The ./data/postgres bind mount hit a WSL2/Docker Desktop bug where the
9p/virtiofs bridge reports normal postgres:postgres 0600 ownership but
the kernel still refuses the postgres user's own open() calls for
write — silent on read, fatal on any WAL write, which took the whole
database down after an unclean shutdown. Switches to a Docker-managed
named volume (bellsystems-postgres-data) that lives inside the WSL2
VM's own filesystem instead. The old bind-mounted data is left in
place at ./data/postgres/ as an untouched backup — that folder was
too wedged by the same bug to even rename, so it's abandoned rather
than deleted.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Matches postgres, which already had restart: unless-stopped — the app
containers were the only ones that didn't come back automatically
after a host reboot or Docker Desktop restart.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
ESP32 devices can't spare the 40KB+ RAM a TLS client needs, so Firebase
Storage's HTTPS-only download URLs were blocking melody downloads. Binaries
are now written to local disk (./data/melody_binaries) and served through a
new unauthenticated /api/melodies/download/{pid} route, exposed publicly on
a separate melodies.bellsystems.net vhost (plain HTTP, no TLS) so the main
console domain can stay HTTPS-only with no exceptions. Preview audio still
uses Firebase Storage since it's only ever fetched by the HTTPS admin UI.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
- Add deploy-host.sh for webhook-triggered docker redeploy
- Update docker-compose.yml and nginx.conf for auto-pull setup
- Fix vite.config.js and admin router for deployment environment
- Fix NVS CRC seed to use 0xFFFFFFFF to match esp_rom_crc32_le
- Add dual-panel flash UI: esptool log + live 115200 serial monitor
- Auto-reset device via RTS after flash (no manual power cycle needed)
- Clean up Header.jsx debug title text
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
bonaminandClaude Sonnet 5