This commit completely resolves the autostart tunnel functionality issues by:
**Core Autostart System**:
- Fixed internal API to return explicit autostart fields to tunnel service
- Implemented automatic endpoint credential resolution during autostart enable
- Enhanced database synchronization with force save and verification
- Added comprehensive debugging and logging throughout the process
**Tunnel Connection Improvements**:
- Enhanced credential resolution with priority: TunnelConnection → autostart → encrypted
- Fixed SSH command format with proper tunnel markers and exec process naming
- Added connection state protection to prevent premature cleanup during establishment
- Implemented sequential kill strategies for reliable remote process cleanup
**Type System Extensions**:
- Extended TunnelConnection interface with endpoint credential fields
- Added autostart credential fields to SSHHost interface for plaintext storage
- Maintained backward compatibility with existing encrypted credential system
**Key Technical Fixes**:
- Database API now includes /db/host/internal/all endpoint with SystemCrypto auth
- Autostart enable automatically populates endpoint credentials from target hosts
- Tunnel cleanup uses multiple kill strategies with verification and delay timing
- Connection protection prevents cleanup interference during tunnel establishment
Users can now enable fully automated tunneling by simply checking the autostart
checkbox - no manual credential configuration required. The system automatically
resolves and stores plaintext credentials for unattended tunnel operation.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
- Fix missing i18n for terminal.terminalWithPath translation key
- Update keyboard shortcuts: remove Ctrl+T conflicts, change refresh to Ctrl+Y, rename shortcut to F6
- Remove click-to-rename functionality to prevent accidental renaming
- Fix drag preview z-index and positioning issues during file operations
- Remove false download trigger when dragging files to original position
- Fix 'Must be handling a user gesture' error in drag-to-desktop functionality
- Remove useless minimize button from file editor and diff viewer windows
- Improve context menu z-index hierarchy for better layering
- Add comprehensive drag state management and visual feedback
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Fixed SSH sessions being automatically removed after a few minutes of
inactivity, causing connection errors when users return to the interface.
## Problems Identified:
### 1. Aggressive Session Timeout
**Issue**: Sessions were cleaned up after only 10 minutes of inactivity
- Too short for typical user workflows
- No warning or graceful handling when timeout occurs
- Users would get connection errors without explanation
### 2. No Session Keepalive Mechanism
**Issue**: No frontend keepalive to maintain active sessions
- Sessions would timeout even if user was actively viewing files
- No periodic communication to extend session lifetime
- No way to detect session expiration proactively
### 3. Server-side SSH Configuration
**Issue**: While SSH had keepalive settings, they weren't sufficient
- keepaliveInterval: 30000ms (30s)
- keepaliveCountMax: 3
- But no application-level session management
## Technical Solution:
### **Extended Session Timeout**
```typescript
// Increased from 10 minutes to 30 minutes
session.timeout = setTimeout(() => {
fileLogger.info(`Cleaning up inactive SSH session: ${sessionId}`);
cleanupSession(sessionId);
}, 30 * 60 * 1000); // 30 minutes
```
### **Backend Keepalive Endpoint**
```typescript
// New endpoint: POST /ssh/file_manager/ssh/keepalive
app.post("/ssh/file_manager/ssh/keepalive", (req, res) => {
const session = sshSessions[sessionId];
session.lastActive = Date.now();
scheduleSessionCleanup(sessionId); // Reset timeout
res.json({ status: "success", connected: true });
});
```
### **Frontend Automatic Keepalive**
```typescript
// Send keepalive every 5 minutes
keepaliveTimerRef.current = setInterval(async () => {
if (sshSessionId) {
await keepSSHAlive(sshSessionId);
}
}, 5 * 60 * 1000);
```
## Session Management Flow:
**Before (Problematic):**
1. User connects → 10-minute countdown starts
2. User leaves browser open but inactive
3. Session times out after 10 minutes
4. User returns → "SSH session not found" error
5. User forced to reconnect manually
**After (Fixed):**
1. User connects → 30-minute countdown starts
2. Frontend sends keepalive every 5 minutes automatically
3. Each keepalive resets the 30-minute timeout
4. Session stays alive as long as browser tab is open
5. Graceful handling if keepalive fails
## Benefits:
✅ **Extended Session Lifetime**: 30 minutes vs 10 minutes base timeout
✅ **Automatic Session Maintenance**: Keepalive every 5 minutes
✅ **Transparent to User**: No manual intervention required
✅ **Robust Error Handling**: Graceful degradation if keepalive fails
✅ **Resource Efficient**: Only active sessions consume resources
✅ **Better User Experience**: No unexpected disconnections
Sessions now persist for the entire duration users have the file
manager open, eliminating frustrating timeout errors.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Fixed race conditions and loading problems when users click folders
or navigation buttons too quickly.
