[cinder][ops][extended-maintenance-sig][public-cloud-sig][enterprise-wg] Cinder to EOL some branches

Elõd Illés elod.illes at est.tech
Thu Dec 5 14:21:11 UTC 2019


Hi Brian,

Thanks for the update! The plan looks good:
- we are not really interested in keeping Newton and Mitaka open
- but supporting to keep open Ocata and Pike - the branches in Extended 
Maintenance

About how long should Ocata and Pike be open:

I think as long as there are interest to push fixes there and is 
possible to keep the CI healthy with "reasonable efforts". That means, 
if for example the CI will be broken on Ocata (due to Zuul v3 changes, 
dependency issues, etc) and it won't be fixed in some time (a couple of 
weeks?) then it is ready to EOL.

Just to clarify: we support this not only to keep the branches open 
until a specific date but to help and encourage others to step up and 
fix issues on branches in the frame of Extended Maintenance, to make EM 
visible and show that it's worth to keep that process going.

Thanks,

Előd


On 2019. 12. 04. 16:40, Brian Rosmaita wrote:
> The tl;dr is that we won't EOL anything until Monday 9 December at the 
> earliest; see below for details.
>
> On 12/3/19 10:00 AM, Elõd Illés wrote:
>> Hi Brian,
>>
>> Thanks in advance for proposing in Cinder meeting to wait to EOL Pike
>> and Ocata.
>>
>> So far I have been looking after stable/* branches.
>
> There seemed to be consensus at today's meeting that we could keep 
> stable/ocata and stable/pike in EM mode, which would be consistent 
> with other openstack projects.
>
>>
>> I had a quick check and it seems that the CI on driverfixes/* branches
>> are not in a good shape for some time. Those branches:
>>
>> - don't have periodic jobs, so we can't see how long they are in this
>> faulty state
>> - the test jobs are very limited
>> - as I see there aren't really arriving patches lately (actually one
>> arrived in newton ~2 months ago)
>>
>> I can try to search for solutions there to fix the gate, but maybe the
>> activity there is less than ocata and pike and if there aren't others
>> who are interested in driverfixes/* branches then maybe those reached
>> the state to move them to EOL.
>
> I agree with your analysis of the state of the driverfixes branches. 
> Would you be interested in getting driverfixes/newton in working 
> order, or are you only interested in us keeping stable/o and /p open?
>
> To be clear, the current idea is:
> - EOL driverfixes/mitaka and driverfixes/newton
> - leave stable/ocata and stable/pike in EM mode
>
>>
>> Thanks again,
>>
>> Előd
>>
>>
>> On 2019. 12. 02. 16:25, Brian Rosmaita wrote:
>>> On 11/27/19 12:02 PM, Elõd Illés wrote:
>>>> Hi,
>>>>
>>>> First of all, sorry for the late response.
>>>>
>>>> About EOLing Ocata and Pike: Extended Maintenance was formed just to
>>>> have a common place for interested parties, vendors, operators, to 
>>>> push
>>>> bugfixes to. Currently these branches are in a good shape, check / 
>>>> gate
>>>> jobs work and as far as I see there are a couple of backports, too 
>>>> (not
>>>> too many, though). So hopefully it's not a waste of efforts. Why don't
>>>> we keep them open as long as the CI works and there are patches? Of
>>>> course, whenever e.g. Ocata branch / CI becomes unmaintainable 
>>>> (zuul v3
>>>> incompatibilities) or there aren't anyone who fixes the issues 
>>>> there, we
>>>> can put it to EOL then.
>>>
>>> Quick question: the driverfixes/mitaka and driverfixes/newton gates
>>> are broken now.  Do you have any interest in getting these working, or
>>> are you only interested in stable/ocata and stable/pike?
>>>
>>>> I write this, because my employer supports Extended Maintenance, and I
>>>> usually try to fix CI issues on stable branches and reviewing patches
>>>> there.
>>>
>>> Thanks for following this strategy, it's good for the community, and
>>> we appreciate it.
>>>
>>>> So maybe I can be some help here, too. Please consider leaving
>>>> branches in Extended Maintenance open as long as they are in a good
>>>> shape and there are bugfixes coming.
>>>
>>> This is good feedback.  As I said in the original email, we don't mind
>>> keeping these open as long as people are actually using them. I'll
>>> put a proposal on this week's Cinder meeting agenda to wait to EOL
>>> ocata and pike until their gates break and no one steps up to fix
>>> them, instead of proactively EOL'ing them now.
>>>
>>> Do let me know your position on the driverfixes branches right away,
>>> though.
>>>
>>>
>>> thanks,
>>> brian
>>>
>>>>
>>>> Thanks,
>>>>
>>>> Előd
>>>>
>>>>
>>>> On 2019. 11. 25. 20:21, Brian Rosmaita wrote:
>>>>> This is a courtesy notice that having received no responses to my
>>>>> email of 28 October [0] proposing to EOL some currently open Cinder
>>>>> branches, and following the policy articulated in [1], at today's
>>>>> Virtual PTG meeting the Cinder project team has decided to put the
>>>>> following stable branches into the End of Life state:
>>>>>     driverfixes/mitaka
>>>>>     driverfixes/newton
>>>>>     stable/ocata
>>>>>     stable/pike
>>>>>
>>>>> I will submit the paperwork to get this process moving one week from
>>>>> today (2 December 2019).
>>>>>
>>>>>
>>>>> cheers,
>>>>> brian
>>>>>
>>>>> [0]
>>>>> http://lists.openstack.org/pipermail/openstack-discuss/2019-October/010385.html 
>>>>>
>>>>>
>>>>> [1]
>>>>> https://docs.openstack.org/project-team-guide/stable-branches.html#end-of-life 
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>
>



More information about the openstack-discuss mailing list