Chris wrote:
You're not missing anything — the Ebooks app is exactly two containers, Calibre-Web and Kavita. There's no Calibre Content Server in the stack today, and nothing in it watches a folder and ingests into the Calibre database, which is why everything lands by hand. Worth knowing: full Calibre already ships inside the Calibre-Web container, not just the web UI — so the gap is the server piece, not Calibre itself. That makes your pipeline a reasonable thing to wire up properly, and I'm logging it as a feature request for Tom; he reads these himself and does act on them. One thing to help him scope it: Readarr isn't a SparkBox app, so how are you running it alongside the box?
Big_Totoro wrote:
Hey Chris, I'm running Readarr as a test under docker in another LXC container, and also under TrueNas which is running in a VM. All on the same server hardware. I'm passing mounts across LXC containers giving sparkbox apps access to the torrents folder and media folders on trueNas and unraid. Unraid is on separate hardware. I'm not doing any kind of storage inside of my sparkbox, it's all external.
Chris wrote:
That detail helps, thanks — logged with Tom, and the all-external, mounts-passed-between-LXC shape is exactly what he'd want for scoping this. One thing that may unblock half of it today: Kavita mounts your Books Folder directly, the same path Calibre-Web gets, so Kavita never touches the Calibre database at all. If you point the Books Folder setting (BOOKSPATH) at the folder Readarr drops epubs into and run sudo sparkbox up, Kavita will scan them where they land — no manual upload for the organise-and-scrape end. Calibre-Web is the one that genuinely needs a metadata.db, and up seeds an empty library there if none exists. So the by-hand step only remains for Calibre-Web itself.