *settings.Settings,
stackProvider WatchdogStackProvider,
notifier *notify.Notifier,
cfg *config.Config,
logger *log.Logger,
) *StorageWatchdog {
return &StorageWatchdog{
settings: sett,
stackProvider: stackProvider,
notifier: notifier,
cfg: cfg,
logger: logger,
pathState: make(map[string]*pathProbeState),
simulatedPaths: make(map[string]bool),
}
}
// isDebug returns true if the logging level is set to "debug".
func (w *StorageWatchdog) isDebug() bool { return w.cfg.Logging.Level == "debug" }
// SetAlertRefresh sets the callback to trigger alert refresh.
func (w *StorageWatchdog) SetAlertRefresh(fn func()) {
w.alertRefresh = fn
}
// SetHubReportPusher sets the callback to push an immediate hub report.
func (w *StorageWatchdog) SetHubReportPusher(fn func()) {
w.pushHubReport = fn
}
// SetRepoUnlocker sets the callback to unlock a restic repo on reconnect.
func (w *StorageWatchdog) SetRepoUnlocker(fn func(ctx context.Context, repoPath string) error) {
w.unlockRepo = fn
}
// Check probes all registered storage paths and reacts to state changes.
// Called by the scheduler every 5 seconds.
func (w *StorageWatchdog) Check(ctx context.Context) error {
paths := w.settings.GetStoragePaths()
if len(paths) == 0 {
return nil
}
for _, sp := range paths {
select {
case <-ctx.Done():
return ctx.Err()
default:
}
state := w.getOrCreateState(sp.Path)
// Rate-limit per-path probes
state.mu.Lock()
if time.Since(state.lastProbeTime) < state.probeInterval {
state.mu.Unlock()
continue
}
state.lastProbeTime = time.Now()
state.mu.Unlock()
// Skip decommissioned drives entirely — no apps reference them
if sp.Decommissioned {
continue
}
// Skip simulated-disconnected paths (handled by debug UI)
if w.isSimulated(sp.Path) {
continue
}
if sp.Disconnected {
w.handleReconnectCheck(ctx, sp)
} else {
w.handleConnectedProbe(sp, state)
}
}
return nil
}
// getOrCreateState returns the in-memory probe state for a path, creating if needed.
func (w *StorageWatchdog) getOrCreateState(path string) *pathProbeState {
w.mu.Lock()
defer w.mu.Unlock()
if s, ok := w.pathState[path]; ok {
return s
}
s := &pathProbeState{
lastStatus: "connected",
probeInterval: defaultProbeInterval,
}
w.pathState[path] = s
return s
}
// handleConnectedProbe probes a connected drive and triggers disconnect if needed.
func (w *StorageWatchdog) handleConnectedProbe(sp settings.StoragePath, state *pathProbeState) {
probeStart := time.Now()
result := system.ProbeStoragePath(sp.Path)
probeLatency := time.Since(probeStart)
state.mu.Lock()
defer state.mu.Unlock()
if w.isDebug() {
state.probeCount++
state.totalLatency += probeLatency
}
if result.Status == system.ProbeConnected {
if state.consecutiveFailures > 0 {
w.logger.Printf("[DEBUG] [STORAGE] Probe recovered for %s after %d failures", sp.Path, state.consecutiveFailures)
}
state.consecutiveFailures = 0
state.lastStatus = "connected"
if w.isDebug() {
state.probeOKCount++
// Every 60 probes (~5 minutes at 5s interval): emit summary
if state.probeCount >= 60 {
avgLatency := state.totalLatency / time.Duration(state.probeCount)
w.logger.Printf("[DEBUG] [STORAGE] Storage watchdog: %s — %d/%d probes OK (last 5m, avg %dms)",
sp.Path, state.probeOKCount, state.probeCount, avgLatency.Milliseconds())
state.probeCount = 0
state.probeOKCount = 0
state.totalLatency = 0
state.lastSummaryTime = time.Now()
}
}
return
}
state.consecutiveFailures++
// Debug: log immediately on unexpected failure (was connected, now failing)
if w.isDebug() && state.lastStatus == "connected" {
w.logger.Printf("[DEBUG] [STORAGE] Storage probe failed for %s (%d/%d before disconnect): %v",
sp.Path, state.consecutiveFailures, probeThreshold, result.Err)
}
w.logger.Printf("[WARN] [STORAGE] Probe failed for %s (%d/%d): %v",
sp.Path, state.consecutiveFailures, probeThreshold, result.Err)
if state.consecutiveFailures >= probeThreshold {
// Release state.mu before calling handleDisconnect (which re-acquires it
// internally). Re-acquire afterwards so the deferred Unlock stays balanced.
