Lee Yarwood recently announced the change to 'unmaintained' status of nova stable/ocata [0] and stable/pike [1] branches, with the clever idea of back-dating the 6 month period of un-maintenance to the most recent commit to each branch. I took a look at cinder stable/ocata and stable/pike, and the most recent commit to each is 8 months ago and 7 months ago, respectively. The Cinder team discussed this at today's Cinder meeting and agreed that this email will serve as notice to the OpenStack Community that the following openstack/cinder branches have been in 'unmaintained' status for the past 6 months: - stable/ocata - stable/pike The Cinder team hereby serves notice that it is our intent to ask the openstack infra team to tag each as EOL at its current HEAD and delete the branches two weeks from today, that is, on Wednesday, 22 July 2020. (This applies also to the other stable-branched cinder repositories, that is, os-brick, python-cinderclient, and python-cinderclient-extension.) Please see [2] for information about the maintenance phases and what action would need to occur before 22 July for a branch to be adopted back to the 'extended maintenance' phase. On behalf of the Cinder team, thank you for your attention to this matter. cheers, brian [0] http://lists.openstack.org/pipermail/openstack-discuss/2020-July/015747.html [1] http://lists.openstack.org/pipermail/openstack-discuss/2020-July/015798.html [2] https://docs.openstack.org/project-team-guide/stable-branches.html
Hi, Sorry for sticking my nose into this thread (again o:)), just a couple of thoughts: - we had a rough month with failing Devstack and Tempest (and other) jobs, but thanks to Gmann and others we could fix most of the issues (except Tempest in Ocata, that's why it is announced generally as Unmaintained [0]) - this added some extra time to show a branch as unmaintained - branches in extended maintenance are not that busy branches, but still, I see some bugfix backports coming in even in Pike (in spite of failing gate in the last month) - Lee announced nova's Unmaintained state in the same circumstances, as we just fixed Pike's devstack - and I also sent a reply that I will continue to maintain nova's stable/pike as it is getting in a better shape now Last but not least: in cinder, there are "Zuul +1"d gate fixes both for Pike [1] (and Queens [2]), so it's not that hopeless. I don't want to keep a broken branch open in any cost, but does it cost that much? I mean, if there is the possibility to push a fix, why don't we let it happen? Right now Cinder Pike's gate seems working (with the fix, which needs an approve [1]). My suggestion is that let Pike still be in Extended Maintenance as it is still have a working gate ([1]) and EOL Ocata as it was already about to happen according to the mail thread [0], if necessary. Also, please check the steps in 'End of Life' chapter of the stable guideline [3] and let me offer my help if you need it for the transition. Cheers, Előd [0] http://lists.openstack.org/pipermail/openstack-discuss/2020-May/thread.html#... [1] https://review.opendev.org/#/c/737094/ [2] https://review.opendev.org/#/c/737093/ [3] https://docs.openstack.org/project-team-guide/stable-branches.html#end-of-li... On 2020. 07. 08. 23:14, Brian Rosmaita wrote:
Lee Yarwood recently announced the change to 'unmaintained' status of nova stable/ocata [0] and stable/pike [1] branches, with the clever idea of back-dating the 6 month period of un-maintenance to the most recent commit to each branch. I took a look at cinder stable/ocata and stable/pike, and the most recent commit to each is 8 months ago and 7 months ago, respectively.
The Cinder team discussed this at today's Cinder meeting and agreed that this email will serve as notice to the OpenStack Community that the following openstack/cinder branches have been in 'unmaintained' status for the past 6 months: - stable/ocata - stable/pike
The Cinder team hereby serves notice that it is our intent to ask the openstack infra team to tag each as EOL at its current HEAD and delete the branches two weeks from today, that is, on Wednesday, 22 July 2020.
(This applies also to the other stable-branched cinder repositories, that is, os-brick, python-cinderclient, and python-cinderclient-extension.)
