Broque Thomas 7b2324f52e Fix config DB lock spam on slow disks (#434) ... User on Docker + HDDs saw "database is locked" errors on every settings save. Two retries spaced 1 second apart isn't enough when an enrichment worker is mid-commit on spinning-rust storage, and each failed attempt logged at ERROR — multiplying the spam. Three changes in config/settings.py: - Centralized connection setup in a new _connect_db() helper that always sets PRAGMA journal_mode=WAL + busy_timeout=30000 + synchronous=NORMAL. NORMAL is the safe pairing with WAL and avoids the per-commit fsyncs that make FULL brutal on HDDs, shrinking the window the competing writer holds the lock. - _save_to_database logs lock errors at DEBUG instead of ERROR. The retry loop owns the user-visible message; otherwise every retry spammed even when the next attempt succeeded. - _save_config now retries 6 times with exponential backoff (0.2 + 0.5 + 1.0 + 2.0 + 4.0s ≈ 7.7s of sleep, on top of the 30s busy_timeout each attempt already runs internally) before logging a single error and falling back to config.json. Closes #434.		2026-04-30 13:04:09 -07:00
..
config.example.json	Add hi-res FLAC to CD quality downsampling in post-processing	2026-03-17 11:53:56 -07:00
settings.py	Fix config DB lock spam on slow disks (#434)	2026-04-30 13:04:09 -07:00