Jellyfin 12.0

Clients that have not seen an update in years are the ones at risk here. This release contains security fixes, so we recommend upgrading rather than staying on an older release once you are ready. As always with major Jellyfin releases, bugs will exist.

The leading 10. never changed and never told anyone anything, so all it did was push the number that actually mattered into the middle position and make "major" releases look minor. Dropping it means the first number moves when the release is big. If you maintain anything that parses Jellyfin version strings - a client, a monitoring check, a deployment script, a container tag pin - this is the item to look at before upgrading. Tuning the new database Jellyfin 10.11.0 finished rebuilding how the library database works, which opened the door for the performance work in this release. We aren't finished, but you should notice a big difference when browsing. One of the largest changes is how playlists and collections are stored. Before, everything in a playlist was kept as one big list inside the playlist itself, and the database couldn't look into that list. Anything we needed to know meant loading the whole playlist and unpacking it first. Getting the item count, working out how much you've watched, or showing a single page all cost the same as loading the entire thing. Editing was the same story, because adding or removing one item meant writing the whole list back out. Every item in a playlist is now its own row. The database can count rows, return one page, and add or remove a single item without touching the rest, so this should be a big improvement on large playlists. Collections and boxsets were stored the same way and get the same fix. Alongside that: Deleting a large number of items at once no longer fails partway through. Continue Watching, Next Up, rewatching, Latest Media for music, artist lookups, and the watched counts on folders should be faster. Heavy database maintenance no longer runs during a library scan, so the two stop competing for the same resources. Most of this shows up as pages that used to freeze no longer freezing. One thing to note is that this work went into how Jellyfin reads your library, not into the library scanner. You may find the scanner is a little quicker as a side effect, but speeding it up was not a goal this time around. What runs on first boot As usual, a new release of Jellyfin comes with multiple database migrations. These migrations will take a while to complete depending on the size of your library and how much bad data has been accumulated. If you would rather run that step deliberately than have it happen on first start the server now accepts --mode MigrateSystem, which performs the upgrade and exits without starting the rest of Jellyfin. Multiple versions for episodes Alternate versions have been a movies-only feature since they were introduced. In 12.0 they work for episodes as well, so a series with a broadcast cut and an extended cut, or a 1080p and a 4K copy of the same episode, can be grouped the way movies always could. This includes resume data that follows the version you were actually watching. This is also the feature behind the required post-upgrade scan: making versions work correctly for episodes meant fixing how version links are stored, and automatically resolved versions have to be rebuilt from the files on disk. Books and comics, properly this time Books have often taken a backseat in favor of video playback in Jellyfin, and this release is the start of an effort to change that. Most of what the Bookshelf plugin used to do now has been moved to the server itself. On the server: Book metadata is read directly from OPF and ComicInfo files, or from ComicBookInfo comments, with no plugin required. Posters are generated for EPUBs and every supported comic archive format, and external covers work for audiobook files. Name, index, year, and series are read from book filenames, and volume and chapter numbers are picked up from comic filenames when they are there. Page counts are extracted from comic archives and PDFs.