Please see [2] for information about the maintenance phases and what action would need to occur before 22 July for a branch to be adopted back to the 'extended maintenance' phase.
On behalf of the Cinder team, thank you for your attention to this matter.
cheers, brian
[0] http://lists.openstack.org/pipermail/openstack-discuss/2020-July/015747.html [1] http://lists.openstack.org/pipermail/openstack-discuss/2020-July/015798.html [2] https://docs.openstack.org/project-team-guide/stable-branches.html
On 7/9/20 12:27 PM, Előd Illés wrote:
Hi,
Sorry for sticking my nose into this thread (again o:)), just a couple of thoughts:
Always happy to see your nose :-)
- we had a rough month with failing Devstack and Tempest (and other) jobs, but thanks to Gmann and others we could fix most of the issues (except Tempest in Ocata, that's why it is announced generally as Unmaintained [0]) - this added some extra time to show a branch as unmaintained - branches in extended maintenance are not that busy branches, but still, I see some bugfix backports coming in even in Pike (in spite of failing gate in the last month) - Lee announced nova's Unmaintained state in the same circumstances, as we just fixed Pike's devstack - and I also sent a reply that I will continue to maintain nova's stable/pike as it is getting in a better shape now
Last but not least: in cinder, there are "Zuul +1"d gate fixes both for Pike [1] (and Queens [2]), so it's not that hopeless.
I don't want to keep a broken branch open in any cost, but does it cost that much? I mean, if there is the possibility to push a fix, why don't we let it happen? Right now Cinder Pike's gate seems working (with the fix, which needs an approve [1]).
We discussed this at the past two Cinder project team meetings, once to think about the idea and again today to make sure there were no second thoughts. I proposed that we would keep Pike open if someone on the cinder stable maintenance team were willing to "adopt" the branch. The silence was deafening. In short, no one on the core team is interested in approving patches for stable/pike, and no one in the wider Cinder project team of active contributors has any objections.
My suggestion is that let Pike still be in Extended Maintenance as it is still have a working gate ([1]) and EOL Ocata as it was already about to happen according to the mail thread [0], if necessary.
We appreciate your suggestion, but the feeling of the Cinder project team is that we should EOL both Pike and Ocata.
Also, please check the steps in 'End of Life' chapter of the stable guideline [3] and let me offer my help if you need it for the transition.
I appreciate your offer. I'll have the EOL patches posted shortly. The only thing I'm not sure about is whether there are zuul jobs in other repositories that are not needed any more. I don't think there are, but I may be having a failure of imagination in deciding where to look.
Cheers,
Előd
[0] http://lists.openstack.org/pipermail/openstack-discuss/2020-May/thread.html#...
[1] https://review.opendev.org/#/c/737094/ [2] https://review.opendev.org/#/c/737093/ [3] https://docs.openstack.org/project-team-guide/stable-branches.html#end-of-li...
On 2020. 07. 08. 23:14, Brian Rosmaita wrote:
Lee Yarwood recently announced the change to 'unmaintained' status of nova stable/ocata [0] and stable/pike [1] branches, with the clever idea of back-dating the 6 month period of un-maintenance to the most recent commit to each branch. I took a look at cinder stable/ocata and stable/pike, and the most recent commit to each is 8 months ago and 7 months ago, respectively.
The Cinder team discussed this at today's Cinder meeting and agreed that this email will serve as notice to the OpenStack Community that the following openstack/cinder branches have been in 'unmaintained' status for the past 6 months: - stable/ocata - stable/pike
The Cinder team hereby serves notice that it is our intent to ask the openstack infra team to tag each as EOL at its current HEAD and delete the branches two weeks from today, that is, on Wednesday, 22 July 2020.
(This applies also to the other stable-branched cinder repositories, that is, os-brick, python-cinderclient, and python-cinderclient-extension.)
Please see [2] for information about the maintenance phases and what action would need to occur before 22 July for a branch to be adopted back to the 'extended maintenance' phase.
