Adguard-home - Liveness probe failed / Connection failed

[mr-onion-2]

Hi

Apologies if this is not the right place to ask, since this is probably more due to my lack of experience with Helm rather than any issue with this chart.

I am running a clean k3s environment on a Pi4. Just the one node for now.

The only values I have overridden in the adguard-home chart are the image tag (armhf-latest) and the existingClaim.

I can port-forward to the pod on port 3000 and access and complete the setup wizard at /install.html, but after this it says the connection was reset. Then eventually the liveness probe fails with e.g. Get http://10.42.0.70:3000/login.html: dial tcp 10.42.0.70:3000: connect: connection refused.

Am I missing something in my environment setup?

Environment:
Raspberry Pi4
5.4.73-v8+ #1360 SMP PREEMPT Thu Oct 29 16:00:37 GMT 2020 aarch64 GNU/Linux
Running k3s version v1.18.9+k3s1 (630bebf9)

Many thanks

[dcplaya]

During the setup wizard, do you keep the port as 3000? Or do you change it to something else?

Do you have an external IP setup for AdGuard?
Do you have ingress set up for it?

[mr-onion-2]

Good call - I had left it as port 80. Changing it to 3000 has done the trick - thanks!

Aside from that I hadn't enabled my configAsCode block so I shouldn't have had the wizard anyway.

Edit - another thing if I may, but if there is a better place for such 'support' questions please do let me know...

I have installed the nginx-ingress and metallb charts, giving me an externalIP of say x.x.x.200. I have mapped this IP to the hostname 'adguard.local' in my hosts file, and this is the what I use to access the admin interface. Port 53 is exposed on the adguard-home-udp service and it is a NodePort, so the node's IP should be the one used as the DNS server right? I tried setting my router to use the node's IP as the DNS server but it does not appear to be working. In fact, a connectivity test from my router to port 53 on the node IP returns a 'connection refused'.

Many thanks

[dcplaya]

The wizard runs if there is no current config, what you enter in the wizard is basically making the starting config, which should be almost identical to the configAsCode. I worried about this exact problem when I made this chart but there was no way around it.

As for the port 53 issue,
You probably want to use your externalIP for the DNS lookups. I dont know your setup exactly but I use the externalIP. One way to check is to look at the kube services for that namespace. kubectl get svc -n NAMESPACE You should see port 53 listed by the adguard-home-tcp and adguard-home-upd along with an IP.

[mr-onion-2]

Yeh I configured my configAsCode block from my old config file but I just hadn't noticed the 'enable' flag, which is what threw me off a bit.

I'll have another play with the port 53 service bit and will close this issue for now as I think it's more likely something wrong in my overall set up. If I do pinpoint the issue I'll put a comment on for future reference.

Thanks again for your help and the great project.

[dcplaya]

Feel free to join the discord! Might get quicker help or just come hangout
https://discord.gg/tfAZPvSS

[mr-onion-2]

Thanks, might just do that :)