// Wrap in a func to guarantee re-lock even if handleDisconnect panics.
func() {
state.mu.Unlock()
defer state.mu.Lock()
w.handleDisconnect(sp, state, result)
}()
}
}
// handleDisconnect reacts to a confirmed drive disconnection.
func (w *StorageWatchdog) handleDisconnect(sp settings.StoragePath, state *pathProbeState, probe system.ProbeResult) {
label := sp.Label
if label == "" {
label = sp.Path
}
w.logger.Printf("[ERROR] [STORAGE] Drive disconnected: %s (%s)", sp.Path, label)
// 1. Find and stop affected stacks
stoppedStacks := w.stopAffectedStacks(sp.Path)
// 2. Mark disconnected in settings (persists to settings.json)
if err := w.settings.SetDisconnected(sp.Path, true, stoppedStacks); err != nil {
w.logger.Printf("[ERROR] [STORAGE] Failed to mark disconnected: %v", err)
}
// 3. Lazy unmount stale mount (if probe timed out — mount is likely hanging)
if probe.Status == system.ProbeTimeout {
w.lazyUnmount(sp.Path)
}
// 4. Update in-memory state
state.mu.Lock()
state.lastStatus = "disconnected"
state.probeInterval = disconnectedProbeInterval
state.consecutiveFailures = 0
state.mu.Unlock()
// 5. Trigger alert refresh
if w.alertRefresh != nil {
w.alertRefresh()
}
// 6. Send notification
w.notifier.NotifyStorageDisconnected(label, stoppedStacks)
// 7. Push immediate hub report
if w.pushHubReport != nil {
go w.pushHubReport()
}
}
// handleReconnectCheck checks if a disconnected drive has been reconnected.
func (w *StorageWatchdog) handleReconnectCheck(ctx context.Context, sp settings.StoragePath) {
// Find the UUID for this path from fstab
// For attach-wizard drives, the UUID is on the raw mount, not the bind mount
mountPath := sp.Path
rawPath, isAttachWizard := system.HasFelhomRawMount(hostFstabPath, sp.Path)
if isAttachWizard {
mountPath = rawPath
}
uuid := system.ParseFstabUUID(hostFstabPath, mountPath)
if uuid == "" {
// No UUID in fstab — can't detect reconnection automatically
return
}
if w.isDebug() {
w.logger.Printf("[DEBUG] [STORAGE] Reconnect check for %s: UUID=%s, mountPath=%s, isAttachWizard=%v",
sp.Path, uuid, mountPath, isAttachWizard)
}
// Check if the UUID block device is present
uuidPath := filepath.Join(hostDevUUIDPath, uuid)
if _, err := os.Stat(uuidPath); err != nil {
return // Drive not reconnected yet
}
label := sp.Label
if label == "" {
label = sp.Path
}
w.logger.Printf("[INFO] [STORAGE] Drive reconnected (UUID found), attempting remount: %s (%s)", sp.Path, label)
if w.isDebug() {
w.logger.Printf("[DEBUG] [STORAGE] UUID %s found at %s, mounting %s (raw=%s, attachWizard=%v)",
uuid, uuidPath, sp.Path, rawPath, isAttachWizard)
}
// Attempt remount
if err := w.remount(sp.Path, rawPath, isAttachWizard); err != nil {
w.logger.Printf("[ERROR] [STORAGE] Remount failed for %s: %v", sp.Path, err)
return // Try again next cycle
}
// Verify with a probe
verifyResult := system.ProbeStoragePath(sp.Path)
if verifyResult.Status != system.ProbeConnected {
w.logger.Printf("[ERROR] [STORAGE] Post-remount probe failed for %s: %v", sp.Path, verifyResult.Err)
if w.isDebug() {
w.logger.Printf("[DEBUG] [STORAGE] Post-mount verification failed for %s: status=%v, err=%v",
sp.Path, verifyResult.Status, verifyResult.Err)
}
return
}
if w.isDebug() {
