I upgraded to v6 today and have encountered an issue where the webview doesn't load/render anymore. It's not a 404 or a redirect, when I navigate to https://my.domain/admin/ I'm just provides the folder structure for the admin folder. Calling the API routes does 404. Using an explicit port does not provide access to the webview, either, as noted below.
I did update the pihole.toml file to replace the pi.hole domain with my own, and confirmed it did select new ports (8080, 8443) as apache is serving other websites on this raspi. This is how v5 worked as well.
Apache does have an entry for my domain, removing that completely breaks access to the /admin folder. The fact I can access that folder via HTTPS is a main reason I'm bewildered.
I had one error show up in the debug log about the cert mismatching, and resolved that with another post.
I do not recall getting a prompt at all about lighttpd during the v6 upgrade, which I saw mentioned in some other posts. It doesn't appear to be needed nor is taking up any ports per the debug log.
The backend service appears to be running just fine, catching queries without issue. This is just an issue with loading the webserver in the browser.
The webserver log information states only the following entry each time it logs:
Initializing HTTP server on ports "8080o,8443os,[::]:8080o,[::]:84"
The lack of further errors has me thinking this isn't actually spinning up the webview at all, just silently initializing or otherwise I'm not accessing the webserver correctly. I attempted to run a repair in case I'm missing webserver files, but nothing changed. Perhaps the debug token can shed some light on the fact I'm simply missing configuration? Or is this just an instance of apache breaking the new webserver setup?
Expected Behaviour:
Webview should load and render after upgrading to v6 with a custom domain.
Actual Behaviour:
No webview is presented, only the folder structure.
Port 80 and 443 was already used by apache, Pi-hole grabbed 8080.
[✓] tcp:0.0.0.0:8080 is in use by pihole-FTL
tcp:127.0.0.1:3306 is in use by mariadbd
[✓] tcp:[::]:84 is in use by pihole-FTL
[✗] tcp:*:80 is in use by apache2 (https://docs.pi-hole.net/main/prerequisites/#ports)
[✓] tcp:[::]:53 is in use by pihole-FTL
tcp:[::]:222 is in use by sshd
[✗] tcp:*:443 is in use by apache2 (https://docs.pi-hole.net/main/prerequisites/#ports)
Thanks for responses! However, as mentioned I did try using the explicit ports, and that did not load the webserver either. Both with my own domain and with pi.hole. While my domain fails to load anything on 8080 or 8443, using pi.hole errors due to a timeout.
I had modified the config to replace pi.hole with my own domain instead, including adding the updated .pem file, so I was hoping to be able to access it via a custom domain as I did on v5. Please note, I have no issues specifying a port if needed! It just isn’t working after upgrading.
I've partially figured this out. I had setup ufw but did not explicitly add any allow rules for 8080/8443. That's half of my problem resolved as I can load, without certs, pi.hole:8080/admin, but I've not yet come across a solution to allow using a custom domain (and it's LE certs) alongside the required ports.
ufw allow 8080/tcp
ufw allow 8443/tcp
The other half of the problem was that my router was also not setup to port the incoming connections to the raspi... since pihole is my DHCP, I didn't consider to update my router settings. Allowing the ports to forward to the raspi for 8080/8443 allows the custom domain name to load on the specific ports.
Additional notes for future travelers, since I'm using apache I could use the following reverse proxy to bypass needing the ports:
ProxyRequests off
SSLProxyEngine on
ProxyPass /admin https://<my.domain.com>:8443/admin
ProxyPassReverse /admin https://<my.domain.com>:8443/admin
ProxyPass /api https://<my.domain.com>:8443/api
ProxyPassReverse /api https://<my.domain.com>: 8443/api
