Possibly incorrect nginx configuration in docs when using jellyfin in a docker container

[Oni]

I've installed jellyfin in a docker container following the official documentation (https://jellyfin.org/docs/general/installation/container/) and using the official image:

docker run -d \
 --name=jellyfin \
 -p 8096:8096 \
 -v /my_config:/config \
 -v /my_cache:/cache \
 -v /my_media:/media:ro \
 -e PGID=my_pgid -e PUID=my_puid  \
 -e TZ=Europe/Rome \
 --restart unless-stopped \
jellyfin/jellyfin:2024031105

If I try to use the nginx config provided in the docs (https://jellyfin.org/docs/general/networking/nginx/) and I try to reach jellyfin container (jellyfin.mydomain.com or jellyfin.mydomain.com/web/) it fails (error 404).

It turns out that the offending line is:

# location block for /web - This is purely for aesthetics so /web/#!/ works instead of having to go to /web/index.html/#!/
    location = /web/ {
        # Proxy main Jellyfin traffic
        proxy_pass http://$jellyfin:8096/web/index.html;
       ...
    }

It works with:

# location block for /web - This is purely for aesthetics so /web/#!/ works instead of having to go to /web/index.html/#!/
    location = /web/ {
        # Proxy main Jellyfin traffic
        proxy_pass http://$jellyfin:8096/web/;
       ...
    }

Is the nginx documentation outdated? Or is the official image messing with "index.html"?

[FrostyZoob]

Piggybacking on this thread since I encountered similar issues when using the supplied nginx reverse proxy config here:

https://jellyfin.org/docs/general/networking/nginx/#nginx-from-a-subdomain-jellyfinexampleorg

tl;dr: Using the nginx config as shown results in a 404. The fix was to simply delete the = / location block entirely. I'm not an nginx expert but it seems unnecessary. Can one of the devs speak to it's purpose?

Details:

I have nginx running on a separate server named docker-prod-01.mydomain.tld
I installed jellyfin via it's official docker container on a server named docker-prod-02.mydomain.tld

I verified that Jellyfin works by going to http://docker-prod-02.mydomain.tld:8096. It finds my media, plays it back...all the things. All good.

(As a quick aside: I noticed that when visiting the base URL, Jellyfin would redirect me to /web/index.html.)

Now I want to get it working through a reverse proxy so I have https and a simpler URL: https://media.mydomain.tld

I have a valid SSL cert for my domain created via acme.sh. It's installed on the nginx server (docker-prod-01) and it's used by other services proxied by nginx. I have verified that it works.

I applied the config shown here for "Nginx from a subdomain (jellyfin.example.org)" and made the bare minimum changes to get it to work for my setup: jellyfin.conf.clean.txt

I'd visit http://media.mydomain.tld and notice a redirect and then...blank page.'

Time to break out curl. I run the following command:

curl -I -L https://media.mydomain.tld

And this is what I get: nginx-borked-location.txt

After some experimenting, I deleted the = / location block entirely, restarted nginx, and visiting https://media.mydomain.tld redirected me to /web/index.html and I got the login screen.

Edit: Here's the output of curl -I -L https://media.mydomain.tld after the fix:
nginx-success.txt

Edit 2: I was using Jellyfin 10.8.13

[Oni]

tl;dr: Using the nginx config as shown results in a 404. The fix was to simply delete the = / location block entirely. I'm not an nginx expert but it seems unnecessary. Can one of the devs speak to it's purpose?

Apparently the NGINX config changed since when I've reported this bug.

The "web/" location is gone, so the "= /" is incorrectly redirecting to the missing "web/" location.