w.logger.Printf("[DEBUG] [STORAGE] Post-mount verification succeeded for %s", sp.Path)
}
w.logger.Printf("[INFO] [STORAGE] Drive successfully remounted: %s (%s)", sp.Path, label)
// Clean stale restic locks
w.cleanResticLocks(ctx, sp.Path)
// Validate stopped stacks — filter to only actually stopped ones
filteredStacks := w.filterStoppedStacks(sp.StoppedStacks)
// Clear disconnected but preserve StoppedStacks for the restart UI
if err := w.settings.SetDisconnected(sp.Path, false, filteredStacks); err != nil {
w.logger.Printf("[ERROR] [STORAGE] Failed to clear disconnected: %v", err)
}
// Update in-memory state
state := w.getOrCreateState(sp.Path)
state.mu.Lock()
state.lastStatus = "connected"
state.probeInterval = defaultProbeInterval
state.consecutiveFailures = 0
state.mu.Unlock()
// Trigger alert refresh
if w.alertRefresh != nil {
w.alertRefresh()
}
// Send notification
w.notifier.NotifyStorageReconnected(label)
// Push immediate hub report
if w.pushHubReport != nil {
go w.pushHubReport()
}
}
// stopAffectedStacks stops all deployed stacks whose HDD_PATH matches the disconnected drive.
func (w *StorageWatchdog) stopAffectedStacks(drivePath string) []string {
if w.stackProvider == nil {
return nil
}
var stopped []string
cleanDrive := filepath.Clean(drivePath)
for _, stack := range w.stackProvider.ListDeployedStacks() {
hddPath := w.stackProvider.GetStackHDDPath(stack.Name)
if hddPath == "" {
continue
}
cleanHDD := filepath.Clean(hddPath)
if cleanHDD != cleanDrive && !strings.HasPrefix(cleanHDD, cleanDrive+"/") {
continue
}
// Don't stop protected stacks
if w.cfg.IsProtectedStack(stack.Name) {
w.logger.Printf("[WARN] [STORAGE] Skipping protected stack: %s", stack.Name)
continue
}
w.logger.Printf("[INFO] [STORAGE] Stopping stack %s (drive disconnected: %s)", stack.Name, drivePath)
if err := w.stackProvider.StopStack(stack.Name); err != nil {
w.logger.Printf("[ERROR] [STORAGE] Failed to stop stack %s: %v", stack.Name, err)
continue // Don't add to stopped list if stop failed
}
stopped = append(stopped, stack.Name)
}
if len(stopped) > 0 {
w.logger.Printf("[INFO] [STORAGE] Stopped %d stack(s) due to drive disconnect: %v", len(stopped), stopped)
}
return stopped
}
// lazyUnmount performs a lazy unmount of a path and its raw mount (if attach-wizard).
func (w *StorageWatchdog) lazyUnmount(path string) {
// For attach-wizard, unmount bind first, then raw
rawPath, isAttachWizard := system.HasFelhomRawMount(hostFstabPath, path)
// Unmount the bind/main path
cmd := exec.Command("umount", "-l", path)
if out, err := cmd.CombinedOutput(); err != nil {
w.logger.Printf("[WARN] [STORAGE] umount -l %s: %v (%s)", path, err, strings.TrimSpace(string(out)))
} else {
w.logger.Printf("[INFO] [STORAGE] Lazy unmounted: %s", path)
}
// Then unmount the raw path if it's an attach-wizard drive
if isAttachWizard && rawPath != "" {
cmd = exec.Command("umount", "-l", rawPath)
if out, err := cmd.CombinedOutput(); err != nil {
w.logger.Printf("[WARN] [STORAGE] umount -l %s: %v (%s)", rawPath, err, strings.TrimSpace(string(out)))
} else {
w.logger.Printf("[INFO] [STORAGE] Lazy unmounted raw: %s", rawPath)
}
}
}
// remount attempts to remount a storage path using fstab entries.
