Extract jail, filter, and action configuration management into separate
domain-focused service modules:
- jail_config_service.py: Jail activation, deactivation, validation, rollback
- filter_config_service.py: Filter discovery, CRUD, assignment to jails
- action_config_service.py: Action discovery, CRUD, assignment to jails
Benefits:
- Reduces monolithic 3100-line module into three focused modules
- Improves readability and maintainability per domain
- Clearer separation of concerns following single responsibility principle
- Easier to test domain-specific functionality in isolation
- Reduces coupling - each service only depends on its needed utilities
Changes:
- Create three new service modules under backend/app/services/
- Update backend/app/routers/config.py to import from new modules
- Update exception and function imports to source from appropriate service
- Update Architecture.md to reflect new service organization
- All existing tests continue to pass with new module structure
Relates to Task 4 of refactoring backlog in Docs/Tasks.md
- Create ErrorBoundary component to handle React render errors
- Wrap App component with ErrorBoundary for global error handling
- Add comprehensive tests for ErrorBoundary functionality
- Show fallback UI with error message when errors occur
Implement TASK F-2: Wrap JailDetailPage jail-control API calls in a hook.
Changes:
- Add start(), stop(), reload(), and setIdle() methods to useJailDetail hook
- Update JailDetailPage to use hook control methods instead of direct API imports
- Update error handling to remove dependency on ApiError type
- Add comprehensive tests for new control methods (8 tests)
- Update existing test to include new hook methods in mock
The control methods handle refetching jail data after each operation,
consistent with the pattern used in useJails hook.
Centralizes ruff linter configuration at project root with consistent
line length (120 chars), Python 3.12 target, and exclusions for
external dependencies and build artifacts.
Logs a warning when the initial setup status request fails, allowing
operators to diagnose issues during the setup phase. The form remains
visible while the error is logged for debugging purposes.
- Add TYPE_CHECKING guards for runtime-expensive imports (aiohttp, aiosqlite)
- Reorganize imports to follow PEP 8 conventions
- Convert TypeAlias to modern PEP 695 type syntax (where appropriate)
- Use Sequence/Mapping from collections.abc for type hints (covariant)
- Replace string literals with cast() for improved type inference
- Fix casting of Fail2BanResponse and TypedDict patterns
- Add IpLookupResult TypedDict for precise return type annotation
- Reformat overlong lines for readability (120 char limit)
- Add asyncio_mode and filterwarnings to pytest config
- Update test fixtures with improved type hints
This improves mypy type checking and makes type relationships explicit.
Task 2: rename 'Failures:' label to 'Failed Attempts:' and update tooltip
to 'Total failed authentication attempts currently tracked by fail2ban
across all active jails' — more accurate than 'Currently failing IPs'.
Task 3: move <ServerHealthSection /> to the top of ServerTab so users
see connectivity status before scrolling through all settings fields.
- Inject __APP_VERSION__ at build time via vite.config.ts define (reads
frontend/package.json#version); declare the global in vite-env.d.ts.
- Render 'BanGUI v{__APP_VERSION__}' in the sidebar footer (MainLayout)
when expanded; hidden when collapsed.
- Rename fail2ban version tooltip to 'fail2ban daemon version' in
ServerStatusBar so it is visually distinct from the app version.
- Sync frontend/package.json version (0.9.0 → 0.9.3) to match
Docker/VERSION; update release.sh to keep them in sync on every bump.
- Add vitest define stub for __APP_VERSION__ so tests compile cleanly.
- Add ServerStatusBar and MainLayout test suites (10 new test cases).
On startup BanGUI now verifies that the four fail2ban jail config files
required by its two custom jails (manual-Jail and blocklist-import) are
present in `$fail2ban_config_dir/jail.d`. Any missing file is created
with the correct default content; existing files are never overwritten.
Files managed:
- manual-Jail.conf (enabled=false template)
- manual-Jail.local (enabled=true override)
- blocklist-import.conf (enabled=false template)
- blocklist-import.local (enabled=true override)
The check runs in the lifespan hook immediately after logging is
configured, before the database is opened.
The sha256Hex helper used window.crypto.subtle.digest(), which is only
available in a secure context (HTTPS / localhost). In the HTTP Docker
environment crypto.subtle is undefined, causing a TypeError before any
request is sent — the setup and login forms both silently failed with
'An unexpected error occurred'.
