GNU bug report logs - #75144
[PATCH] machine: Implement 'hetzner-environment-type'.

Previous Next

Package: guix-patches;

Reported by: Roman Scherer <roman <at> burningswell.com>

Date: Fri, 27 Dec 2024 16:48:02 UTC

Severity: normal

Tags: patch

Done: Ludovic Courtès <ludo <at> gnu.org>

Bug is archived. No further changes may be made.

Full log


View this message in rfc822 format

From: Roman Scherer <roman <at> burningswell.com>
To: Maxim Cournoyer <maxim.cournoyer <at> gmail.com>
Cc: Josselin Poiret <dev <at> jpoiret.xyz>, Simon Tournier <zimon.toutoune <at> gmail.com>, Mathieu Othacehe <othacehe <at> gnu.org>, Ludovic Courtès <ludo <at> gnu.org>, Tobias Geerinckx-Rice <me <at> tobias.gr>, Roman Scherer <roman <at> burningswell.com>, Christopher Baines <guix <at> cbaines.net>, 75144 <at> debbugs.gnu.org
Subject: [bug#75144] [PATCH] machine: Implement 'hetzner-environment-type'.
Date: Fri, 07 Feb 2025 14:00:37 +0100
[Message part 1 (text/plain, inline)]
Hi Maxim,

yes, it is fully automated. What happens is:

- a server is provisioned through the Hetzner API
- the the server is booted into the rescue system via the API
- partitions are setup in the rescue system (enlarged)
- a minimal Guix system is installed
- then the server re-booted, starting the minimal Guix system
- then the machine-ssh-environment takes over and applies the final system configuration
- this all is done once, when the server is initially provisioned

Previsouly I tried the guix-infect.sh approach that installs a Guix
system on top of a debian/ubuntu image, but I found this was very
brittle (issues with dns when you remove /etc, etc.). From my experience
working with this I found the approach with the rescue system both more
reliable and faster.

Does this mnake sense?

Roman

Maxim Cournoyer <maxim.cournoyer <at> gmail.com> writes:

> Hi Roman,
>
> Roman Scherer <roman <at> burningswell.com> writes:
>
> [...]
>
>> Things to improve another day:
>>
>> - Get Hetzner to add a Guix image to their collectin of supported images. That
>> would remove the need for using the rescue system to install an initial Guix system.
>>
>> - Installing the initial Guix system via the rescue system is kind of slow
>> (especially if there are no substituyes), and done in sequence. I'm not sure
>> how this could be parallelized with how things are invoke by guix deploy.
>
> Forgive my ignorance, but I thought the idea of a deploy <machine>
> environment type was to allow fully provisioning the OS via the service
> API?
>
> I haven't reviewed the change yet; perhaps you mean that currently such
> provision must happen by going through the rescue system path (but is
> still automated by this new environment type?)
[signature.asc (application/pgp-signature, inline)]

This bug report was last modified 122 days ago.

Previous Next


GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.