|
the cleanest way to get what youre describing is to stop thinking of the windows pc as the server at all. keep calibre on your laptop/desktop as the place where you import, tag and fix metadata, and let your always-on home box be the thing that actually serves the library 24/7. that way the windows machine can sleep whenever and remote access still works.
for the server side, calibre-web (or COPS if you want something lighter) is the usual pick over the built-in content server because it has proper user logins and a nicer browser UI for reading on a phone. the part that bites people is exactly the sync mess you saw earlier in this thread, those .fuse_hidden files happen because calibre-web has the database open while a sync is writing to it. the trick is to push the library to the server one direction only and ideally pause calibre-web during the copy, or just accept a short reload after. dont let two things write to the same metadata.db at once and most of the corruption worries go away. and definitely put a login + https in front of it before you open any ports.
if you dont feel like wiring up docker, reverse proxy, certs and logins by hand, full disclosure im involved in an open source project called syncloud that does the boring parts for you. its a self-hosting platform with one-click installs and calibre-web is one of them, it handles the https and auth side automatically and runs fine on a raspberry pi or an old mini pc, so it might save you the yak-shaving. either way the local-calibre-plus-server-side-calibre-web split is the setup youre looking for.
|