func (w *StorageWatchdog) remount(path, rawPath string, isAttachWizard bool) error {
// Clean any stale mount entries first
exec.Command("umount", "-l", path).Run()
if isAttachWizard && rawPath != "" {
exec.Command("umount", "-l", rawPath).Run()
}
if isAttachWizard && rawPath != "" {
// Mount raw first, then bind
cmd := exec.Command("mount", "-T", hostFstabPath, rawPath)
if out, err := cmd.CombinedOutput(); err != nil {
return fmt.Errorf("mount raw %s: %v (%s)", rawPath, err, strings.TrimSpace(string(out)))
}
w.logger.Printf("[INFO] [STORAGE] Mounted raw: %s", rawPath)
cmd = exec.Command("mount", "-T", hostFstabPath, path)
if out, err := cmd.CombinedOutput(); err != nil {
return fmt.Errorf("mount bind %s: %v (%s)", path, err, strings.TrimSpace(string(out)))
}
w.logger.Printf("[INFO] [STORAGE] Mounted bind: %s", path)
} else {
cmd := exec.Command("mount", "-T", hostFstabPath, path)
if out, err := cmd.CombinedOutput(); err != nil {
return fmt.Errorf("mount %s: %v (%s)", path, err, strings.TrimSpace(string(out)))
}
w.logger.Printf("[INFO] [STORAGE] Mounted: %s", path)
}
return nil
}
// cleanResticLocks runs restic unlock on the primary repo for a drive path.
func (w *StorageWatchdog) cleanResticLocks(ctx context.Context, drivePath string) {
repoPath := filepath.Join(drivePath, primaryResticSubpath)
locksDir := filepath.Join(repoPath, "locks")
entries, err := os.ReadDir(locksDir)
if err != nil || len(entries) == 0 {
return // No locks dir or no lock files
}
w.logger.Printf("[INFO] [STORAGE] Found %d restic lock file(s) in %s, running unlock", len(entries), repoPath)
if w.unlockRepo != nil {
if err := w.unlockRepo(ctx, repoPath); err != nil {
w.logger.Printf("[WARN] [STORAGE] Restic unlock failed for %s: %v", repoPath, err)
}
}
}
// filterStoppedStacks validates that stacks in the list still exist as deployed stacks.
func (w *StorageWatchdog) filterStoppedStacks(stackNames []string) []string {
if w.stackProvider == nil || len(stackNames) == 0 {
return nil
}
deployed := make(map[string]bool)
for _, s := range w.stackProvider.ListDeployedStacks() {
deployed[s.Name] = true
}
var result []string
for _, name := range stackNames {
if deployed[name] {
result = append(result, name)
}
}
return result
}
// SafeDisconnect performs a safe disconnect of a storage path.
// Stops affected apps, syncs filesystem, and unmounts the drive.