## Problems Identified:
### 1. Race Conditions in Path Changes
**Issue**: Fast clicking folders/back button caused multiple simultaneous requests
- useEffect triggered on every currentPath change
- No debouncing for path changes (only for manual refresh)
- Multiple loadDirectory() calls executed concurrently
- Later responses could overwrite earlier ones
### 2. Concurrent Request Conflicts
**Issue**: loadDirectory() had basic isLoading check but insufficient protection
- Multiple requests could run if timing was right
- No tracking of which request was current
- Stale responses could update UI incorrectly
### 3. Missing Request Cancellation
**Issue**: No way to cancel outdated requests when user navigates rapidly
- Old requests would complete and show wrong directory
- Confusing UI state when mixed responses arrived
## Technical Solution:
### **Path Change Debouncing**
```typescript
// Added 150ms debounce specifically for path changes
const debouncedLoadDirectory = useCallback((path: string) => {
if (pathChangeTimerRef.current) {
clearTimeout(pathChangeTimerRef.current);
}
pathChangeTimerRef.current = setTimeout(() => {
if (path !== lastPathChangeRef.current && sshSessionId) {
loadDirectory(path);
}
}, 150);
}, [sshSessionId, loadDirectory]);
```
### **Request Race Condition Protection**
```typescript
// Track current loading path for proper cancellation
const currentLoadingPathRef = useRef<string>("");
// Enhanced concurrent request prevention
if (isLoading && currentLoadingPathRef.current !== path) {
console.log("Directory loading already in progress, skipping:", path);
return;
}
```
### **Stale Response Handling**
```typescript
// Check if response is still relevant before updating UI
if (currentLoadingPathRef.current !== path) {
console.log("Directory load canceled, newer request in progress:", path);
return; // Discard stale response
}
```
## Flow Improvements:
**Before (Problematic):**
1. User clicks folder A → currentPath changes → useEffect → loadDirectory(A)
2. User quickly clicks folder B → currentPath changes → useEffect → loadDirectory(B)
3. Both requests run concurrently
4. Response A or B arrives randomly, wrong folder might show
**After (Fixed):**
1. User clicks folder A → currentPath changes → debouncedLoadDirectory(A)
2. User quickly clicks folder B → currentPath changes → cancels A timer → debouncedLoadDirectory(B)
3. Only request B executes after 150ms
4. If A somehow runs, its response is discarded as stale
## User Experience:
✅ Rapid folder navigation works smoothly
✅ Back button rapid clicking handled properly
✅ No more loading wrong directories
✅ Proper loading states maintained
✅ No duplicate API requests
✅ Responsive feel with 150ms debounce (fast enough to feel instant)
The file manager now handles rapid user interactions gracefully without
race conditions or loading the wrong directory content.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Fixed the brief jarring flash between SSH connection and file list display:
## Problem
During file manager connection process:
1. SSH connection completes → setIsLoading(false)
2. Brief empty/intermediate state displayed (jarring flash)
3. useEffect triggers → setIsLoading(true) again
4. Directory loads → setIsLoading(false)
5. Files finally displayed
This created a jarring user experience with double loading states.
