[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]
Re: [atomic-devel] Multi-node deployment strategies
- From: Ohad Levy <olevy redhat com>
- To: Colin Walters <walters verbum org>, SGhosh <sghosh redhat com>, atomic-devel projectatomic io
- Subject: Re: [atomic-devel] Multi-node deployment strategies
- Date: Wed, 03 Dec 2014 09:46:01 +0200
On 12/03/2014 05:23 AM, Colin Walters wrote:
> On Mon, Nov 10, 2014, at 06:42 PM, SGhosh wrote:
>
>> On the prod side - for host deployment - would like to see Foreman support
>
> I looked at this briefly, from what I can tell they don't have cloud-init support.
We actually have cloud init support in general, and additional code per
provider (ec2, openstack, libvirt etc).
> An open ticket I found here is http://projects.theforeman.org/issues/8289
yes, ovirt is lacking due the api binding, this I assume will be fixed
for 1.8.
> That's not the end of the world; kickstart can be used for as normal for physical installs, but a major goal I'd like to achieve for the Fedora 22 cycle is making it easy to use cloud-init on bare metal. I'll post more about that soon, but it'd unify things a lot more across deployment models.
that's interesting in general, who would be providing the metadata service?
>
> The Foreman puppet bits requiring Ruby on the host are going to be kind of a showstopper on that side - I think there's been some investigation of puppet in privileged containers?
We are slowly adding more CM agents support (salt, chef etc) is there
any specific requirement / need?
for example, we support deploying rhev-h, but simply the puppet features
are irrelevant for this kind of deployment.
Ohad
[Date Prev][Date Next] [Thread Prev][Thread Next]
[Thread Index]
[Date Index]
[Author Index]