On behalf of the Cinder team, thank you for your attention to this matter.
cheers, brian
[0] http://lists.openstack.org/pipermail/openstack-discuss/2020-July/015747.html
[1] http://lists.openstack.org/pipermail/openstack-discuss/2020-July/015798.html
[2] https://docs.openstack.org/project-team-guide/stable-branches.html
Hi Brian, Sorry to hear that :( As a side note, maybe it's good to mention, that bugfixes are getting merged in Pike now as gate is fixed (e.g. in Nova [1], where I am stable core, and also in Neutron[2]). Maybe before you EOL Cinder's Pike, it would be nice to review & merge at least the open patches [3], I can help with the review as soon as the gate fixing patch [4] has merged (which I have already reviewed :)). To be honest I haven't reviewed yet the other patches because I reviewed first the gate fixing ones and waited them to get merged. Anyway, I'm always happy to help with stable reviews, at least from stable core point of view (but I can only give +1 for patches in Cinder). About the Cinder zuul jobs in EOL candidate branches: I'll go through the zuul jobs in Pike and Ocata in Cinder to look for unused job definitions and propose deletion patch if there are such. Thanks, Előd [1] https://review.opendev.org/#/q/project:openstack/nova+branch:stable/pike+sta... [2] https://review.opendev.org/#/q/project:openstack/neutron+branch:stable/pike+... [3] https://review.opendev.org/#/q/project:openstack/cinder+branch:stable/pike+s... [4] https://review.opendev.org/#/c/737094/ On 2020. 07. 22. 21:28, Brian Rosmaita wrote:
On 7/9/20 12:27 PM, Előd Illés wrote:
Hi,
Sorry for sticking my nose into this thread (again o:)), just a couple of thoughts:
Always happy to see your nose :-)
- we had a rough month with failing Devstack and Tempest (and other) jobs, but thanks to Gmann and others we could fix most of the issues (except Tempest in Ocata, that's why it is announced generally as Unmaintained [0]) - this added some extra time to show a branch as unmaintained - branches in extended maintenance are not that busy branches, but still, I see some bugfix backports coming in even in Pike (in spite of failing gate in the last month) - Lee announced nova's Unmaintained state in the same circumstances, as we just fixed Pike's devstack - and I also sent a reply that I will continue to maintain nova's stable/pike as it is getting in a better shape now
Last but not least: in cinder, there are "Zuul +1"d gate fixes both for Pike [1] (and Queens [2]), so it's not that hopeless.
I don't want to keep a broken branch open in any cost, but does it cost that much? I mean, if there is the possibility to push a fix, why don't we let it happen? Right now Cinder Pike's gate seems working (with the fix, which needs an approve [1]).
We discussed this at the past two Cinder project team meetings, once to think about the idea and again today to make sure there were no second thoughts. I proposed that we would keep Pike open if someone on the cinder stable maintenance team were willing to "adopt" the branch. The silence was deafening. In short, no one on the core team is interested in approving patches for stable/pike, and no one in the wider Cinder project team of active contributors has any objections.
My suggestion is that let Pike still be in Extended Maintenance as it is still have a working gate ([1]) and EOL Ocata as it was already about to happen according to the mail thread [0], if necessary.
We appreciate your suggestion, but the feeling of the Cinder project team is that we should EOL both Pike and Ocata.
Also, please check the steps in 'End of Life' chapter of the stable guideline [3] and let me offer my help if you need it for the transition.
I appreciate your offer. I'll have the EOL patches posted shortly. The only thing I'm not sure about is whether there are zuul jobs in other repositories that are not needed any more. I don't think there are, but I may be having a failure of imagination in deciding where to look.
Cheers,
Előd
[0] http://lists.openstack.org/pipermail/openstack-discuss/2020-May/thread.html#...
[1] https://review.opendev.org/#/c/737094/ [2] https://review.opendev.org/#/c/737093/ [3] https://docs.openstack.org/project-team-guide/stable-branches.html#end-of-li...