## Root Cause
- initializeSSHConnection() only handled SSH connection
- File directory loading was handled separately in useEffect
- Gap between connection completion and directory loading caused UI flash
## Solution
**Unified Connection + Directory Loading:**
- Modified initializeSSHConnection() to load initial directory immediately after SSH connection
- Added initialLoadDoneRef to prevent duplicate loading in useEffect
- Loading state now remains true until both connection AND directory are ready
**Technical Changes:**
- SSH connection + initial directory load happen atomically
- useEffect skips initial load, only handles path changes
- No more intermediate states or double loading indicators
## Flow Now:
1. setIsLoading(true) → "Connecting..."
2. SSH connection establishes
3. Initial directory loads immediately
4. setIsLoading(false) → Files displayed seamlessly
**User Experience:**
✅ Smooth single loading state until everything is ready
✅ No jarring flashes or intermediate states
✅ Immediate file display after connection
✅ Maintains proper loading states for path changes
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Added missing translation key for database security settings:
Problem:
- AdminSettings.tsx used t("admin.encryptionEnabled")
- Translation key was missing from both English and Chinese files
- Caused missing text in database security encryption status display
Solution:
- Added "encryptionEnabled": "Encryption Enabled" to English translations
- Added "encryptionEnabled": "加密已启用" to Chinese translations
- Maintains consistency with existing encryption-related translations
Files updated:
- src/locales/en/translation.json
- src/locales/zh/translation.json
Now the database security section properly displays encryption status
with correct i18n support in both languages.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Fixed "ReferenceError: t is not defined" when creating new files/folders:
Problem:
- CreateIntentGridItem and CreateIntentListItem components used t() function
- But neither component had useTranslation hook imported
- Caused runtime error when trying to create new files or folders
Solution:
- Added const { t } = useTranslation(); to both components
- Fixed hardcoded English text in CreateIntentListItem placeholder
- Now uses proper i18n translation keys for all UI text
Changes:
- CreateIntentGridItem: Added useTranslation hook
- CreateIntentListItem: Added useTranslation hook + fixed placeholder text
- Both components now properly use t('fileManager.folderName') and t('fileManager.fileName')
Now file/folder creation works without console errors and supports i18n.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Fixed two critical database issues causing user creation errors and data loss:
## Issue 1: Lazy Encryption Migration Error
**Problem**: TypeError: Cannot read properties of undefined (reading 'db')
**Root Cause**: AuthManager called getSqlite() before database initialization
**Solution**: Added databaseReady promise await before accessing SQLite instance
Changes in auth-manager.ts:
- Import and await databaseReady promise before getSqlite() call
- Ensures database is fully initialized before migration attempts
- Prevents "SQLite not initialized" errors during user login
## Issue 2: Data Loss After Backend Restart
**Problem**: All user data wiped after backend restart
**Root Cause**: Database saves were skipped when file encryption disabled
**Solution**: Added fallback to unencrypted SQLite file persistence
Changes in database/db/index.ts:
- Modified saveMemoryDatabaseToFile() to handle encryption disabled scenario
- Added unencrypted SQLite file fallback to prevent data loss
- Added data directory creation to ensure save path exists
- Enhanced logging to track save operations and warnings
## Technical Details:
- saveMemoryDatabaseToFile() now saves data regardless of encryption setting
- Encrypted: saves to .encrypted file (existing behavior)
- Unencrypted: saves to .sqlite file (new fallback)
- Ensures data persistence in all configurations
- Maintains 15-second auto-save and real-time trigger functionality
These fixes ensure:
✅ User creation works without backend errors
✅ Data persists across backend restarts
✅ Lazy encryption migration completes successfully
✅ Graceful handling of encryption disabled scenarios
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Root cause: Duplicate handleFileOpen function definitions caused symlinks
to be treated as regular files instead of navigating to their targets.
