[libvirt-ci PATCH 11/12] config: Move the virt-install settings from install.yml to the config

[Andrea Bolognani]

On Mon, 2020-05-11 at 10:08 +0200, Erik Skultety wrote:
> On Thu, May 07, 2020 at 06:56:44PM +0200, Andrea Bolognani wrote:
> > On Wed, 2020-05-06 at 14:06 +0200, Erik Skultety wrote:
> > > +  # Settings mapping to the virt-install options - see virt-install(1).
> > > +  # It is strongly recommended that you keep the following at their default
> > > +  # values to produce machines which conform to the upstream libvirt standard,
> > > +  # unless you have a reason to do otherwise.
> > > +  #
> > > +  # Sizes are expressed in GiB.
> > > +  #
> > > +  virt_type: kvm
> > > +  arch: x86_64
> > > +  machine: pc
> > > +  cpu_model: host-passthrough
> > > +  vcpus: 2
> > > +  memory_size: 2
> > > +  disk_size: 15
> > > +  storage_pool: default
> > > +  network: default
> > 
> > Now is a great time to delete group_vars/all/install.yml, since
> > you've just made it obsolete and we don't want to have the same
> > information stored in two places.
> > 
> > Note that there's one use of install_* variables that would be broken
> > by doing so in playbooks/update/templates/bashrc.j2, but I've already
> > posted a patch[1] that takes care of that, so as long as that goes in
> > before your series we don't have to worry about it :)
> 
> Would it? I knew about the MAKEFLAGS thing, but I expected that to simply work
> once I pass the extra variables,

Yeah, you're right, it *would* work: I was mistakenly thinking that
the additional variables would only be used when installing, but I
see now that any time Ansible is called it would have access to them
as well.

Anyway, I've already merged the other patch, so it no longer
matters :)

-- 
Andrea Bolognani / Red Hat / Virtualization