[openstack-dev] [kolla] Re: Questions about kolla

Steven Dake (stdake) stdake at cisco.com
Mon Mar 30 13:42:13 UTC 2015


Michal,

All great questions.  I have copied the OpenStack mailing list since these technical questions can benefit everyone.

From: <Stachowski>, Michal <michal.stachowski at intel.com<mailto:michal.stachowski at intel.com>>
Date: Monday, March 30, 2015 at 6:21 AM
To: Steven Dake <stdake at cisco.com<mailto:stdake at cisco.com>>
Cc: "Chylinski, Arek" <arek.chylinski at intel.com<mailto:arek.chylinski at intel.com>>
Subject: Questions about kolla

Hi Steven

I am interested in Kolla project and I’ve got few questions about it. Could You tell me:

·         What is state of whole project? What issues do You have?

Most issues we run into are actually docker issues.  For example, we had a bug where mariadb was not set to always-restart.  Rebooting a node would crater docker.  This is but one example of many.  To be fair the Docker community is super responsive to bug reports and debugging the problems I find :)


·         What is scope of Kolla? Which services do You plan dockerize?

We plan to dockerize any service needed to deploy OpenStack.  We haven’t decided if that includes ceph, since ceph may already be dockerized by someone else.  But it does include the HA services we need as well as the rest of the OpenStack services.


·         As I can see in docker-registry You’re providing centos and fedora images, but in files from git repositorie You are using only centos. Is It mean that fedora images are depracted?

I just don’t build fedora images because they are larger.  To build fedora images, in
.buildconf set
PREFIX=fedora-rdo-

And it will build fedora image


·         Do You plan release Kolla as a part of whole OpenStack (for example: in Liberty release)?

Kilo 2015.1.0 will be our first release


·         Do You plan support Mesos instead of Kubernetes?


Kubernetes has been deprecated because it doesn’t provide super privileged containers.


·         Do You plan support containers with bridge instead of host network mode?


Docker-proxy adds about 20 microseconds to each network packet.  While that doesn’t sound like much it adds up.  The main reason we are not using docker-proxy and containers with bridging has to do with the fact we couldn’t get Neutron to work in this environment in the past (when we were using k8s).  Neutron creates its own network namespaces, which requires host networking mode.

Regards
-steve

Thanks for Your help ☺

Kind regards
Michał Stachowski
Undergrad Intern Technical
Software Assurance Administrator – Cloud Platform Group – Data Center Group
Intel Corporation - Gdańsk


---------------------------------------------------------------------
Intel Technology Poland sp. z o.o.
ul. Słowackiego 173 | 80-298 Gdańsk | Sąd Rejonowy Gdańsk Północ | VII Wydział Gospodarczy Krajowego Rejestru Sądowego - KRS 101882 | NIP 957-07-52-316 | Kapitał zakładowy 200.000 PLN.

Ta wiadomość wraz z załącznikami jest przeznaczona dla określonego adresata i może zawierać informacje poufne. W razie przypadkowego otrzymania tej wiadomości, prosimy o powiadomienie nadawcy oraz trwałe jej usunięcie; jakiekolwiek przeglądanie lub rozpowszechnianie jest zabronione.
This e-mail and any attachments may contain confidential material for the sole use of the intended recipient(s). If you are not the intended recipient, please contact the sender and delete all copies; any review or distribution by others is strictly prohibited.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20150330/3fd3afa8/attachment.html>


More information about the OpenStack-dev mailing list