Fix: pass raw passwords directly to the API. The backend already applies
bcrypt, which is sufficient. No stored hashes need migration because
setup never completed successfully in the HTTP environment.
* frontend/src/pages/SetupPage.tsx — remove sha256Hex call
* frontend/src/api/auth.ts — remove sha256Hex call
* frontend/src/pages/__tests__/SetupPage.test.tsx — drop crypto mock
* frontend/src/utils/crypto.ts — deleted (no remaining callers)
Task 0.1: Create database parent directory before connecting
- main.py _lifespan now calls Path(database_path).parent.mkdir(parents=True,
exist_ok=True) before aiosqlite.connect() so the app starts cleanly on
a fresh Docker volume with a nested database path.
Task 0.2: SetupRedirectMiddleware redirects when db is None
- Guard now reads: if db is None or not is_setup_complete(db)
A missing database (startup still in progress) is treated as setup not
complete instead of silently allowing all API routes through.
Task 0.3: SetupGuard redirects to /setup on API failure
- .catch() handler now sets status to 'pending' instead of 'done'.
A crashed backend cannot serve protected routes; conservative fallback
is to redirect to /setup.
Task 0.4: SetupPage shows spinner while checking setup status
- Added 'checking' boolean state; full-screen Spinner is rendered until
getSetupStatus() resolves, preventing form flash before redirect.
- Added console.warn in catch block; cleanup return added to useEffect.
Also: remove unused type: ignore[call-arg] from config.py.
Tests: 18 backend tests pass; 117 frontend tests pass.
Add key={selectedActiveJail.name} and key={selectedInactiveJail.name} to
JailConfigDetail and InactiveJailDetail in JailsTab.tsx so React unmounts
and remounts the detail component whenever the selected jail changes,
resetting all internal state including the loadedRef guard.
Rename fail2ban-dev-config jail.d/bangui-sim.conf and filter.d/bangui-sim.conf
to manual-Jail.conf. Update section header, filter reference, and comments in
both files. Update JAIL constant and header comment in check_ban_status.sh.
Update comments in simulate_failed_logins.sh. Replace all bangui-sim
occurrences in fail2ban-dev-config/README.md.
Rename GET/PUT /api/config/actions/{name} to /actions/{name}/raw in
file_config.py to eliminate the route-shadowing conflict with config.py,
which registers its own GET /actions/{name} returning ActionConfig.
Add configActionRaw endpoint helper in endpoints.ts and update
fetchActionFile/updateActionFile in config.ts to use it. Add
TestGetActionFileRaw and TestUpdateActionFileRaw test classes.
- Replace vague 'System Recovered' message with 'Configuration Rolled Back'
and actionable text describing the rollback outcome
- Replace 'Manual Intervention Required' with 'Rollback Unsuccessful' and
specific instructions: check jail.d/{name}.local, fix manually, restart
- Add test_activate_jail_rollback_deletes_file_when_no_prior_local to cover
rollback path when no .local file existed before activation
- Mark all three tasks complete in Tasks.md
- Delete RecoveryBanner.tsx component and its test
- Remove RecoveryBanner from MainLayout
- Remove onCrashDetected prop from ActivateJailDialog, JailsTab
- Remove fetchPendingRecovery, rollbackJail API functions
- Remove configJailRollback, configPendingRecovery endpoints
- Remove PendingRecovery type
- Add has_local_override field to InactiveJail model
- Update _build_inactive_jail and list_inactive_jails to compute the field
- Add delete_jail_local_override() service function
- Add DELETE /api/config/jails/{name}/local router endpoint
- Surface has_local_override in frontend InactiveJail type
- Show Deactivate Jail button in JailsTab when has_local_override is true
- Add tests: TestBuildInactiveJail, TestListInactiveJails, TestDeleteJailLocalOverride
- jail_service.restart(): replace invalid ["restart"] socket command with
["stop"], matching fail2ban transmitter protocol. The daemon is now
stopped via socket; the caller starts it via subprocess.
- config_file_service: expose _start_daemon and _wait_for_fail2ban as
public start_daemon / wait_for_fail2ban functions.
- restart_fail2ban router: orchestrate stop (socket) → start (subprocess)
→ probe (socket). Returns 204 on success, 503 when fail2ban does not
come back within 10 s. Catches JailOperationError → 409.
- reload_fail2ban router: add JailOperationError catch → 409 Conflict,
consistent with other jail control endpoints.
