Don't set proxy_bind and instead rely on default behaviour.

[Irishsmurf]

Setting proxy_bind to $server_addr can cause issues if the connecting client is attempting to connect to a different IP protocol address than the backend supports (IPv6 vs IPv4)

This removes explicitly setting the proxy_bind setting, and instead relies on the default nginx behaviour (of binding to the same protocol as the upstream service)

Description

calibreweb currently doesn't work with IPv6 as internally, CalibreWeb binds to an IPv4 address 127.0.0.1 - and if you attempt to hit Nginx with a client using IPv6 - $server_addr will be an IPv6 address - preventing it to talk to the upstream IPv4 address

Fixes issues:

• Issue [BUG] Calibreweb Nginx configuration doesn't work for IPv6 #1071 1071

Proposed Changes:

• Changing /scripts/nginx/calibreweb.sh to remove setting proxy_bind, and instead rely on default behaviour.

Change Categories

• Bug fix

Checklist

• [✅] Branch was made off the develop branch and the PR is targetting the develop branch
• [✅] Docs have been made OR are not necessary
• [✅] Changes to panel have been made OR are not necessary
• [✅] Code conforms to project structure (See more)
• [✅] Prints to terminal are handled (See more)
• [✅] I have commented my code, particularly in hard-to-understand areas
• [✅] Testing was done
  • [✅] Tests created or no new tests necessary
  • [✅] Tests executed

Test scenarios

I manually modified the Nginx Configuration.

1. Without the change, it errors with 500 Internal Error
2. With the change, the page loads over both IPv4 and IPv6

Architectures

	amd64	armhf	arm64	Unspecified
Jammy
Focal
Bookworm
Bullseye
Buster
Raspbian	⚫️		⚫️	⚫️

✅❎ Passed

🛠🛠 TODO

❌❌ Currently failing

[liaralabs]

Thanks for doing all the legwork on this.

Can you please supply a script for the update directory and add a case to the existing calibreweb.sh?

I'll leave the exact mechanism up to you, but the gist is that we should check wither proxy_bind is in the nginx conf file and set the config correctly.

A simple mechanism would be to simply trigger the nginx/calibreweb.sh script as it should do what we need to already

[Irishsmurf]

Hey,

I've updated calibreweb.sh which adds a case which will overwrite an existing nginx config for CalibreWeb with one which removes the proxy_bind setting.

I took inspiration from mango.sh - which also overwrites an existing nginx config.

Is there something specific that's needed to be done to trigger this, or will this just run when calling box update?
I took a look at the update directory and it seems everything is merely tied to an app-specific shell script - and there's no triggers that I can see outside of that.

[liaralabs]

Yeah, you don't need to worry about that -- the update/upgrade hooks run all scripts in the update directory so nothing further than writing the hook is involved.