On 2020. 07. 08. 23:14, Brian Rosmaita wrote:
Lee Yarwood recently announced the change to 'unmaintained' status of nova stable/ocata [0] and stable/pike [1] branches, with the clever idea of back-dating the 6 month period of un-maintenance to the most recent commit to each branch. I took a look at cinder stable/ocata and stable/pike, and the most recent commit to each is 8 months ago and 7 months ago, respectively.
The Cinder team discussed this at today's Cinder meeting and agreed that this email will serve as notice to the OpenStack Community that the following openstack/cinder branches have been in 'unmaintained' status for the past 6 months: - stable/ocata - stable/pike
The Cinder team hereby serves notice that it is our intent to ask the openstack infra team to tag each as EOL at its current HEAD and delete the branches two weeks from today, that is, on Wednesday, 22 July 2020.
(This applies also to the other stable-branched cinder repositories, that is, os-brick, python-cinderclient, and python-cinderclient-extension.)
Please see [2] for information about the maintenance phases and what action would need to occur before 22 July for a branch to be adopted back to the 'extended maintenance' phase.
On behalf of the Cinder team, thank you for your attention to this matter.
cheers, brian
[0] http://lists.openstack.org/pipermail/openstack-discuss/2020-July/015747.html
[1] http://lists.openstack.org/pipermail/openstack-discuss/2020-July/015798.html
[2] https://docs.openstack.org/project-team-guide/stable-branches.html
On 7/23/20 10:56 AM, Előd Illés wrote: [snip]
Maybe before you EOL Cinder's Pike, it would be nice to review & merge at least the open patches [3], I can help with the review as soon as the gate fixing patch [4] has merged (which I have already reviewed :)). To be honest I haven't reviewed yet the other patches because I reviewed first the gate fixing ones and waited them to get merged. Anyway, I'm always happy to help with stable reviews, at least from stable core point of view (but I can only give +1 for patches in Cinder).
We've held off on merging anything because if we're going to EOL it anyway, what's the point? -- and we didn't want to reset the 6-month 'unmaintenance' clock. But if: (1) the gates are really working, and (2) the community agrees that we can make a set of final commits to stable/pike and then immediately EOL it -- I think that would be reasonable, especially since it would allow us to merge the fixes for OSSN-0086 into stable/pike, which would be nice (though the patches have been available in Gerrit for anyone who wants them). There are no open reviews for python-cinderclient or python-brick-cinderclient-ext, so we don't have to worry about those repos. With respect to (1), I've got two test patches to make sure the stable/pike cinder and os-brick gates are functional today: - https://review.opendev.org/730959 - https://review.opendev.org/731196 I don't mean to be unreasonable, but if I have to do more than 2 rechecks on either of those to get them to pass, I have no interest in proceeding to step 2. (They both must pass because the ossn-0086 fix must be applied to both cinder and os-brick or it doesn't fix anything.) With respect to (2), the policy reads: "After a project/branch exceeds the time allocation as Unmaintained, or a team decides to explicitly end support for a branch, it will become End of Life." [0] My reading of that "or" is that we would *not* have to wait another 6 months to declare Pike EOL given that the Cinder team has explicitly decided to end support for that branch. If anyone interested in this matter reads the document differently, now would be a good time to speak up. [0] https://opendev.org/openstack/project-team-guide/src/commit/5a8b34fbba7c0744... And, just to be clear about what patches are eligible: cinder: - https://review.opendev.org/737094 - https://review.opendev.org/733662 - https://review.opendev.org/734725 - https://review.opendev.org/734723 - https://review.opendev.org/729604 os-brick: - https://review.opendev.org/733615 - https://review.opendev.org/740318 No other reviews will be considered for inclusion. I put a -W on the "Cinder: EOL Pike" patch while we think this over. But one way or another, the cinder project stable/pike branches will be EOL by this time next week.