func (w *StorageWatchdog) SafeDisconnect(ctx context.Context, path string) (stoppedStacks []string, err error) {
sp := w.findStoragePath(path)
if sp == nil {
return nil, fmt.Errorf("storage path %q not found", path)
}
if sp.Disconnected {
return nil, fmt.Errorf("drive already disconnected")
}
if sp.Decommissioned {
return nil, fmt.Errorf("drive is decommissioned — no apps to stop")
}
label := sp.Label
if label == "" {
label = sp.Path
}
w.logger.Printf("[INFO] [STORAGE] Safe disconnect requested: %s (%s)", path, label)
// 1. Stop affected stacks
stoppedStacks = w.stopAffectedStacks(path)
// 2. Sync filesystem
exec.Command("sync").Run()
// 3. Unmount
rawPath, isAttachWizard := system.HasFelhomRawMount(hostFstabPath, path)
// Unmount bind/main
cmd := exec.Command("umount", path)
if out, umountErr := cmd.CombinedOutput(); umountErr != nil {
// Try lazy unmount as fallback
w.logger.Printf("[WARN] [STORAGE] umount %s failed, trying lazy: %v", path, umountErr)
cmd = exec.Command("umount", "-l", path)
if out, umountErr = cmd.CombinedOutput(); umountErr != nil {
return stoppedStacks, fmt.Errorf("umount %s failed: %v (%s)", path, umountErr, strings.TrimSpace(string(out)))
}
}
// Unmount raw if attach-wizard
if isAttachWizard && rawPath != "" {
cmd = exec.Command("umount", rawPath)
if out, umountErr := cmd.CombinedOutput(); umountErr != nil {
cmd = exec.Command("umount", "-l", rawPath)
if out, umountErr = cmd.CombinedOutput(); umountErr != nil {
w.logger.Printf("[WARN] [STORAGE] umount raw %s failed: %v (%s)", rawPath, umountErr, strings.TrimSpace(string(out)))
}
}
}
// 4. Mark disconnected
if setErr := w.settings.SetDisconnected(path, true, stoppedStacks); setErr != nil {
w.logger.Printf("[ERROR] [STORAGE] Failed to mark disconnected: %v", setErr)
}
// 5. Update in-memory state
state := w.getOrCreateState(path)
state.mu.Lock()
state.lastStatus = "disconnected"
state.probeInterval = disconnectedProbeInterval
state.consecutiveFailures = 0
state.mu.Unlock()
// 6. Trigger alert refresh
if w.alertRefresh != nil {
w.alertRefresh()
}
// 7. Notify and push hub report
w.notifier.Notify("storage_safe_disconnect", "info",
fmt.Sprintf("Meghajtó biztonságosan leválasztva: %s", label), "")
if w.pushHubReport != nil {
go w.pushHubReport()
}
w.logger.Printf("[INFO] [STORAGE] Safe disconnect completed: %s — drive can be removed", path)
return stoppedStacks, nil
}
// Reconnect attempts to remount a disconnected storage path.
func (w *StorageWatchdog) Reconnect(ctx context.Context, path string) (stoppedStacks []string, err error) {
sp := w.findStoragePath(path)
if sp == nil {
return nil, fmt.Errorf("storage path %q not found", path)
}
if !sp.Disconnected {
return nil, fmt.Errorf("drive is not disconnected")
}
label := sp.Label
if label == "" {
label = sp.Path
}
// Check UUID availability
mountPath := sp.Path
rawPath, isAttachWizard := system.HasFelhomRawMount(hostFstabPath, sp.Path)
if isAttachWizard {
mountPath = rawPath
}
uuid := system.ParseFstabUUID(hostFstabPath, mountPath)
if uuid != "" {
uuidPath := filepath.Join(hostDevUUIDPath, uuid)
if _, statErr := os.Stat(uuidPath); statErr != nil {
return nil, fmt.Errorf("drive not detected (UUID %s not found) — ensure the drive is physically connected", uuid)
}
}
// Attempt remount
if mountErr := w.remount(path, rawPath, isAttachWizard); mountErr != nil {
return nil, fmt.Errorf("mount failed: %w", mountErr)
}
// Verify
verifyResult := system.ProbeStoragePath(path)
if verifyResult.Status != system.ProbeConnected {
return nil, fmt.Errorf("mount appeared to succeed but probe failed: %v", verifyResult.Err)
}
// Clean restic locks
w.cleanResticLocks(ctx, path)
// Validate stopped stacks
filteredStacks := w.filterStoppedStacks(sp.StoppedStacks)
// Clear disconnected, preserve stopped stacks for restart UI
if setErr := w.settings.SetDisconnected(path, false, filteredStacks); setErr != nil {
w.logger.Printf("[ERROR] [STORAGE] Failed to clear disconnected: %v", setErr)
}
// Update in-memory state
state := w.getOrCreateState(path)
state.mu.Lock()
state.lastStatus = "connected"
state.probeInterval = defaultProbeInterval
state.consecutiveFailures = 0
state.mu.Unlock()
// Trigger alert refresh
if w.alertRefresh != nil {
w.alertRefresh()
}
// Notify
w.notifier.NotifyStorageReconnected(label)
if w.pushHubReport != nil {
go w.pushHubReport()
}
w.logger.Printf("[INFO] [STORAGE] Reconnect completed: %s", path)
return filteredStacks, nil
}
// RestartStoppedApps restarts apps that were auto-stopped due to a drive disconnect.
