Fixed
-
Prowlarr no longer syncs zero indexers (#763, #788, #794) — since v1.13.x the Prowlarr sync read each indexer's categories from a top-level field that Prowlarr's API does not populate (the category list lives under the indexer's
capabilities). Every indexer was rejected as having "no book/audiobook categories", and — worse — previously synced indexers were then deleted on each sync, wiping indexer config. The sync now reads categories fromcapabilities, scoped against a registered book application (Readarr/LazyLibrarian) where one exists and falling back to the indexer's own book/audiobook capability categories for standalone Prowlarr setups. The syncer also no longer removes indexers when a sync matches nothing at all, so a future filter regression cannot wipe indexer config again. -
qBittorrent 5.1.4 category health checks no longer fail (#793) — qBittorrent 5.1.4 can return a boolean
download_pathflag in category payloads, which made Bindery fail to decode the categories response (cannot unmarshal bool into Go struct field) and fail the category health check. Non-string category path fields are now ignored while the string save path is still read. -
Readarr migration imports author names correctly (#784) — the Readarr import queried a non-existent
Authors.Namecolumn; in Readarr's schema the author name lives inAuthorMetadata, joined viaAuthors.AuthorMetadataId. The import now performs that join, so migrated authors keep their names. -
Telemetry server stages its database backup next to the database (#777) — the telemetry server wrote its backup to
/tmp, which could fail or cross a filesystem boundary; the backup is now staged alongside the live database.