About the Cinder zuul jobs in EOL candidate branches: I'll go through the zuul jobs in Pike and Ocata in Cinder to look for unused job definitions and propose deletion patch if there are such.
Thanks, I appreciate it. [snip]
On 2020. 07. 23. 21:43, Brian Rosmaita wrote:
On 7/23/20 10:56 AM, Előd Illés wrote: [snip]
Maybe before you EOL Cinder's Pike, it would be nice to review & merge at least the open patches [3], I can help with the review as soon as the gate fixing patch [4] has merged (which I have already reviewed :)). To be honest I haven't reviewed yet the other patches because I reviewed first the gate fixing ones and waited them to get merged. Anyway, I'm always happy to help with stable reviews, at least from stable core point of view (but I can only give +1 for patches in Cinder).
We've held off on merging anything because if we're going to EOL it anyway, what's the point? -- and we didn't want to reset the 6-month 'unmaintenance' clock. But if:
(1) the gates are really working, and
The gate jobs are really working, as I said before :)
(2) the community agrees that we can make a set of final commits to stable/pike and then immediately EOL it --
I think that would be reasonable, especially since it would allow us to merge the fixes for OSSN-0086 into stable/pike, which would be nice (though the patches have been available in Gerrit for anyone who wants them).
There are no open reviews for python-cinderclient or python-brick-cinderclient-ext, so we don't have to worry about those repos.
With respect to (1), I've got two test patches to make sure the stable/pike cinder and os-brick gates are functional today: - https://review.opendev.org/730959 - https://review.opendev.org/731196 I don't mean to be unreasonable, but if I have to do more than 2 rechecks on either of those to get them to pass, I have no interest in proceeding to step 2. (They both must pass because the ossn-0086 fix must be applied to both cinder and os-brick or it doesn't fix anything.)
With respect to (2), the policy reads: "After a project/branch exceeds the time allocation as Unmaintained, or a team decides to explicitly end support for a branch, it will become End of Life." [0] My reading of that "or" is that we would *not* have to wait another 6 months to declare Pike EOL given that the Cinder team has explicitly decided to end support for that branch. If anyone interested in this matter reads the document differently, now would be a good time to speak up.
Yes, you read and understand the documentation correctly, we don't have to wait another 6 months (see the patch that introduced this wording: https://review.opendev.org/#/q/I92542012108f0aa07e28968479cdaddf7e06301d ).
[0] https://opendev.org/openstack/project-team-guide/src/commit/5a8b34fbba7c0744...
And, just to be clear about what patches are eligible: cinder: - https://review.opendev.org/737094 - https://review.opendev.org/733662 - https://review.opendev.org/734725 - https://review.opendev.org/734723 - https://review.opendev.org/729604
os-brick: - https://review.opendev.org/733615 - https://review.opendev.org/740318
No other reviews will be considered for inclusion.
I put a -W on the "Cinder: EOL Pike" patch while we think this over. But one way or another, the cinder project stable/pike branches will be EOL by this time next week.
Thanks for thinking this over. OK, Cinder will go EOL next week, I understand.
About the Cinder zuul jobs in EOL candidate branches: I'll go through the zuul jobs in Pike and Ocata in Cinder to look for unused job definitions and propose deletion patch if there are such.
Thanks, I appreciate it.
[snip]
Thanks, Előd
On 7/8/20 5:14 PM, Brian Rosmaita wrote: [snip]
This email will serve as notice to the OpenStack Community that the following openstack/cinder branches have been in 'unmaintained' status for the past 6 months: - stable/ocata - stable/pike
The Cinder team hereby serves notice that it is our intent to ask the openstack infra team to tag each as EOL at its current HEAD and delete the branches two weeks from today, that is, on Wednesday, 22 July 2020.
As promised, here are the EOL patches: ocata: https://review.opendev.org/742513 pike: https://review.opendev.org/742523
participants (2)
-
Brian Rosmaita
-
Előd Illés