Chris wrote:
Since Jellyfin works through the same chain, this is almost certainly a Home Assistant thing rather than SparkBox or your proxy setup. Home Assistant refuses connections that don't look like they're coming straight from a trusted address once it's behind a reverse proxy - and that shows up exactly as a blank 'no available server' page. In Home Assistant's own configuration.yaml, under the http: section, you need usexforwardedfor: true and trustedproxies listing your proxy's IP (not just allowing http, which you already did). That's the piece people usually miss. Give that a try and let us know.
Dioniso wrote:
Hey, sorry I kinda vanished... 馃槄 just to update you on the matter. I actually tried your suggestion but it didn't solve the problem and I actually started experiencing other unrelated bugs with some integrations. I got really frustrated and ended installing HAOS on a different device, never got any problem ever since and the routing works a charm. I concluded that HA seams not really stable in its docker version, but it's just my experience. Thanks for the help!
Chris wrote:
Thanks for coming back to tell us, and glad Home Assistant is running smoothly now. Sorry my earlier tip didn't get you there. Looking at it again, I pointed you at the wrong layer. The blank "no available server" page doesn't come from Home Assistant, Nginx Proxy Manager or SparkBox. It's the message the proxy on your VPS (Pangolin) shows when it can't reach the target its entry points at, so those visits never got as far as your box. The trustedproxies setting I mentioned is real, but it fixes a different symptom: a plain "400: Bad Request" page from Home Assistant itself. That's why changing it made no difference for you. Nothing to do now, since HAOS on its own device works. If you ever move Home Assistant back onto the box, the quick way to tell which problem you have is to load the page through your domain once, then run: sudo docker logs --since 10m sb-homeassistant 2&1 | grep -c "not set-up for reverse proxies" Above 0 means it's the Home Assistant setting. 0 means the problem is on the Pangolin side. I've added both cases to what the assistant knows, so the next person with this setup gets pointed at the right layer the first time.