Follow-up polish to #5478. The new launcher warned at Node 25+ and told
users to install Node 22 LTS on a native-module failure. Start Haven.bat
warns at 27+, says Haven is tested on 18-26, and points at Node 26 LTS.
Two launchers in the same repo giving opposite advice is worse than
either one alone, and the fix for a native module that won't load is a
newer Node, not an older one.
The 15-second readiness loop probed /api/health with curl only. On
minimal server images without curl (and on hosts that lost it after a
reflush), every probe failed with exit 127 while the server was already
up and serving the frontend — and at the 15s mark the launcher killed
the healthy process and printed 'Server failed to start', leaving only
a 'terminated gracefully' trail in the logs. Reported on the community
server (worked in a desktop VM, failed on the freshly set up host —
the VM had curl, the server image did not).
- Probe with node itself: it is guaranteed present because it runs the
server. No curl/wget dependency; https first, http fallback, self-
signed certs accepted.
- A probe failure alone never kills a live server anymore. If readiness
is unconfirmed after the window but the process is up, say so and
leave it running.
- A dead server process is detected immediately (not after 15 silent
seconds) and the message points at the real log instead of guessing.
- Window widened to 30s for slow first migrations.
- Node version gate now matches the Windows launcher: functional
better-sqlite3 load check (with one rebuild attempt) instead of a
hard version refusal, so Node 24 works where the native module loads.
Tested: healthy start with curl+wget absent on Node 24 (probe confirms,
LIVE banner); the old script reproduced the reported kill-at-15s against
a serving instance; a crashing server now reports the real failure in
~1 second with exit 1.
The lsof kill block was intended to clean up stale Haven processes on restart, but it kills anything bound to the configured port, including unrelated processes. Per community feedback, if the port is in use the server should simply fail to start with a clear EADDRINUSE error rather than silently killing whatever is there.
- Fix noise gate/AI suppression lost after mic device switch (#212)
- Fix screen share + webcam not renegotiated on voice-rejoin (#213)
- Start Haven.bat / start.sh now read PORT from .env (#214)
- SSL cert generation errors are no longer suppressed (#214)
- Desktop download links updated to v1.1.7
- Tighten engines field to <24.0.0 in package.json
- Add runtime version guard at top of server.js with clear error message
- Add version check to start.sh matching what Start Haven.bat already does
- Update Dockerfile from node:20-alpine to node:22-alpine
- Add FORCE_HTTP=true env option to skip built-in SSL (for Caddy/nginx/Traefik)
- Update startup scripts (bash, bat, Docker) to respect FORCE_HTTP
- Disable HSTS header when FORCE_HTTP is active
- Add auto_assign column to roles table with DB migration
- Default 'User' role is auto-assigned on registration
- Admin UI: toggle auto-assign per role, visual indicator
- join-channel auto-assigns flagged roles to new members
- Fix window.app reference for desktop voice hook
- Database, .env, certs, and uploads now stored outside the code directory
- Auto-migration moves existing data from old locations on first launch
- SSL certs auto-detected in data dir (no .env config needed)
- New src/paths.js module as single source of truth for all data paths
- HAVEN_DATA_DIR env var for custom data location override
- Updated Start Haven.bat, start.sh, README, GUIDE, .env.example
- Bumped version to 1.1.0
ancsemi