func (w *StorageWatchdog) RestartStoppedApps(path string) (started, failed []string) {
sp := w.findStoragePath(path)
if sp == nil || sp.Disconnected {
return nil, nil
}
stacks := w.settings.GetStoppedStacks(path)
if len(stacks) == 0 {
return nil, nil
}
for _, name := range stacks {
w.logger.Printf("[INFO] [STORAGE] Starting stack %s (drive reconnected: %s)", name, path)
if err := w.stackProvider.StartStack(name); err != nil {
w.logger.Printf("[ERROR] [STORAGE] Failed to start stack %s: %v", name, err)
failed = append(failed, name)
} else {
started = append(started, name)
}
}
// Clear stopped stacks list
if err := w.settings.ClearStoppedStacks(path); err != nil {
w.logger.Printf("[ERROR] [STORAGE] Failed to clear stopped stacks: %v", err)
}
return started, failed
}
// ── Debug simulation methods ─────────────────────────────────────────
// isSimulated returns true if the path is in simulated-disconnect state.
func (w *StorageWatchdog) isSimulated(path string) bool {
w.simulatedMu.RLock()
defer w.simulatedMu.RUnlock()
return w.simulatedPaths[path]
}
// SimulateDisconnect simulates a drive disconnection without actually unmounting.
// Runs disconnect steps 1,2,4,5,6,7 (skips step 3: lazyUnmount).
// Returns the list of stopped stacks.
func (w *StorageWatchdog) SimulateDisconnect(ctx context.Context, path string) ([]string, error) {
sp := w.findStoragePath(path)
if sp == nil {
return nil, fmt.Errorf("storage path %q not found", path)
}
if sp.Disconnected {
return nil, fmt.Errorf("drive already disconnected")
}
if sp.Decommissioned {
return nil, fmt.Errorf("drive is decommissioned")
}
label := sp.Label
if label == "" {
label = sp.Path
}
w.logger.Printf("[INFO] [STORAGE] [DEBUG-SIM] Simulating disconnect: %s (%s)", path, label)
// Mark as simulated so the watchdog skips probing this path
w.simulatedMu.Lock()
w.simulatedPaths[path] = true
w.simulatedMu.Unlock()
// Step 1: Stop affected stacks
stoppedStacks := w.stopAffectedStacks(path)
// Step 2: Mark disconnected in settings
if err := w.settings.SetDisconnected(path, true, stoppedStacks); err != nil {
w.logger.Printf("[ERROR] [STORAGE] [DEBUG-SIM] Failed to mark disconnected: %v", err)
}
// Step 3: SKIPPED (no lazyUnmount — drive stays physically mounted)
// Step 4: Update in-memory state
state := w.getOrCreateState(path)
state.mu.Lock()
state.lastStatus = "disconnected"
state.probeInterval = disconnectedProbeInterval
state.consecutiveFailures = 0
state.mu.Unlock()
// Step 5: Trigger alert refresh
if w.alertRefresh != nil {
w.alertRefresh()
}
// Step 6: Send notification
w.notifier.NotifyStorageDisconnected(label, stoppedStacks)
// Step 7: Push hub report
if w.pushHubReport != nil {
go w.pushHubReport()
}
w.logger.Printf("[INFO] [STORAGE] [DEBUG-SIM] Disconnect simulated: %s — %d stack(s) stopped", path, len(stoppedStacks))
return stoppedStacks, nil
}
// SimulateReconnect undoes a simulated disconnection.
