modules: introduce hostctl (static dns for development)

[blaggacao]

hostctl module allows to temporarily set static dns entries for development per project

[hostctl]
enable = true
dns."test.domain.local" = "172.0.0.1"
dns."shared.domain.link-local" = "169.254.0.5"

What we get:

$ nix develop
warning: Git tree '/home/blaggacao/ghq/github.com/numtide/devshell' is dirty
[hostctl] Update static dns entries...
[sudo] password for blaggacao:
[hostctl] +----------+--------+-------------+--------------------------+
[hostctl] | PROFILE  | STATUS |     IP      |          DOMAIN          |
[hostctl] +----------+--------+-------------+--------------------------+
[hostctl] | devshell | on     | 169.254.0.5 | shared.domain.link-local |
[hostctl] | devshell | on     | 172.0.0.1   | test.domain.local        |
[hostctl] +----------+--------+-------------+--------------------------+
🔨 Welcome to devshell

[dns]

  hostctl       - Your dev tool to manage /etc/hosts like a pro!

[formatters]

  nixpkgs-fmt   - Nix code formatter for nixpkgs

[general commands]

  hello         - prints hello
  menu          - prints this menu

[linters]

  golangci-lint - golang linter

[utilites]

  hub           - github utility

$

[roberth]

This seems hard to get right, particularly because when the user runs two dev-shells and exits one.

Modifying the system is a bit heavy-handed for a development shell. I was secretly hoping for an unprivileged solution when I saw this PR. Perhaps (userns) containers are a more appropriate solution for most use cases.

[blaggacao]

There is indeed a long term heavily politically involved solution to this: https://github.com/apetresc/nss-hostslocal. That would require lobbying for changing the way how linux system generally resolve hostfiles these days.

This particular PR is not going to get merged for this reason.

However, in a dev team setting, where you can exert certain powers over an individual's machine, this is a solution that (somewhat) works.

[zimbatm]

Do you think this PR is worth pursuing? It would need to have some tests.

[blaggacao]

It might. I can try to free some time in the next few month, please keep nudging in case I'm not reactive 😄

[bobvanderlinden]

Have you considered using something like userhosts? https://github.com/figiel/hosts I packaged it in the past for the purpose of using it inside a shell.

It uses ldpreload to override a number of the DNS resolving functions. It doesn't need root and is basically set/reset using envvars.

That said, at the moment it doesn't support osx. Also, if you have a webbrowser open already, it won't pick up the new hosts, as they are only available within the shell.

[bobvanderlinden]

I created #225 based on libuserhosts as a different implementation of configuring hosts entries.