Open Stack

[+ openstack-dev as suggested]

Steven,

Thank you for the candid comments and suggestions! We are revising our
mission to be openstack-centric as we plan our next phase. We will follow
the suggested steps by you and other TC reviewers that make sense for our
project.

On the specific question on Ansible playbook, it is good that we understand
why Kolla's Ansible is different. In Compass, we are trying to deprecate
ours and adopt os-ansible-deployment instead.

more in line:

On Mon, Aug 10, 2015 at 5:47 PM Steven Dake (stdake) <stdake at cisco.com>
wrote:

> Xicheng.  Comments inline.  This discussion should probably be on the
> openstack-dev mailing list, but I understand if you don’t feel comfortable
> asking such questions there.  IMO this is part  of the problem with Compass
> in big tent…
>
> I have copied Sam Yaple, one of our core reviewers because I think this is
> relevant to him.
>
> From: Xicheng Chang <mr.xchang at gmail.com>
> Date: Monday, August 10, 2015 at 2:46 PM
> To: Steven Dake <stdake at cisco.com>
> Cc: Weidong Shao <weidongshao at gmail.com>, Xicheng Chang <
> Xicheng.chang at huawei.com>
> Subject: Ansible playbook for openstack
>
> Hi Steven,
>
> This is Xicheng from Compass dev-team. I heard from Weidong that Kolla has
> its own Ansible playbooks for deployment and I would like to learn more
> about the following aspects:
>
> - What is the github url of this ansible project?
>
>
> http://github.com/stackforge/kolla
>
> We are not just an ansible project.  The goal of our project is deployment
> using thin container technology and providing a reference implementation of
> deployment tooling.  We fully expect TripleO will provide an integration
> with Kolla using puppet to our thin container technology.
>
>
> - Does it have anything to do with the upstream
> stackforge/os-ansible-deployment?
>
>
> No we are a completely separate project.  We can’t use OSADs ansible
> scripts because our Ansible implementation is tightly integrated with our
> container technology.  Maybe some day we can merge but getting the two
> independently formed communities on board with such a merger would be
> difficult.
>
>
>
> - What effort did your team make on ansible playbooks in order to get
> inducted to the openstack big-tent? I think it is required to use upstream
> deployment cookbooks/manifests/playbooks.(we currently have our own
> deployment repo at
>
> github.com/stackforge/compass-adapters)
>
>
> My take on the TC is they are willing to accept many different deployment
> tools into the Big Tent assuming they offer something unique and are a
> legitimate OpenStack project, following OpenStack processes, with a
> properly diversified community.
>
> Kolla’s big tent application for reference:
> https://review.openstack.org/206789
>
> I hope you don’t mistake my directness at answering the question I think
> you really want answered for rudeness, but as a casual observer of the
> Compass big tent application I noticed the following problems:
>
>
>    1. You have your own domain name vs using OpenStack infrastructure for
>    user interaction and marketing.  Openstack has all our infrastructure to
>    engage the community in one place not spread them all out all over.
>
> Good suggestion. We will move all related content to our wiki page on
OpenStack. This also helps us as it is easier to keep a single location up
to date and consistent than two.

>
>    1. According to Jay, compass installs all different kinds of
>    infrastructure.  This is a non-starter.  Generic tools have always been
>    rejected for openstack namespace.  It shows a lack of commitment to
>    OpenStack’s success and a desire to “hedge your bets”.
>
> The project is focused on OpenStack, but deployment spans many thing not
strictly OpenStack.  Just as Ironic installs baremetal systems, ours deploy
them.  Ironic is expected to be key in Compass going forward.  If a
customer wants to deploy a baremetal instance with a container with a
legacy app, we don't expect Compass to not allow it, but we won't be
providing special tooling to specifically enable it, either.

>
>    1. This email is an example of not participating in the open.  You
>    would get the same exact response from me on openstack-dev.  It would be
>    better if the entire community could learn from my opinions rather then a
>    couple people.
>
> correcting it right now.

>
>    1. I get that other systems like open daylight, NFV, and all the other
>    new networking systems are becoming popular.  It is appealing to want to
>    compete with these in the same deployment tool that deploys OpenStack.
>    That is a nonstarter.
>
> As I mentioned above,  the scope of Compass, as an (hoping to be)
Openstack project is being adjusted to focus on OpenStack deployment.

>
>    1. I think including CEPH is fine.  Everyone is in love with ceph.  We
>    are going to use it in Kolla for our persistent storage.  No openstack
>    project solves this problem in a suitable way.
>
> A course of action to correct the deficiencies pointed out by the TC:
>
>
>    1. Commit entirely to OpenStack in every way possible
>    2. Deprecate your website that is not an OpenStack infrastructure
>    property
>    3. Remove extensible parts of your system and make your tool only
>    deploy OpenStack and possibly do bare metal deployment using Ironic (not
>    cobbler)
>    4. Copy this email to the mailing list as an example of how
>    communication in the open should be done and you have learned from it :)
>    5. Focus only on OpenStack not  other projects like ODL, NFV, etc
>    which are essentially competitors to OpenStack’s basic service Neutron.
>    OpenStack has enough challenges without throwing ODL into the deployment.
>    I speculate this is why Kyle commented on the review even though he is not
>    on the TC.
>    6. Make a new mission that keeps your scope focused on OpenStack based
>    upon what your development community wants.  Let your community define the
>    mission.  Example: https://etherpad.openstack.org/p/kolla-manifesto
>
>
> I’m not certain making these changes will result in acceptance by the TC.
> They may find other deficiencies that need to be corrected once these are
> corrected.  This course of action may wreck your strategic business plans
> and may result in also not being accepted to Big Tent.  But this is the
> course of action I’d take if job #1 was OpenStack Big Tent.  YMMV. :)
>
> Bet the farm on OpenStack deployment.  Make other deployment tools for
> other systems.  It should be natural that a deployment tool only deploys
> the system it is targeted towards.
>
> I hope someday compass can enter the Big Tent.  I want choices in
> OpenStack deployment – it makes OpenStack more attractive.  Of course I
> want people to choose Kolla, but that’s a personal mission sort of thing.
>
> Hope this was helpful.
>
> Regards
> -steve
>
>
>
> Thank you,
> Xicheng
>
>
> --
> *-Xicheng Chang*
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20150812/b633435b/attachment.html>