Problem:
- Line 575: Correct implementation with symlink handling
- Line 1401: Incorrect duplicate that overrode the correct function
- Double-clicking symlinks opened them as files instead of following links
Solution:
- Removed duplicate handleFileOpen function (lines 1401-1436)
- Preserved correct implementation with symlink navigation logic
- Added recordRecentFile call for consistency
Now symlinks properly:
- Navigate to target directories when they point to folders
- Open target files when they point to files
- Use identifySSHSymlink backend API for resolution
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Fixed WebSocket connection issue where SSH terminals couldn't connect
despite correct credentials. Root cause was port mismatch - terminals
were trying to connect to port 8081 while SSH service runs on 8082.
Changes:
- Desktop Terminal: Updated WebSocket URL to use port 8082
- Mobile Terminal: Updated WebSocket URL to use port 8082
- File Manager continues using port 8081 for HTTP API (unchanged)
This ensures all SSH terminal connections route to the correct service port.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Phase 1: Database file migration (startup)
- Add DatabaseMigration class for safe unencrypted → encrypted DB migration
- Disable foreign key constraints during migration to prevent constraint failures
- Create timestamped backups and verification checks
- Rename original files instead of deletion for safety
Phase 2: Lazy field encryption (user login)
- Add LazyFieldEncryption utility for plaintext field detection
- Implement gradual migration of sensitive fields using user KEK
- Update DataCrypto to handle mixed plaintext/encrypted data
- Integrate lazy encryption into AuthManager login flow
Key improvements:
- Non-destructive migration with comprehensive backup strategy
- Automatic detection and handling of plaintext vs encrypted fields
- User-transparent migration during normal login process
- Complete migration logging and admin API endpoints
- Foreign key constraint handling during database structure migration
Resolves data decryption errors during Docker updates by providing
seamless transition from plaintext to encrypted storage.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
Replace JSON-based backup system with SQLite export/import functionality:
**Export Features:**
- Generate SQLite database files with complete user data
- Export all tables: SSH hosts, credentials, file manager data, settings, alerts
- Include OIDC configuration and system settings (admin only)
- Password authentication required for data decryption
- Direct browser download instead of file path display
**Import Features:**
- Incremental import with duplicate detection and skipping
- Smart conflict resolution by key combinations:
- SSH hosts: ip + port + username
- Credentials: name + username
- File manager: path + name
- Re-encrypt imported data to current user's keys
- Admin-only settings import (including OIDC config)
- Detailed import statistics with category breakdown
**Removed:**
- Database backup functionality (redundant with export)
- JSON export format
- File path-based workflows
**Security:**
- Password verification for all operations
- SQLite file format validation
- Proper error handling and logging
- Admin permission checks for settings
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
- Merge production-deps and native-builder stages to eliminate duplication
- Remove redundant intermediate layers that were consuming Docker space
- Add aggressive cleanup (rm -rf ~/.npm /tmp/* /var/cache/apk/*)
- Reduce overall image size and build-time space requirements
Fixes:
- ENOSPC errors during COPY operations from multiple build stages
- Excessive Docker layer accumulation from duplicate dependency installs
- Reduced disk space usage during multi-stage builds
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
- Fix chmod command syntax to properly set permissions for multiple directories
- Use && to chain chmod commands instead of space-separated arguments
- Ensure /app/config, /app/ssl, and /app/data have correct 755 permissions
Fixes syntax error that would cause Docker build failures.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
- Add comprehensive .dockerignore to exclude unnecessary files from Docker context
- Exclude .git directory to prevent large Git objects from being copied
- Exclude node_modules, logs, temp files, and other build artifacts
- Reduce Docker image size and build time significantly
Fixes:
- ENOSPC error during Docker build due to large .git directory
- Excessive Docker image size from unnecessary files
- Build context transfer time and resource usage
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
- Pre-create /app/ssl directory in Dockerfile to prevent runtime creation failures
- Set proper permissions for /app/ssl, /app/config, and /app/data directories
- Ensure all required directories exist before application startup
Fixes:
- ENOSPC error when creating SSL directory at runtime
- Permission issues with auto-generated .env file writing
- Container restart loops due to initialization failures
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <[email protected]>
ZacharyZcRandClaude