Problems with nginx reverse proxy and calibre

[HuckinFappy]

I've recently set up nginx as a front end reverse proxy for all my internal servers (plex, the *arrs, calibre/calibre-web). I'm sure this is probably an nginx config question, but I'm starting here since I got all of the other subdomains to work over HTTPS using certbot and LetsEncrypt, except for calibre 7.2.

All servers run in docker containers on an Ubuntu 23.10 (mantic) host and nginx version 1.24.0-1ubuntu1, and are cleanly accessible from the internal network with no problems without the proxy...the host they run on is not using a firewall at this time (the router is, of course). With ports 443, 8080, 8081, and 8181 open on the router (the latter 3 opened during debugging, probably unnecessary) I get to the server, but I get a KasmVNC error:

(domain name blacked out for privacy)


Here's my basic nginx configuration for this subdomain, based on the Calibre documentation, but with SSL configuration added:



The Chrome developer tools show the following:


I don't know if this is a red herring or if there's websocket configuration I need to add (that is not mentioned in the documentation though), but that would be new to me, so i could use some assistance if so. I suspect this may be the issue, since trying Calibre 6.x and 5.x versions give similar errors.

I was able to set up the proxy to forward to the content server instead (port 8081), but that's not the interface I'm looking for. I use Calibre-web for that functionality, and am looking for the main Calibre UI for full remote library maintenance when on the go

Thanks for any ideas, and let me know if more information is needed.

[kovidgoyal]

I have no idea what that is. There is no main.bundle.js in the calibre content server.

[HuckinFappy]

I was right...it was the websockets. I needed to add the following to the location block for my nginx virtual server:

proxy_http_version 1.1;
proxy_set_header Upgrade $http_upgrade;
proxy_set_header Connection $http_connection;

[GinoGinelli]

Dead post but commenting to say I was having the same problem and this was the solution.

I did not seem to need 2 of the directives mentioned and it worked fine with only
proxy_set_header Upgrade $http_upgrade;

I could remove the following 2 and still load calibre fine:
proxy_http_version 1.1;
proxy_set_header Connection $http_connection;

This was using lscr.io/linuxserver/calibre and I just used NGINX 1.18 from the default ubuntu repo (on ubuntu focal)

Cheers HuckinFappy

[pandako]

Adding that I further needed to change the KasmVNC settings (in the web UI where the error is) so that the "websocket path" value was "calibre/websockify" (i.e. prepending my nginx proxy path) instead of the default value "websockify". Otherwise the web GUI was trying to connect to domain.com/websockify instead of domain.com/calibre/websockify.

[p8d]

Originally Posted by GinoGinelli

Dead post but commenting

The word here is necro. In use as necro'ing occasionally without apostrophe or necrod.

[DNSB]

Originally Posted by p8d

The word here is necro. In use as necro'ing occasionally without apostrophe or necrod.

Not sure if posting in a thread where the last post from 2023 was on 2023-Dec-31 qualifies as necro posting. And the posts you are complaining about were from a poster on 2024-Apr-26 who found the contents of the thread helped solve their problem and another poster further commenting on changes they made for fix the same issue on 2024-May-09.