One Caddy serves the whole container and POST /load replaces the entire
config, so two instances pushing would clobber each other. Now only the
instance with CADDY_MANAGER=true (production) pushes, seeds routes from the
Caddyfile, and accepts route mutations (others get 403). /api/auth/config
exposes caddyManaged so the non-owner Settings UI shows the Caddy section
read-only. The installer sets the flag on the production .env only.
When a route's upstream starts with https://, buildCaddyfile emits a
transport http { tls_insecure_skip_verify } block so Caddy connects over TLS
and accepts the self-signed certificate typical of backends like Semaphore.
Added a UI hint explaining the https:// prefix.
GET /api/caddy/routes now returns the route array directly instead of
{ system, custom }. Frontend state is CaddyRoute[] initialised to [],
rendered with a simple .map() and an empty-state message — mirroring how
bookings are loaded and displayed.
Routes now load immediately from DB without waiting for the Caddy Admin API
status check (which can take up to 2s timeout). A dedicated useEffect on
caddyEnabled replaces the unreliable fire-and-forget call inside loadSettings.
POST /api/caddy/routes now returns 409 if the hostname already exists,
preventing duplicate DB entries that cause Caddy's "ambiguous site definition" error.
loadCaddyRoutes uses Promise.allSettled so a failure in the status check
can no longer silently prevent the routes list from loading.
Add inline edit for custom routes (Pencil icon → inline form with all fields).
Log route add/update/delete/import to the logs table (type: system) so
operations appear in the Logbook. Fix loadCaddyRoutes() called without await
after settings save, causing a race between the success message and route list.
Add a Database section under Settings (split into Integrations/System
tabs) showing SQLite file size, last-modified date, a proportional
table-usage bar and per-table row counts. Supports downloading a
consistent backup and importing a .db file that overwrites the entire
database, with an explicit overwrite warning and confirmation.
Backend adds GET /api/database/info, GET /api/database/backup and
POST /api/database/import; DB_FILE is now exported from server-db.
caddy_prod_domain and caddy_dev_domain are already handled by the Proxmox deploy
process. The Caddy integration is a generic TLS proxy for additional services
(Semaphore, Netbox, etc.) — the custom routes list is the sole mechanism.
- JSON panel header (bg-slate-900) no longer flips to light gray inside
the dark terminal block — scoped override keeps it #161b22
- Title text and copy button styled consistently for dark context
- Orange Ansible status card gets proper light-mode colors (orange-50 bg,
orange-200 border, orange-600 text)
- application/json badge tweaked to text-indigo-400 for consistency
Ansible trigger successes now logged as type 'booking' so they appear
in the default filter view. Removed the redundant 'All incl. System'
filter button.
BookingDetailsModal: remove static playbook template and fake simulator,
keep only the JSON REST response panel. Settings: drop max-w-2xl,
wrap integration cards in lg:grid-cols-3 so Azure, CheckMK and
Semaphore sit side by side on wide screens.
- Background scheduler checks every 30s for bookings that need setup or teardown
- Per-lab Semaphore template IDs stored on the labs table
- Booking flags track which jobs have been triggered and their Semaphore job IDs
- Immediate teardown triggered when an active booking is cancelled
- Settings UI section for Semaphore API URL, token, and project ID
- Lab template form fields for setup/teardown template IDs
- BookingDetailsModal shows live Ansible job status with manual trigger buttons
Replace per-device CheckMK URL field with a global, IP-based lookup.
The sync job fetches all host configs from CheckMK once per cycle,
matches each device by IP address, and updates its status accordingly.
Devices not found in CheckMK are reset to 'unknown'.
- Add checkmk_enabled / checkmk_api_user settings; toggle in Settings
mirrors the Entra ID pattern (fields dim when disabled)
- Sync job uses self-scheduling setTimeout so interval changes apply
without a server restart; POST /api/checkmk/sync for manual triggers
- Status changes and a per-cycle summary are written to the Logbook
- Remove checkMkUrl from Device type, form, list view, and detail panel;
status badge and CheckMK panel only render when CheckMK is enabled
- Booking offline warning suppressed when CheckMK is disabled
- Topology status dot color driven purely by device.status
- Header now shows Production/Development with color-coded dot via import.meta.env.PROD
- LinkDashboard: click-to-edit description inline (blur/Enter to save, Escape to cancel)
- LoginPage: fix Azure button label to English
- Remove unused motion and autoprefixer dependencies
- Add SQLite settings table with getSetting/setSetting/getAllSettings helpers
- Implement Azure OAuth2 authorization code flow via @azure/msal-node
- Add public GET /api/auth/config endpoint for frontend activation check
- Add admin-only GET/PUT /api/settings API with masked secret fields
- CheckMK sync reads credentials from DB settings (env vars as fallback)
- New Settings.tsx: Entra ID and CheckMK configuration cards
- LoginPage: "Sign in with Microsoft" button, shown only when Azure is active
- App.tsx: OAuth callback handling (?token=/?auth_error=), Settings tab for admins
