av1155/houndarr v1.11.0

on GitHub

Added

• HOUNDARR_LOG_RETENTION_DAYS env var configures the search log retention threshold (default 30; 0 disables automatic purges, 7 to 365 overrides the default). (#586)
• Helm chart ships a startupProbe with a 10-minute default budget so first-boot migrations on Kubernetes finish before the liveness probe kicks in; failureThreshold and periodSeconds are exposed in values.yaml. (#586)

Changed

• Dashboard aggregations cache for 20 seconds in-process so polling tabs share one DB scan; mutation routes invalidate immediately, and the HTMX poll drops new triggers while a previous response is still in flight. (#586)
• SQLite connections are reused through a connection pool with the modern operational PRAGMA stack (synchronous=NORMAL, temp_store=MEMORY, larger per-connection cache, memory-mapped reads, capped WAL growth); PRAGMA optimize runs after migrations and after each daily retention pass. (#586)

Fixed

• From-source installs now refuse to start with a clear error message when the compiled app.built.css is missing instead of silently 404-ing every stylesheet request, and the from-source install guide documents the pnpm install and pnpm run build-css step required since 1.10.0. (#582)
• Dashboard no longer hangs on databases with large search_log tables: two new composite indexes serve the v14 cooldown back-fill and the dashboard aggregation queries instead of full-table scans, and the back-fill itself only touches cooldown rows still at the default 'missing' stamp. (#586)