Re: [atomic-devel] (no subject)

[arnaud gaboury <arnaud gaboury gmail com>]

On Thu, May 31, 2018 at 8:09 PM Jonathan Lebon <jlebon redhat com> wrote:

Hi Arnaud,

On Thu, May 31, 2018 at 7:51 AM, arnaud gaboury
<arnaud gaboury gmail com> wrote:
> I decided to go Fedora Atomic and installed it on one VM. I have now the
> perfect setup for what I want and will now install Atomic with these setup
> to the other VM. What is the best way to clone my current machine and
> install to other? I couldn't find a simple method which would make sens to
> me. I was thinking of using git, but i can't find any guide, so my guess is
> this is not the proper way.

The "canonical" way is to use cloud-init for setting up auth, and then ansible
for configuration. So the first step in your case would be to convert your
customizations to an ansible playbook, and then provisioning is essentially
boot + run playbook. You can then use git to version your playbook. :)

If you'd like to modify the host in fundamental ways (e.g. adding/removing a lot
of base packages), you could also compose your own OSTrees. The Fedora
Atomic Host images are based on the manifest files from here:

https://pagure.io/fedora-atomic

I was maybe not clear enough and should have added that all my VM run already a fresh Atomic Fedora delivered by the hosting company.

So my idea is to "share" what I added in one machine (/etc settings and layered packages, let's call the machine "origin") to other machines (targeted VM) in an automatic way.

It seems that articles about automate building atomic host[0] will indeed do the trick. In this way, I will just need to install/configure origin and build the other according to. Same for any future changes.

The only thing I was thinking of adding is to put in the "middle" a github repo to store the machine state. This would be the path: machine origin ---> github repo "My OSTree" --> targeted VM. But maybe it is not possible and best is machine origin --> targeted VM.

Am I in the right path for what I want to do?

[0]https://www.projectatomic.io/blog/2017/02/automate-building-atomic-host/

There's no native way right now to just "clone" an existing machine and
duplicate it across VMs. You might have luck with e.g. snapshotting and then
using something like `virt-sysprep`, though I haven't tried it myself with AH.