- Tests: add TestJailControls.test_restart_* (3 cases), TestReloadFail2ban
502/409 cases, TestRestartFail2ban (5 cases), TestRollbackJail (6
integration tests verifying file-write, subprocess invocation, socket-
probe truthiness, active_jails count, and offline-at-call-time).
The container init script (init-fail2ban-config) copies jail.conf from the
image's /defaults/ on every start, overwriting any direct edits. The correct
fix is jail.local, which is not present in the image defaults and therefore
persists across restarts.
Changes:
- Add Docker/fail2ban-dev-config/fail2ban/jail.local with [DEFAULT] overrides
for banaction = iptables-multiport and banaction_allports = iptables-allports.
fail2ban loads jail.local after jail.conf so these values are available to
all jails during %(action_)s interpolation.
- Untrack jail.local from .gitignore so it is committed to the repo.
- Fix TypeError in config_file_service: except jail_service.JailNotFoundError
failed when jail_service was mocked in tests because MagicMock attributes are
not BaseException subclasses. Import JailNotFoundError directly instead.
- Mark BUG-001 as Done in Tasks.md.
This commit implements fixes for three independent bugs in the fail2ban configuration and integration layer:
1. Task 1: Detect UnknownJailException and prevent silent failures
- Added JailNotFoundError detection in jail_service.reload_all()
- Enhanced error handling in config_file_service to catch JailNotFoundError
- Added specific error message with logpath validation hints
- Added rollback test for this scenario
2. Task 2: Fix iptables-allports exit code 4 (xtables lock contention)
- Added global banaction setting in jail.conf with -w 5 lockingopt
- Removed redundant per-jail banaction overrides from bangui-sim and blocklist-import
- Added production compose documentation note
3. Task 3: Suppress log noise from unsupported backend/idle commands
- Implemented capability detection to cache command support status
- Double-check locking to minimize lock contention
- Avoids sending unsupported get <jail> backend/idle commands
- Returns default values without socket calls when unsupported
All changes include comprehensive tests and maintain backward compatibility.
Adds ability to reload or restart fail2ban service from the Server tab UI.
Backend changes:
- Add new restart() method to jail_service.py that sends 'restart' command
- Add new POST /api/config/restart endpoint in config router
- Endpoint returns 204 on success, 502 if fail2ban unreachable
- Includes structured logging via 'fail2ban_restarted' log entry
Frontend changes:
- Add configRestart endpoint to endpoints.ts
- Add restartFail2Ban() API function in config.ts API module
- Import ArrowSync24Regular icon from Fluent UI
- Add reload and restart button handlers to ServerTab
- Display 'Reload fail2ban' and 'Restart fail2ban' buttons in action row
- Show loading spinner during operation
- Display success/error MessageBar with appropriate feedback
- Update ServerTab docstring to document new buttons
All 115 frontend tests pass.
The Log tab provided a service health panel and log viewer. These are
consolidated into the Server tab with a new ServerHealthSection component
that encapsulates all log-related functionality.
- Extract service health panel and log viewer into ServerHealthSection component
- Add severity-based log line color coding (ERROR=red, WARNING=yellow, DEBUG=gray)
- Implement log filtering, line count selection, and auto-refresh controls
- Scroll to bottom when new log data arrives
- Render health metrics grid with version, jail count, bans, failures
- Show read-only log level and log target in health section
- Handle non-file targets with informational banner
- Import ServerHealthSection in ServerTab and render after map thresholds
- Remove LogTab component import from ConfigPage
- Remove 'log' from TabValue type
- Remove Log tab element from TabList
- Remove conditional render for LogTab
- Remove LogTab from barrel export (index.ts)
- Delete LogTab.tsx and LogTab.test.tsx files
- Update ConfigPage docstring
All 115 frontend tests pass (8 fewer due to deleted LogTab tests).
The Map tab provided a form for editing world-map color thresholds
(low, medium, high). Moving this into the Server tab consolidates all
server-side configuration in one place.
- Add map color thresholds section to ServerTab with full validation
- Load map thresholds on component mount with useEffect
- Implement auto-save for threshold changes via useAutoSave hook
- Display threshold color interpolation guide
- Remove MapTab component import from ConfigPage
- Remove 'map' from TabValue type
- Remove Map tab element from TabList
- Remove conditional render for MapTab
- Remove MapTab from barrel export (index.ts)
- Delete MapTab.tsx file
- Update ConfigPage test to remove MapTab mock
All 123 frontend tests pass.
