[Pkg-nginx-maintainers] Bug#1021393: closed by Thomas Ward <teward at thomas-ward.net> ()

[Thomas Ward]

Stuart,

With my Ubuntu Core Dev hat on and Ubuntu Server Team hat on, the Ubuntu 
team decided that disabling IPv6 on a server is an administrator 
decision that is "not the defaults" and the default installation 
configuration is designed as that - the 'default' that functions in a 
standard-expected installation.  Full disabling of IPv6 was considered 
"atypical default configuration".  This was discussed at length in 
multiple bugs on this nature.

With my Debian Maintainer hat on:

I've discussed this in the #debian-devel channel on IRC, and the 
responses I have are: "IPv6 is always enabled, you should use it in the 
default config", and "if you disable ipv[46] totaly, I don't think one 
can expect the system to work without bugs."

The consensus is identical among multiple users and Debian Developers, 
as the configuration you have chosen *diverges* from the defaults and is 
a "user/admin decision" that the user/admin will have to accept 
consequences for.

As such, this echoes the Ubuntu Server Team's decision but at the Debian 
level, and we're not changing this for Bullseye or any of the other 
supported Debian releases.

You need to configure your deployment scripts to ignore the first 
package installation failure, replace your configuration entirely, and 
*then* `apt install -f` or similar in your deploy script to *replace* 
the configuration.

---

(TL;DR, The default configuration file is made to adapt to *default* 
Debian environments, and it's non-standard to disable ipv[46] entirely 
at a sysctl and similar level, so you can *expect* multiple packages to 
have bugs when these nonstandard setups are being used.  ALTERNATIVELY, 
your deployment script should install nginx first, change the 
configuration, disable IPv6, then reboot.  That's how you *should* do 
things with an autodeploy script).



Thomas Ward

Ubuntu Core Developer (https://launchpad.net/~teward)

Debian Maintainer (for nginx package among others)

On 10/7/22 14:37, Stuart Culligan wrote:
>
> Ok I guess I should have been more clear. This is a debian package 
> issue in that the debian package is linking the default site default 
> -> /etc/nginx/sites-available/default Which contains 2 listen lines as 
> I listed before
>
> listen 80 default_server; listen [::]:80 default_server; After digging 
> further into this I discovered it was IPv6 line that was causing the 
> issue. listen [::]:80 default_server; We disable IPv6 on our servers. 
> I did not run into this on Ubuntu 18.04. It appears there was a lot of 
> changes to how debian packaged nginx between the 2 OS releases. With 
> these changes there appears to be a lot of changes to the default. I 
> suggest not trying to link the default and not attempting to do a 
> restart. Just a suggestion though.
>
> Now I am trying to find a way to get passed this issue as we build an 
> entire server via debian packages. I can not have a package error out 
> in my build process.
>
> On 10/7/2022 11:45 AM, Debian Bug Tracking System wrote:
>> This is an automatic notification regarding your Bug report
>> which was filed against the nginx-common package:
>>
>> #1021393: error in package nginx-common_1.18.0-6ubuntu14.2_all.deb for ubuntu 22.04 with default enabled site.
>>
>> It has been closed by Thomas Ward<teward at thomas-ward.net>.
>>
>> Their explanation is attached below along with your original report.
>> If this explanation is unsatisfactory and you have not received a
>> better one in a separate message then please contact Thomas Ward<teward at thomas-ward.net>  by
>> replying to this email.
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/pkg-nginx-maintainers/attachments/20221007/37743431/attachment.htm>