func (w *StorageWatchdog) SimulateReconnect(ctx context.Context, path string) error {
if !w.isSimulated(path) {
return fmt.Errorf("path %q is not in simulated-disconnect state", path)
}
sp := w.findStoragePath(path)
if sp == nil {
return fmt.Errorf("storage path %q not found", path)
}
label := sp.Label
if label == "" {
label = sp.Path
}
w.logger.Printf("[INFO] [STORAGE] [DEBUG-SIM] Simulating reconnect: %s (%s)", path, label)
// Remove from simulated set
w.simulatedMu.Lock()
delete(w.simulatedPaths, path)
w.simulatedMu.Unlock()
// Verify drive is actually still mounted (it should be since we never unmounted)
verifyResult := system.ProbeStoragePath(path)
if verifyResult.Status != system.ProbeConnected {
return fmt.Errorf("drive probe failed after simulation clear: %v", verifyResult.Err)
}
// Clean restic locks
w.cleanResticLocks(ctx, path)
// Validate stopped stacks
filteredStacks := w.filterStoppedStacks(sp.StoppedStacks)
// Clear disconnected, preserve stopped stacks for restart UI
if err := w.settings.SetDisconnected(path, false, filteredStacks); err != nil {
w.logger.Printf("[ERROR] [STORAGE] [DEBUG-SIM] Failed to clear disconnected: %v", err)
}
// Update in-memory state
state := w.getOrCreateState(path)
state.mu.Lock()
state.lastStatus = "connected"
state.probeInterval = defaultProbeInterval
state.consecutiveFailures = 0
state.mu.Unlock()
// Trigger alert refresh
if w.alertRefresh != nil {
w.alertRefresh()
}
// Send notification
w.notifier.NotifyStorageReconnected(label)
if w.pushHubReport != nil {
go w.pushHubReport()
}
w.logger.Printf("[INFO] [STORAGE] [DEBUG-SIM] Reconnect simulated: %s", path)
return nil
}
// PathDebugStatus holds per-path probe state for the debug page.
type PathDebugStatus struct {
Path string `json:"path"`
Label string `json:"label"`
Status string `json:"status"`
Simulated bool `json:"simulated"`
ProbeOK bool `json:"probe_ok"`
DebounceCount int `json:"debounce_count"`
DebounceMax int `json:"debounce_max"`
LastProbe time.Time `json:"last_probe"`
AvgLatencyMs float64 `json:"avg_latency_ms"`
ProbeCount int `json:"probe_count"`
ProbeOKCount int `json:"probe_ok_count"`
}
// GetDebugStatus returns per-path probe state for the debug page.
func (w *StorageWatchdog) GetDebugStatus() []PathDebugStatus {
paths := w.settings.GetStoragePaths()
result := make([]PathDebugStatus, 0, len(paths))
w.mu.Lock()
defer w.mu.Unlock()
for _, sp := range paths {
if sp.Decommissioned {
continue
}
ds := PathDebugStatus{
Path: sp.Path,
Label: sp.Label,
DebounceMax: probeThreshold,
}
if sp.Disconnected {
ds.Status = "disconnected"
} else {
ds.Status = "connected"
}
ds.Simulated = w.isSimulatedLocked(sp.Path)
if state, ok := w.pathState[sp.Path]; ok {
state.mu.Lock()
ds.DebounceCount = state.consecutiveFailures
ds.LastProbe = state.lastProbeTime
ds.ProbeOK = state.lastStatus == "connected"
ds.ProbeCount = state.probeCount
ds.ProbeOKCount = state.probeOKCount
if state.probeCount > 0 {
ds.AvgLatencyMs = float64(state.totalLatency.Milliseconds()) / float64(state.probeCount)
}
state.mu.Unlock()
}
result = append(result, ds)
}
return result
}
// isSimulatedLocked checks simulation state without acquiring simulatedMu
// (caller must hold w.mu or be ok with a racy read for debug display).
func (w *StorageWatchdog) isSimulatedLocked(path string) bool {
w.simulatedMu.RLock()
defer w.simulatedMu.RUnlock()
return w.simulatedPaths[path]
}
// findStoragePath returns the storage path entry for a given path, or nil.
func (w *StorageWatchdog) findStoragePath(path string) *settings.StoragePath {
for _, sp := range w.settings.GetStoragePaths() {
if sp.Path == path {
return &sp
}
}
return nil
}