[openstack-dev] [Fuel] ML2 support in Fuel deployment risks

Mike Scherbakov mscherbakov at mirantis.com
Wed Jun 18 08:08:27 UTC 2014


Ok. If don't have more or less working solution by the beginning of next
week, let's start a second track to consume ML2 part only as a mitigation
plan.


On Tue, Jun 17, 2014 at 2:29 AM, Dmitry Borodaenko <dborodaenko at mirantis.com
> wrote:

> Correction/clarification: call tomorrow is about
> multiple-cluster-networks, and it was a bad idea on my part to try to
> hijack that with neutron-ml2 discussion. Lets not do that and continue
> discussing the blueprint spec in gerrit, and hopefully by Thursday
> Andrew will have enough code out there to make the discussion more
> concrete. Link:
> https://review.openstack.org/99807
>
> On Mon, Jun 16, 2014 at 3:07 PM, Dmitry Borodaenko
> <dborodaenko at mirantis.com> wrote:
> > Mike,
> >
> > We discussed this in our team syncup meeting earlier today. The
> > agreement was that HA is the biggest risk with the current approach.
> > However, keeping our current state of divergence from upstream (and
> > even exagerrating it further) leaves us with a huge technical debt, so
> > the tradeoff between that and potential new neutron deployment issues
> > is not that obvious. Andrew is confident that he can port our HA
> > deployment code over to the current upstream puppet-neutron by the end
> > of this week, he's now updating the spec per review comments from the
> > team and we will have another meeting tomorrow morning (8am PT) to go
> > over all concerns and risks.
> >
> > -DmitryB
> >
> > On Mon, Jun 16, 2014 at 3:57 AM, Mike Scherbakov
> > <mscherbakov at mirantis.com> wrote:
> >> Fuelers, Andrew,
> >> I've talked to Sergey V. today about ML2 support in Fuel. Our current
> >> approach [1] is to port upstream puppet module for Neutron which has
> support
> >> of ML2, however our Neutron module is significantly diverged from
> upstream
> >> one (at least for Neutron HA deployment capabilities), as far as I
> >> understand. Basically, there is a risk that we will get unstable Neutron
> >> deployment in 5.1. Also, unless we have ML2, we are blocking others who
> rely
> >> on it, for example Mellanox.
> >>
> >> To mitigate the risk, there is a suggestion to start the work in two
> >> parallel tracks: one is to continue porting of upstream puppet module,
> and
> >> another one - port the only ML2 part into Fuel Neutron puppet module.
> This
> >> will not take much time, but will allow us to have 5.1 reliable and
> with ML2
> >> in case of instability after porting external module.
> >>
> >> Your opinion on this?
> >>
> >> [1] https://review.openstack.org/#/c/99807/1/specs/5.1/ml2-neutron.rst
> >> --
> >> Mike Scherbakov
> >> #mihgen
> >>
> >>
> >> _______________________________________________
> >> OpenStack-dev mailing list
> >> OpenStack-dev at lists.openstack.org
> >> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
> >>
> >
> >
> >
> > --
> > Dmitry Borodaenko
>
>
>
> --
> Dmitry Borodaenko
>
> _______________________________________________
> OpenStack-dev mailing list
> OpenStack-dev at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>



-- 
Mike Scherbakov
#mihgen
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20140618/f91f2ffa/attachment.html>


More information about the OpenStack-dev mailing list