[OpenStack-DefCore] [openstack-dev] [defcore][glance] Glare not defcore ready

Nikhil Komawar nik.komawar at gmail.com
Fri Apr 1 19:51:59 UTC 2016


Rochelle, thanks for replying. Glad to hear all the voices.

Response inline.

On 4/1/16 3:24 PM, Rochelle Grober wrote:
> Hi folks.
>
> I'm chiming in here from a systems engineering perspective.  I recently discovered that OpenStack-client is trying to build cross-project consistency into its design.  As such, it is opinionated, but this is good.  The glance team might consider also consulting the OSC team to ensure api structure compatibility with the OpenStack client.
>

+1 on consulting the OSC. I think the initiative is trying to capture as
much consistency across the OS ecosystem as feasible. So, definitely OSC
is a good & thus important pointer.

> Yeah, this is a lot to rollup, but the more consistency across projects, the friendlier it is to deployers and users, so heading that way soonest, but when/where it makes sense would be a *good thing*.
>

Makes sense on the applicability on the purist principles (in this case
consistency approach to the design).

I would to like to chat a bit more on the system engg. perspective as
well. I think there's a bit more to the design that consistency but all
the discussions will be quite subjective/abstract right now and
inappropriate for the ML discussions (possible cause of another set of
confusions).

If possible, let's all gather at the summit.

> --Rocky
>
>
> -----Original Message-----
> From: Nikhil Komawar [mailto:nik.komawar at gmail.com] 
> Sent: Friday, April 01, 2016 10:38 AM
> To: OpenStack Development Mailing List (not for usage questions)
> Cc: Mikhail Fedosin; defcore-committee at lists.openstack.org; Flavio Percoco
> Subject: Re: [OpenStack-DefCore] [openstack-dev] [defcore][glance] Glare not defcore ready
>
> Thank you for your emails Flavio and Mike. It's really good to get a
> clarity out there.
>
> Hence, yes, the intent of the DefCore meeting was to get more "clarity"
> on the entire situation and making sure that the project proceeds with
> compliant standards. However, meetings can be informal and if anyone
> perceived anything differently, I would like to apologize from my end.
> I'm happy to clarify more things. Please feel free to ping me, send me
> email or ask for chat if you do think that's necessary.
>
> One important thing that I wanted to clarify for Newton, our top
> priorities are 1) working with the Nova team for adoption of the Glance
> v2 API 2) moving ahead and fast on the import refactor work. All of
> these are strongly tied together API hardening and ensuring we support
> interoperability requirements.
>
> Looking forward to move collaboration with the DefCore committee in the
> future.
>
> On 4/1/16 1:03 PM, Mikhail Fedosin wrote:
>> Hi Flavio! Thank you for the clarification.
>>
>>     I do realize that I missed both meetings and that logs from one of
>>     them are not
>>
>>     complete. I apologize if I've misinterpreted the intentions here.
>>     I do think
>>
>>     engaiging with DefCore as early in the process as possible is good
>>     but I'd also
>>
>>     like to clarify the intentions here before this escalates (again)
>>     into more
>>
>>     confusion about what Glance's future looks like.
>>
>>
>> I want to tell youthat the intention of the DefCore meeting was not to
>> confuse more on the work, rather it was to get clarity on all the
>> constraints that we are stuck with. Currently we intend to keep our
>> focus on interoperability issues this cycle - API hardening being our
>> first priority, along with early adoption from Murano and Community
>> App Catalog.
>>
>> And also I want to assure the community that Glare is being developed
>> consistent with the API WG principles and in such a way that it could
>> be included in DefCore at the appropriate time.
>>
>> Best regards,
>> Mikhail Fedosin
>>
>> On Fri, Apr 1, 2016 at 6:02 PM, Flavio Percoco <flavio at redhat.com
>> <mailto:flavio at redhat.com>> wrote:
>>
>>     Greetings,
>>
>>
>>
>>     I missed yday's Glance meeting but I went ahead and read the logs.
>>     While I was
>>
>>     at it, I read a sentence from Erno (under the Glare updates topic)
>>     that caught
>>
>>     my eye:
>>
>>
>>
>>             14:06:27 <jokke_> About that. I got couple of pings last
>>     night asking wtf is
>>
>>             going on. Could we please stop selling Glare as
>>     replacement for Glance at
>>
>>             least until we have a) stable API and b) some level of
>>     track record/testing
>>
>>             that it actually is successfully working
>>
>>
>>
>>     I went ahead and looked for the defcore meeting logs[0] (btw,
>>     seems like the bot
>>
>>     died during the meeting) to get a better understanding of what
>>     Erno meant (I
>>
>>     assumed the pings he mentioned came from the meeting and then
>>     confirmed it).
>>
>>
>>
>>     From the small piece of conversation I could read, and based on
>>     the current
>>
>>     status of development, priorities and support, I noticed a few
>>     "issues" that I
>>
>>     believe are worth raising:
>>
>>
>>
>>     1. Glare's API is under discussion and it's a complementary
>>     service for Glance.
>>
>>     [1] 2. Glare should not be a required API for every cloud, whereas
>>     Glance is and
>>
>>     it should be kept that way for now. 3. Glare is not a drop-in
>>     replacement for
>>
>>     Glance and it'll need way more discussions before that can happen.
>>
>>
>>
>>     I do realize that I missed both meetings and that logs from one of
>>     them are not
>>
>>     complete. I apologize if I've misinterpreted the intentions here.
>>     I do think
>>
>>     engaiging with DefCore as early in the process as possible is good
>>     but I'd also
>>
>>     like to clarify the intentions here before this escalates (again)
>>     into more
>>
>>     confusion about what Glance's future looks like.
>>
>>
>>
>>     So, to summarize, I don't think Glare should be added in DefCore
>>     in the near
>>
>>     future. The Glance team should focus on fixing the current
>>     interoperability
>>
>>     issues before we'll be able to actually try to build on top of the
>>     current API.
>>
>>
>>
>>     Hope the above makes sense,
>>
>>     Flavio
>>
>>
>>
>>     [0]
>>     http://eavesdrop.openstack.org/meetings/defcore/2016/defcore.2016-03-30-16.00.log.txt
>>
>>     [1] https://review.openstack.org/#/c/283136
>>
>>
>>
>>     -- 
>>
>>     @flaper87
>>
>>     Flavio Percoco
>>
>>
>>     __________________________________________________________________________
>>
>>     OpenStack Development Mailing List (not for usage questions)
>>
>>     Unsubscribe:
>>     OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>>     <http://OpenStack-dev-request@lists.openstack.org?subject:unsubscribe>
>>
>>     http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
>>
>>
>>
>>
>>
>>
>> __________________________________________________________________________
>> OpenStack Development Mailing List (not for usage questions)
>> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>>
>>
>
>
>


-- 

Thanks,
Nikhil





More information about the Defcore-committee mailing list